DIY ICX453AQ, Sony sensor CCD camera

[luka]

Hi Patrice
:welcome:
Thank you for clearing up several things on the Ukrainian forum as they answered some questions that I had as well.


1. I got proper quotes for the PCB manufacture from PCBway. For our board with size 56x88mm we get:
5x = $10
10x = $10
15x = $25
20x = $28
25x = $31
30x = $33
(all prices in US$)

Shipping with DHL is $25, HK post = $11. DHL will deliver within 2-3 days, HK post within 4-6 weeks. PCB manufacturing usually takes 2-3 days only.


2. Brendan, ChipQuik is the weapon of my choice. It works well even with just a soldering iron and no heat gun. Once I won a bet by desoldering 100-QFP chip with just ChipQuik and a soldering iron :)

3. Brendan, what do you think about modifying the PCB to have a separate power supply for the Peltier? That way we could use 12V or even 16V for the Peltier and 7-ish V for the rest of the circuit (I forgot what the real minimum is, will have to confirm on the Ukrainian forums). It would greatly reduce the heat generation on the PCB but we would need an extra pin. We would also need to use a common ground for both, somehow...

[pat30]

Hello,
I asked the price for the minimum 5 pieces thus € 10 for PCB and 9 € for shipping by HK post, delivered in 16 days after the order.

The Cam86 can be powered with 7.5V minimum; for Peltier I took it as voltage regulator:
Ultra-small-size-DC-DC-step-down-power-supply-module-3A-adjustable (https://fr.aliexpress.com/item/Free-Shipping-1PCS-Ultra-small-size-DC-DC-step-down-power-supply-module-3A-adjustable-step/32341006065.html?spm=2114.13010608. 0.0.L5cDL9)

Cam will be powered by 9V and 5V Peltier; there is makes -30 ° C under ambient temperature, that is enough well for me.

[gehelem]

Hue guys, you're so fast !!
Just to inform : little progress on cam84/linux driver, buffer reading seems to work now
Still no image, on step at a time...
(But i don't want to bother you all with these old cam84 stuff, i can share on CN instead...)

[wasyoungonce]

Patrice and Luka.

Yes reading the specs on the regulators. The lower 3.3V and 5V LDOs (from ST) appear have 2.0V drop out LDO while others have a 1.7V LDO. Yet this depends on mfgr spec sheets (could also change with mfgrs)...ST 2016 sheets say this but their 2009 sheets say 1.7 LDO across the board!

I'm guessing we can run at 8V minimum ok..allowing for small irregularities and mfgr part spec differences. Patrice mentioned 9V...yes that fine!

The idea of running the TEC from 12V separate source. Yes gents this has merit. What are we looking at, placing this TEC DC/DC LDO outside the case and running its output inside? Are we not using the ATMEGA328 FET TEC controller?

Luka and Patrice...is that just plain PCB price..non populated?

Patrice have you run Peltier tests? What sized TEC did you use? Sorry for hitting you up its just I was just about to run some to see what sort of delta I could get. Looking at a 30mm sq TEC.

Gilles...we would love to hear about the cam84 you have put a lot of work in it..show it off:thumbsup:

edit:
Just a question...what are the Ukrainians using (https://translate.google.com.au/translate?sl=auto&tl=en&js=y&prev=_t&hl=en&ie=UTF-8&u=http%3A%2F%2Fwww.astroclub.kiev.u a%2Fforum%2Findex.php%3Ftopic%3D289 29.2060&edit-text=) to take darks to evaluate/test their cameras...MaximDL?

[pat30]

Sorry, I do not understand non populated ??? free components?
this is what I got for that price:

http://www.webastro.net/upload/images/8654-1474041415.jpg

Yes, a lot of testing with Peltier 15x30 (tes1-6303) and 40x40 (tec12706), much heat sink aluminum, fan and different voltage.
Without too much detail, the best results (-32°C without sensor) were with tes1-6303 voltage 5V 2A and fan 30x30 or 40x40 voltage 9V not 12V .

No, this is not MaximDL is a special program: Cam View utility (http://astroccd.org/2016/09/cam86-software-pack-01/)

[wasyoungonce]

Thanks Patrice...not populated means...no components on the PCB or "Bare".

Cam-view utility...I've looked at that software pack and didn't think to ask. Thanks.:thumbsup:

I'm about to play with a 30mm SQ TEC TES1-12703 and another TEC1-12703 (40mm SQ). I think the 40mm one is too big but would like to hear about how you TECs went.

We need to agree on cooling (for our IIS cam86) before ordering PCBs. Your experience could save us lots of work! This is the last thing for us to do before ordering the PCBs.

Again many thanks

Brendan

[luka]

Sorry, must keep it short.
Regarding the separate Peltier power VT1 controls the GND to the Peltier there are few options. There is no space on the PCB to insert an DC-DC converter so we could:
1. Have the converter "dangling" in the case, hot-glued to the side (more heat inside the case but simpler if we have space).
2. Have the converter outside the case. Outside the case could mean attached to the camera (I could 3D print a small housing) or even completely detached, on the ground (floor), with a custom-made 3-pin power cable going to the camera.
3. Use two separate power supplies with shared ground, one for Peltier and one for the rest of the camera.

[pat30]

I put pictures to understand:

The heat sink is insert to the box
205124

Peltier
205125

fingers with fixation for DS18b20
205126

bedroom for CCD with Silicagel
205127

for DC-DC converter, I choose the option 2 it's very small converter

[wasyoungonce]

Hi Luka...exactrly what I was thinking...although haven't decided with way to go yet.

Uploaded new Cam86 schematic, drawings etc. Changed C45 to 10uF, 1206, X5R series. as done by Grim. Just increasing bypass capacity. He said any cap more than or equal to 10uF.

Oh I tried to update the BOM but think I made a hash of it putting in RS numbers cost etc. Might be best that one person only does these BOM changes. I'd rather it be you as I am totally lost using that "live BOM". I can send you details of schematic/PCB board changes made then you can make the BOM changes?

Is this ok? You may have reason for not wanting to do this but you have put a lot of effort into this BOM and I'd hate to ..."udo it with one foul stroke"!

Finally finished the Royel station. Had a few old caps to change, calibrate the irons and reverse engineer the PCBs, sub PCBs etc and identify all parts. So if I damage something I know what I need to fix it. Now for some nice juicy resistance tweezers. Oh The Hakko 936's are not redundant...yet!;)

Just waiting for the Chip Quik and I'm ready.


Brendan

[pat30]

I forgot to say, during my tests, I also tried 2 stage Peltier with 15x30 and 40x40, the problem is that the hot side is too hot and they need a large heat sink.

The main thing is not to have the too hot side, therefore find a compromise with the voltage.

The GND DC-DC converter is common with GND voltage, that way the Peltier is always controlled by VT1, just the positive leaving from the converter, I think this is the best solution.

[perdrix]

For any UK folks following this thread, the G106 case can be found at Rapid Electronics under their own brand.

https://www.rapidonline.com/rapid-g106-115x65x30mm-ip65-diecast-box-30-1310

Cheers
Dave

[luka]

Patrice, did you measure the temperature in between the two Peltiers?

[luka]

Brendan, I generated the latest BOM from the schematic and will double-check each part and update my list with the plan to order the missing parts and then ship you spares. I assume we won't be making many more changes to the parts...

Also we could use a boost DC-DC converter instead of a buck converter, i.e to increase the voltage for the Peltier instead of going down. Efficiencies are usually similar but higher voltage means lower current for the pins for the same power. It all depends on the choice of Peltier...

edit:
I keep forgetting to mention this, running cam86 from 7.6V or 8V may not be simple as such power supplies do not exist (unless we build a power supply ourselves). On the other hand 9V power supplies can be bought off the shelf and would make life much simpler.

[pat30]

At the junction of the two, no but the coldest temperature with 2 stage was not super -36 ° C.

[luka]

Just a guess but is it possible that your Peltiers are not matched, i.e. that the small Peltier generates more heat that the large Peltier can remove. The junction temperature will be quite high then and both Peltiers would be running very inefficiently.

Also I did a calculation of the sensor heat load a while ago. Let me find it, it may be of use for someone...

[luka]

Sorry, just to clarify, -36C was the maximum delta T or the temperature of the cold side?

[pat30]

Delta, otherwise it would have good :)

[flolic]

Damn, it's really hard to follow this on 4 different forums :rofl:

About peltiers, this is my post from CN, explaining how I assembled my own 2-stage TEC:

"Today I tested my DIY 2-stage peltier module. For the first stage, I cut TES1-12704 30x30mm module into 20mm width. On top of that I glued with thermal conductive epoxy another module, 20x15mm TEC1-2304. Both modules are connected in series. Results (without CCD) are promising, with 8V power supply, module draw 2 amps and can easily reach 60 degrees C temperature difference. 15 watts of generated heat I can easily dissipate with my CNC milled housing.
CCD will be thermally insulated and in a Argon filled chamber. Now I am worried about thermal sinking through the CCD pins...
More tests will follow."


So, with peltiers we must concentrate on efficiency, not on a power. CCD power dissipation is next to nothing, main thermal losses are thermal sinking through pins, and through convection. Convection problem can be solved by thermally insulating whole cold side, including sensor. Heat transfer through pins is much harder to reduce. On my next build my plan is to use FFC (flat flex cable) to connect sensor with PCB. That will completely eliminate thermal sinking.

[flolic]

Patrice, what's your plan for pushing enough air over those heatsink fins?

[luka]

I found my notes regarding the sensor heat generation, including the heat transfer through pins:
-----------------
The sensor itself is probably negligible. 7mA@15V = ~0.1W.

I am more worried about the heat leaking from the main PCB via the sensor pins. It is a metal connection via 34 pins. Let's try to get a reasonable estimate:
Q = k * A * dT / d (ignore convection, radiation)
k = specific heat = 66 for beryllium copper
A = area (sensor pin = 0.46mm diameter (assume round), A = Pi * r^2 = 0.17 * 10^-6 m^2)
dT = delta T = 40C
d = thickness - how tall are our pins? Assume 10mm??? Longer pins will help.
We get Q = 0.05W for beryllium copper per pin. Multiply by 34 pins to get Q = 1.5W.

So it looks like our heat load for dT = 40C is about 1.6W. Double that as a safety margin to 3.2W.
(for dT = 60C we have 2.3W, 4.6W with double safety margin)
-------------------------------

and then here are more notes regarding a particular TEC from ebay:
-------------------------------
TEC2-19003, 12V, 3A, Qmax = 35W, Tmax = 80C, dual stage, 30x30x6.5mm
We should expect best case scenario of deltaT = 73C. Sounds really good if you believe the ebay specs.
Note that this deltaT would require the total heat load that our cooler needs to deal with 5.2W + (12V*3A) = 42W(note that 5.2W is the doubled heat load at 73C temperature differential).
That will require a smallish CPU cooler. However, we will likely never run the Peltier at this power.
For comparison QHY8L uses 30W max power.

[flolic]

A little bit of off topic, I present you few pictures to show how I solved cooling and housing on my KAI4022 camera.

Back side of housing is slim but efficient CPU cooler with copper insert. I also used DIY 2-staged peltier module, sandwiched between cooler and aluminium cold finger. Cold finger is screwed to the cooler with 4 plastic M3 screws. Unfortunately, I somehow lost pictures of that detail... :(

That camera need approx. 3.2 amps (on 12V supply) for cooling, but can easily reach 48C below ambient temperature :) So I can image at -20C sensor temperature through all year.

[pat30]

I said exactly the same thing GEHELEM yesterday! :lol:

60 ° C difference is huge, very good :thumbsup:

the fan will be set just above the PCB hole and a lot of holes on the box, my first tests were good, we'll see if it will be like when it's all fitted together.
I put thermal paste between heatsink and case to help a little.

At worst, if that's not enough I would do a tube with the 3D printer to push air into the PCB hole with 1 fan and I would put a second fan to extract.

what I found odd during my testing, if I put the fan 12V (normal voltage) decreased delta : -27 ° C with fan 12V and -32 ° C with fan 9V , even fan, just different voltage means less air flow.

This afternoon I had a new room CCD bigger, with resistance to heat integrated (1.2W) for glass .

[wasyoungonce]

Thank Luka. Yes a 9V DC supply is much easier than a 8V one,

Like your TEC info...does this include ambient temp changes? Summer time can add 25ºC to the outside HSF.

42W TEC yes we can manage this wit ha reasonable computer HSF!

Patrice and Flip...thanks for your info

[pat30]

Just a message to warn you, I am working the box and wanting to put the connector in box (DC and USB) I realized that the box and too small, so you can leave the cable output or take a larger box to put connector in it, ref Hammond 1550Z111.

[wasyoungonce]

Hi Patrice

yes we have been thru this issue. The G106 case is really a little too small and it's difficult to find connectors IP66/IP67/IP68 rated. We wanted this as our case will be sealed.

I see your case is sealed for the sensor (http://www.iceinspace.com.au/forum/showpost.php?p=1273704&postcount=258)so this is not an issue for you but finding connector that will fit....its a headache. Even our on-board connector (JST-EH) (https://www.google.com.au/imgres?imgurl=http%3A%2F%2Fwww.nano elektronika.si%2Ffileadmin%2Fcatalo g%2Fimages%2Fjst-KONEKTOR__2Pin_05.jpg&imgrefurl=http%3A%2F%2Fwww.nanoelek tronika.si%2Fen%2Fkonektor-jst-2p-b2b-eh-alfsn-17623%2F&docid=Y85kmThrbfdTyM&tbnid=aAspw572sYxvsM%3A&w=800&h=600&bih=801&biw=1920&ved=0ahUKEwjg6-eilsLPAhWEjJQKHU98A50QMwgjKAYwBg&iact=mrc&uact=8) height is barely enough space to fit.

We are fixing these issues before we build but yeas you are correct, the case is too small and probably needs a dedicated case like http://www.cloudynights.com/topic/497530-diy-astro-ccd-16-bit-color-6mpx-camera/?p=7330965. Every time I see this I quiver with envious jealousy!:thumbsup:

Brendan

[wasyoungonce]

:DHere's what $64.55 buys you these days!

Anyone know of a cheaper alternate or supplier!:shrug:

Oh I also cleaned my antistatic mat:D

[Gary47]

Hi Brendan,
Would it make life easier if I made a sealed break out box to attach to the rear of the camera?
I had considered this as a way of mounting a vac/charging valve but didn't give it much further thought but it could give you more flexibility with size and positioning of connector(s).
Gary

[luka]

Brendan, E14 stocks Chip Quik. Not sure about the pricing. Ages ago I used to buy it from SparkFun but they don't stock it any more.

edit:
What is the length that you got for $65?

[osult]

Hi Brendan

I havent seen any communication regarding me gaining access to the files on Google drive :help:
maybe you can submit the request again

BTW - an interesting site on all things peltier www.tecpeltier.com
in particular article "everything you've wanted to know about tecs"
Thank you.
Tom

[wasyoungonce]

Morning gents

Gary reality the case size is the issue me thinks adding a sealed volume to a sealed volume wound make it "Frankenstein"....maybe not we can just make it with the G106 but it really is a square peg into a round hole. Next iteration of camera...a dedicated case?

Luka..Ele14 wanted $95 for the lead free 4.5ft Chip Quik kit. I paid $45 + $16 post (from Oritech) for the lead 4.5ft kit. I hate lead free stuff!

Tom I'll re-send a link to you. I looked at our GD and it has you as a shared person...no idea I'll send another. Thanks for the link I'll have a read.

Brendan

[luka]

To all the people with the GD access, I have created a new subfolder called Incoming where everyone has write access. If you want to contribute any files to the project, this is the place to upload them. For example, new ideas, different design files, photos, documentation...




Brendan, the increased ambient temperature could make the TEC less efficient but it really depends on the temperature of the hot side. With spare cooling capacity on hot summer nights we could just increase the PWM frequency to compensate.

[wasyoungonce]

Flip that is beautiful work. Did you press in the copper HSF insert into housing or just sealing it with a gasket?

Brendan

[wasyoungonce]

Thanks Luka...I guess we have to now start locking in connectors we are going to use, screws, ...etc. I planned for the JST-EH 4 way for power and 3 way for USB (2.5mm pitch, note note 2.25mm, 3A, in schematics). We can pick these up cheaply and pre-wired from fleabay (http://www.ebay.com.au/sch/i.html?_from=R40&_trksid=p2050601.m570.l1313.TR0.TRC 0.H0.XJST+EH+.TRS0&_nkw=JST+EH+&_sacat=0). They usually sell in lots but we can buy and on send them. c I tried to put this on the BOM but only mucked it up.....you'll see! Tried to put it on a separate page but as said I did a dogs breakfast of it. We probably need this info up for all.

The other connector is the housing. Your idea of the Altronics P9850 (http://www.altronics.com.au/p/p9850-amphenol-ltw-5-pin-4a-m12-screw-on-chassis-male-ip68-waterproof/) appears to fit the bill with sealing and mount hole dia. We should be able to use this and get a good seal.

We still haven't finalised a purge for the housing. This and finalise cooling...edit: oh and a machine socket for the sensor. We looked at some on fleabay but should now be thinking about buying them.

[luka]

Brendan, I already got a whole bunch of sensor sockets from AliExpress. These ones (https://www.aliexpress.com/snapshot/7939055376.html?orderId=77195555480 457). Unsure of the material...
I am not sure how to check if it is beryllium copper... any ideas?

Also I just noticed that Altronics is (mainly) Perth based, 3 out of 6 stores are in Perth. I could get the connectors for all of us and send them with the rest of the parts to you. I can get discount with our trade account and if don't have a local shop you could also save few $ on shipping.

[luka]

Faddy's comment from the Ukrainian forums:
So let's keep 33uF for C45 with 1206 footprint (I assume he meant 33uF and not 30uF).

Also this is quite important for once we start the assembly:

[wasyoungonce]

That's excellent RE: sockets. Beryllium copper...jeeez that goes a long way back but I think it was a swab test sent for spectroscopic analysis. Problem is that its used everywhere, but really only becomes an issue when it becomes ingested and inhaled. Of course F111 pavetack and F18 FLIR used beryllium mirrors. It wasn't an issue unless "bird strike"...which happened once on an F111 mirror assy. All I remember is everyone ducking and weaving as to what to do and we were the depot repair so they dumped it on us and ran away!

If you mean should we avoid this material....I do not think it's possible as it indeed has its uses that cannot be replaced and indeed is usually not an issue unless it becomes a dust inhalant. Aka all the F18 hinged panels had beryllium copper contact fingers and pads. For a reason, it provides good conductive contact and spring tension as well as low corrosion and wear constraints. It wasn't an issue on the F18 until someone workout the main canon seal in the nose/pantograph area was a beryllium copper derivative and since this damn thing rotated and produced dust...my my what an issue!

Yes the preferred value is 33uF for the capacitor.

Altronics are in Vic...but alas not near me.

I was watching this on the Ukrainian forum but couldn't work out what they were saying. Good work indeed. Basically they are scaling the ADC prior to the sensor! Well done!

Brendan

edit: I have to say I have never purchased form alibaba but the Ukrainians and you have proved it to be ok ....maybe I'm too insular!

[luka]

Brendan, we want beryllium copper. Its thermal conductivity is much lower than copper (about 1/6 if I remember correctly) and this will significantly reduce the heat transfer from the PCB to the sensor via pins.
And, as you said, the chance of poisoning is virtually zero in the metal form.

Unfortunately the description on AliExpress only says
"material: metal"
Not very helpful.

[perdrix]

Hi guys, have you chosen a Peltier cooler and fan yet? If so do you have the mechanical arrangements for fitting these defined?

Wish the original guys in Ukraine would provide more information on what they did for cooling.

I'm in two minds ATM as I have an EOS60Da, but of course that isn't cooled, whereas this would be and 6.1MP is not an issue for DSO work.

I'm working on getting a sensor right now.

Dave

[luka]

Hi David, Brendan has been playing with a Peltier from ebay but no decision has been made yet as far as I know. Everybody has been looking at the photos posted on the Ukrainian forums for ideas :-)

[gehelem]

Hi gents,

just to inform on my nights to come :
I have 2 PCBs, and ordered my components at Mouser, with Patrice's BOM.
BTW : i can access to GDrive, thank you Brendan

Waiting time = Questions time :
Do you guys have a working board ? i mean, you're now talking about boxes and cooling options, but imho i would do first things first, ie having a working circuit, no ?
(sorry to be rude)

Gilles

[pat30]

I bought these connectors, I can provide an answer, they are ferrous metal because they are attracted to magnets ;).

Can you send me the link to GDrive? although I soon finished (lack that DD7) I would see your changes.

[luka]

Patrice, can you PM me or Brendan your email address.

Gilles, we (few of us in Australia) don't have anything built yet. Brendan redrew the board and we are still thinking how to improve it to reduce noise. But basically the PCBs are ready to be ordered.

I already have all the parts here apart from few capacitors which we were are in the process of changing.

Now, several people have already built the camera and it works great. The proof of concept has been done so there is no doubt that we will need cooling sooner or later. That's why we are already thinking about the cooling and other aspects of the build.

[gehelem]

Ahhhh, ok i didn't get that, sorry again
I admire those like you who are able to go further
I'm just a follower, trying to contribute by showing noobs like me can "just" follow :)
btw, Patrice, i've ordered the bigger g106 wich is 5,5cm.
We will see.
Gilles

[pat30]

with the large box, you can put any connections, example (http://www.mouser.fr/Search/ProductDetail.aspx?qs=2Egjv1eCyqsue IykMk8zSA%3d%3d) for ip67

[wasyoungonce]

Just start to play with some TECs now a 40mm SQ (TEC1-12703) and 30mm SQ (TES1-12703). Problem is everyone would probably use their own HSFs so everyone's results may vary.

I expect the 30mm TEC to provide adequate cooling. We don't need to cool it to absolute zero...-5ºC would be adequate.

[osult]

Hi All,

I was looking at some ic sockets in Farnell which suggest the contacts are beryllium copper but after checking the product with manufacturer - it turns out only a small part of the contact is beryllium copper and the contact shell is brass - https://www.harwin.com/products/D8864-42/

The following is an alternative way to minimise the heat load transfer to the sensor:
Certain pins on ICX453AQ are repeated i.e. I think there are about 14 pins on the sensor labelled either Gnd or nc.
If all except for one of the pins on the ic socket corresponding to Gnd and NC are trimmed to prevent them connecting to the PCB and drawing heat from the board. The trimmed ic socket pins could then be connected together to the single remaining GND pin on the ic socket connecting into the PCB. The result is one pin connected to the PCB rather than 14.

Tom

[wasyoungonce]

Hi Tom I understand your idea but also think the sensor needs these common gnd points, not so much the N/C pins but the gnd is must.

On the subject of cooling...getting disappointing results with the 30mm SQ TEC. Supposed to be:TES1-12703

Voltage: 12V
Umax(V) : 15.4V
Imax(A) : 2A
Qmax(W): 27W
△TmaxQC=0(C):67 (max temp diff between module sides)
Dimensions : 30mm x 30mm x 3.2mm
Red = +ve, blk = -ve. Writing (or red lead on right) = cold side
Suggested power = 12 x 2 = 24Watts. (ideal)

Actual power = 13.45 x 2.9 = 39watts (max measured)




I only got down to around 3ºC (2.5ºC min on cold heatsink) with hot side HSF at 32ºC max. Now the hot side is difficult to measure as the cooling fans is blowing on the probe which cools it and gives bad readings, but the HSF never got hot enough, with calibrated finger mk1.

The cold side had a small heatsink on it and I even placed the cold side in a plastic case to retain this cooling. Yet it reached min ~2.5ºC and went no further.

All this was done with an ambient ~22ºC, PSU 13.44V and max draw from 3A to 2.7A. No PWM just straight cooling. I did try some PWM with a motor controller and the results pretty much mirrored this.

Which tells me either I'm doing it wrong or this TEC is not working to specs....Hmmm go figure. I tried larger fans and HSFs all the same or even worse.:shrug:

[luka]

Tom, I am not sure what disconnecting all ground pins but one would do to the sensor. While in theory it should be the same, for example those pins may be used internally to reduce the noise. There must be a reason why they are all ground pins and not NC (not connected) pins. But again, that is just a guess unless someone compares the sensor performance.

Brendan, it sounds like you are doing the right thing. What happens if you remove the small heatsink from the cold side? How could would it get without a heat load? I would expect for it to get covered in ice quickly.

[wasyoungonce]

Actually I haven't done that yet I'll do so.

Brendan

[pat30]

Brendan, repeat the test but with 5 or 6 Volts power supply

[wasyoungonce]

Hi Pat...good point.

I will have to fiddle my variable PSU only does ~1A. I might have a 9V old DC PSU around. Other than that...I have a 13.4V DC unit...oh and a computer PSU I made variable output....ha just the ticket!

I'll get back.

Brendan

[wasyoungonce]

Ok now we are getting somewhere.

Ran the TEC at 6.95V DC, pulled 1.5A max, ambient23ºC have cold side at .5ºC (...now at .1ºC) hot side at 22.8ºC. I have a small heatsink on the cold side (makes it easier to clamp together) and this probably adds some thermal mass anyway it is not working much better. This cold side is in a small plastic box to try retain some thermal efficiency.

It appears the hot side on the heatsink isn't getting hot enough I expected far hotter! Which I guess is limiting the cold side? My guess is that this should be able to get down to -5ºC but it's being limited atm...by what?

Anyway thanks Pat and Luka now I have a point I can work on.

Oh Kemal emailed me and said he had a quote from PCBway to make the PCBs and populate them with parts. See below. He was interested in going this way as he cannot do the PCB soldering.

Basically he wants to know if anyone is interested? If you are contact me and I'll put you in contact with him. IMHO this is not necessarily a bad way to go although I felt their parts cost was a little high...anyway if enough people want this ...maybe this is the way ahead?

Anyone?
I have the quote from pcbway(china).
They make the pcbs for the advertised price, so 10USD for 10 pieces.

For the assembly; 1 unit is 155USD 10 units are 180USD and 20 units are 335USD

For the components; 1 unit is 213USD 10 units are 861USD and 20 units are 1722USD

I haven’t spoken about the shipping cost but probably they will make it free with government post or maybe even with DHL.

As you can see it is not viable to order only one unit, it has to be more. If 20 people are in it will be 103.85USD per person which is a good price. But customs would become an issue, at least in Turkey it would. Chinese sellers declare lower values if asked but the quantity would raise suspicions. I’m guessing it is similar in Australia. Although for extra money they would ship them separately to each individual.

Cheers, Kemal


Brendan

[wasyoungonce]

Out of interest sake adjusting the TEC voltage to 6.0V DC (keeping the HSF fan @12V DC) saw a warming on the cold side heatsink to around 1ºC. I suspect between 12V DC and 7 V DC there is an optimum efficiency setting for this TEC.

I had read TECs were better at around 50% their max ratings voltage wise but have never explored this.

[pat30]

hi brendan,
my peltier to have the best results with 5V / 2 amp you really try different voltage, you do not have the possibility of further lowering the voltage?

and too powerful fan with bad results also, it must do many tests

[wasyoungonce]

Pat I thought the cold side is limited and dependant on how much heat you could move away from the hot side. My hot side isn't hot enough meaning the HSF is almost too good.

I'll try a different voltage. I have a variable computer PSU and I can tap into the voltage rails and vary them up/down. Lots of settings to try.;)

edit:
Ok @5.2V DC TEC power, @1.1A, cold side 1ºC (and dropping slowly), hot side 19.7ºC (and dropping slowly), HSF fan @12V DC. I think I'm bleeding cold side to hot side .....the hot side is getting cooler I'm using a computer heatsink metal retaining spring clip and long allen key head drive shaft to hold it together. This is outside the box, I can feel it getting cold and hot side is getting cooler. I though heat flowed from hot to cold but anyway its definitely effecting it all.!

But @5V the TEC is not as good as at 6.95V! More to do on this obviously

[pat30]

Yes,the power it's just for down faster but we have the time ;)

[luka]

Brendan, did I understand correctly that your hot and cold heatsinks are joined by metal? That will transfer the heat. Another thing to consider is if the cold side is getting warmed up by the fan from the hot side?

Did you use any heat transfer compound?

How big is the heat sink? If it is large, you could try running the thing at full power without the fan and KEEP AN EYE ON THE HEATSINK TEMPERATURE. Don't let it get too hot or you will kill the TEC. And don't do this without the thermal compound.

[wasyoungonce]

Hi Luka...its just a metal spring clip used in many older computer HSFs to clamp the HSF down. Contact surface of this is small very small but yes it will loose some efficiency between Hot/cold sides. This is a "rough" test just to find a sweet spot and if the TEC is suitable.

I'm still fiddling with TEC voltages. There does appear to be a sweet spot of efficiency for the TEC. I am getting better cooling using ~7V compared to 5V or 13V.

Yes I'm using thermal grease. I'll fiddle with HSF fans but I'll try to find the optimum TEC voltage first as I think this is the key.

Brendan


edit:
Ahh getting better results now:
TEC @7.08V, TEC I draw 1.54A, cold side @ -4.2ºC/hot side HSF 22.4ºC. HSF fan @12V DC. So its working better much better
TEC @7.42V, TEC I draw 1.6A, cold side @ -5VºC/hot side HSF 22.4ºC. HSF fan @12V DC. So its working even better and deeper cooling.

Repeating above voltages but throttling fan @same as TEC voltage:
TEC @7.1V, TEC I draw 1.52A, cold side @ -4.1ºC/hot side HSF 24.1ºC. HSF fan @7.1V DC. So throttling allowing hot side fan to increase this a few degrees doesn't appear to help cooling capacity.
TEC @7.43V, TEC I draw 1.6A, cold side @ -4.7ºC/hot side HSF 24.6ºC. HSF fan @7.43V DC.

Now I'm at a small wall!
My variable computer PSU can only vary up/down a small range as I modified under/over voltage trip points. So I only have so much range in adjustment. Using the 5V rail I can get from 4.5V to 7.42V and the 12V rail from 11V to 18V. So I've run out of adjustment to run the TEC. I need and 8V and 9V source. Sigh... it never ends. Ireally need to try these voltages. Wonder if I can hook between 3.3V rail and 12V rail?

[wasyoungonce]

Ok here are results for 30mm SQ TEC TES1-12703.

1st the fan. Increasing hotside fan voltage increase cooling capacity...it spins the fan faster re-moving hotside heat more rapidly. But, at some voltages (9~11V DC) its quite acceptable to run the fan from the same voltage as the TEC and the decrease in cooling capacity is minimal.

2nd. The TEC does appear to operate best around 9.0V ~10.5V DC (11V is ok as well). Its difficult to gauge as my PSU can only go down to 10.6 on 12V rail and up to 7.43V DC on 5V rail. But the TEC at 10.6V reached -7.0C easily where my controller switched off...it's a low as it can control.

So best cooling appears to be 9.0~11V on TEC fan at either 12V or at same voltage as the TEC. Increasing above this see the capacity drop off, the cold side isn't as cold! 9.0V ~10V would be optimal as the I draw would be ~ 2 amps and this voltage suits the camera PSUs as they have to drop less across them, aka they heat less.

I'd expect to see the TEC reach lower temps when put in a thermally resistant housing...aka on the case with HSF.

It doesn't appear to matter which HSF I use the TEC never go hot enough ...max HSF hot side measured was 29.6ºC TEC at 10.6V fan @12V.

So this TEC is good enough for purpose to -6~-7V cooling capacity from 9~10.5V DC TEC voltage and with the fan voltage the same or 12V.

Of course these are my fans and HSFs would the results be the same of others?

Next is the 40mm SQ TEC but...this might be too larger for the case and I draw might be too high?

[luka]

What was the ambient temperature or what was maximum deltaT?

[wasyoungonce]

Sorry just uploaded doc to the GD on this. I didn't calc delta T yet. Suffice to say the TEC didn't perform to expected/stated values.

Was busy trying to get as much info I could. Unfortunately I ran into issues with my PSU. That said I had enough to extrapolate optimum desired TEC voltages.

[luka]

Good work Brendan. It looks like the best deltaT you were getting was about 30°C. This will give about -5°C on a hot Australian summer night which is fine but I would like to have more spare cooling capacity (I like to overdesign things :D).

How heavy is your heatsink/fan?

Few thoughts:


The Cam86 uses PWM to control the TEC. Optimally we don't want to run the cooling at full power (I would guess 40-80% PWM would be the optimum) meaning that deltaT will be less than in your tests.
PWM may not be compatible all fans, depending on the fan type. We may not be able to run the fan at the same voltage as the TEC.
Performance is probably what we should have expected. Firstly, the manufacturers like to overestimate the performance and also take measurements in optimum conditions. Secondly, the stated max deltaT of 67°C is with no heat load.

I am thinking getting one of the double-stacked Peltier from ebay and doing some testing. Should have done it straight away as now I have to wait for delivery. I have uploaded a list of candidates to GD. ATM my favourite is:
TEC2-19003
dT max = 83°C
Imax = 3A, Vmax = 12V (or 16V???)
Qcmax △T=0(W) = 15W
30x30x6.5mm
(5.78V gives dT=65°C)
for about $20 on ebay or $15 on aliexpress.

Hmm... what do you think?

[wasyoungonce]

Hi Luka

I think this TEC will perform better wit ha proper set-up...aka in the case and with a copper cold finger, I used the small heatsink on the cold side to act as a thermal resistance. The Hotside HSF is a small computer unit. Seeing the hotside never really got past 30ºC much you can guess it's more than suitable.

But I was concerned to see the differential was only 30ºC. But using the current voltage this showed it to be at 27W which was the Qmax for the TEC. I think this TEC can give some more and I think its right in the range of supply voltages I cannot do at,...aka 9V DC ~ 10V DC. I say this because the rated I of the TEC is supposed to be 2A and I noted at the voltages I had it at either 1.6A or 2.4A!

I agree using this TEC doesn't leave much headroom for PWM control. Would have liked to see -10ºC on cold side. I need to buy another cheap TEC controller one that goes lower than -7ºC! Also do something about my PSUs. I have a linear adjustable...but it won't supply the current required. Need to rectify this as well.

With this info on this TEC...unless I can squeeze -10ºC out of it I don't think its suitable because it leaves no spare capacity!

Yes a dual stage may be the way to go. Looking at your suggestion now.

Brendan

[wasyoungonce]

Luka or anyone else. Does anyone have an idea on a suitable TEC module I can buy and try? Looking for something max 3A ~9~11 VDC and probably a Delta T of 40ºC. Looking to cool to at least -10ºC using minimal volts and current.

Also anyone know of a decent TEC around these parameters?

Actually any voltage from 9~12VC is ok but wanted to limit I draw and footprint size. Mainly looking for a 30mm SQ TEC size.

Also anyone one else have one of those 12V cooler controllers one that can go to at least -40ºC? Mine, WH7016R only went to -7C and I need a lower temperature control! I'd prefer the ones that have a separate fan output as you can set the fan to keep running when TEC is off...helps stock thermal shock.

Yes this is not PWM control but they are useful for testing TECs.

I also have to quickly get a variable DC supply or up to 5A. My "ad Hoc systems" need a better solution!

Sorry for the ask but I have to get onto this all quickly as I'm dragging the chain a little.

Brendan

[flolic]

If Cam86 PWM controller firmware is same as in Cam85, PWM frequency is too low to connect fan parallel to TEC. In the range of approx. 0.5Hz or even lower.

[wasyoungonce]

Yep sorry I didn't comment on this before but yes base freq may not be suitable for fans...especially the mag-lev computer types. Generally its fine to run the fan from 9~12V ...depends on the HSF performance. Generally the faster the fan the better cooling but some lower fan speeds are acceptable. Depends..If we get a suitable TEC @9v DC we have to test if the fan can remove the heat enough at this voltage. If we are talking >2.5A...I'd say yes....at a guess.

Just going to buy a cheap TEC controller wit ha bigger range doe TEC testing. Its no PWM...but you get a good idea is the TEC can fit the range we need.

[Gary47]

Bren,
I have a couple of TEC1-1270240 units, 40 mm square, supposed to have a Dt of 70 deg[URL="http://www.thermonamic.com/TEC1-12702-English.pdf"]
Connected to my 3 amp 0 to 30 v supply I find that Dt maxes at around 20 deg with a 50 mm x 50 mm x 1.5 mm plate for a load with a small HSF. Hot side temp does not go over 30 deg ( ambient 22 deg. Cold side temp drops to temp in about 4 min.
Temp 3.2 deg @ 9 V 1.88 A
Temp 2.4 deg @10 v 2.18 A
Temp 2.4 deg @ 11 v 2.3 A
Temp 2.5 deg @ 12 V 2.48 A
As you can see with only a small load the performance is a mile away from the stated performance and increasing voltage beyond 10 volt does nothing but burn more power.
I'll try it with some insulation as soon as I get a chance.
PS Test was with fan running at TEC voltage.

[luka]

Brendan, before I forget, I believe that there is a "typo" in your Peltier specs. TES1-12703 should have Imax = 3A and not 2A. The last two digits of the model number represent the Imax in Amps.
I noticed that lots of descriptions of Peltiers on ebay have copy/paste errors, even for the same or for different models. The numbers can, at best, be used as a rough guide. And I am sure that the manufacturers are not telling the truth anyway...

I doubt that a single stage Peltier will get the wanted temperature differential. Few pages back in this thread Patrice had deltaT of 32ºC without sensor with TES1-6303 (3A TEC running at 2A). I don't remember if the Ukrainians got it much better with a single TEC.

Filip got deltaT of 60ºC (see here (http://www.iceinspace.com.au/forum/showpost.php?p=1273785&postcount=268)) with a custom dual-TEC stack, again running his TECs at 1/2 and 2/3 of max current.

Dual TECs have the maximum deltaT of about 80ºC compared to 60ºC of single modules. That is a 20ºC better starting point. It also means that the TEC does not have to work as hard to achieve the wanted deltaT which means that it is operating in more efficient region. Keep in mind that our heat load is very low.

Looking at power consumption:
Your TEC uses 15.4*2 + 28 = 58.8W of power to remove 28W of heat.
(Or if I am correct and it is 3A than it uses 74W)
TEC2-19003 (which I mentioned few posts back) uses 16*3 + 15 = 63W of power to remove 15W of heat.

For 2A case the TEC2-19003 is only 5W more power hungry but it is more efficient for our purpose and optimised to get bigger temperature differential. This is what we are after.
That's why I liked the TEC2-19003 and will order one tonight to do some tests.

Also it is difficult to compare each other tests as everybody is using different heatsinks/heat loads/fans/power etc.

[wasyoungonce]

Hi Gary and Luka

Pretty much then what I found:
10.6 (optimal voltage) x 2.4 (amps) + ~30 (QTemp measured) = 55W....which is pretty much as you said.

I agree this TEC performs close to specs but not quite but is not enough for our needs. We need either a much better performing single stage...which probably won't happen given volts and amps. I have a 40mm SQ TEC I've yet to try though. Gary I found just putting the cold side in a box helped retain some cooling performance. Actually a lot of performance. The figures I obtained would be better in our set-up given better insulation etc. But Luka is right....we need a TEC to coll to around -5º~10ºC at around 40~60% PWM power. This TEC will not achieve this.

If the 40mm TEC falls short then we must go dual TEC. I might try something like flip did cut up my 40mm one and fit it to the 30mm....FrakenTEC! :lol:

Luka, do you want me to obtain a dual TEC2-19003 as well...aww heck I'll just do it as well! I'm also ordering a better controller...like the WH7016K. Yes they are not PWM controllers but they are ok for running tests to get an idea. Also I keep running out of Temp probes!:lol:

I'll do something about my PSUs as well....gawd its a long road!

Thanks Gents

Brendan

[glend]

From my cold finger dslr work i know how important the bond is between the TEC and the sensor back, i trust you guys are using high quality thermal paste like Artic Silver, some materials are not very efficient and the pads while potentually less messy are lower in efficency. Good 55-60w TECs can require as much as 6amps and i never used stackibg fir that reason. I am amazed at how my ASI1600MM-C manages to produce a delta T of 45C out of a 2amp input, on a chip just a little smaller than APS-C. If you can work that out it makes remote site operation much easier on power consumption.

Keep at it, i enjoy your discussion.

[wasyoungonce]

Hi Glen

I feel like a drunk idiot wandering in the dark. We'll get there...someday.

I'm only using silicon thermal compound atm. Will do to something better when we build the damn thing. This thread sure does provide plenty of laughs!:D

New controller WH17016C and TEC2-19003 ordered.:thumbsup:

Now...about my power supply....:shrug:

[wasyoungonce]

Been on the hunt for a variable power supply today! Lots to be had quite cheaply but some do appear to be a little inferior. That said they are cheap. Lets say a 32V 10A unit from ~$100~$150 fleabay aust.

But I was looking at EEvblog and basically they said most suffer from switch on overshoot although new units, particularity the switchmode units are better. Also Instek and Mastech units are better and most other brands are copied from these.

All this said...I was eyeing off a Uni-T UTP1035, 32V, 5A unit (http://www.hxlstore.com/531293784733.shtml#).

Has anyone have an opinion on all these variable PSUs? Particularly from experience. I need to buy one soon so I cannot wait for one to pop up on fleabay.

Anyone? :shrug:

[gehelem]

Hi Brendan,
here is my lab PSU :
https://www.aliexpress.com/wholesale?catId=0&initiative_id=SB_20161018032157&SearchText=+KA3005D
(https://www.aliexpress.com/wholesale?catId=0&initiative_id=SB_20161018032157&SearchText=+KA3005D)
i'm quite happy with it, it works well for a few month now.
Gilles.

[luka]

Brendan, I am not sure about the Uni-T unit but at the lab we used GW (Instek) GPS series power supplies (http://www.gwinstek.com/en-global/products/DC_Power_Supply/Single_Channel_DC_Power_Supplies/GPS-Series). Mainly GPS-3030 and GPS-1850. Nothing but good to say about them. We probably had > 30 units running all the time and only 2 broke during my 10 years of work there... and the failures were caused by electrolytic capacitors leaking and were simple to fix. The noise levels on the DC output was extremely low and they could also be computer controlled.

[gehelem]

Forgot to say i'm also trying two of these little babies :
https://fr.aliexpress.com/item/DP30V3A-Constant-Voltage-and-Current-Step-down-Programmable-Power-Supply-Module-Buck-Voltage-Converter-LCD-Display/32431128844.html?spm=2114.06010108. 3.9.20iEx0&ws_ab_test=searchweb0_0,searchweb20 1602_1_116_10065_117_10068_114_115_ 10069_113_10017_10080_10082_10081_1 0060_10061_10062_10056_10055_10054_ 10059_10078_10079_10073_10072_10070 _421_420_10052_10053_10050_10051,se archweb201603_3&btsid=8168ea94-98bf-43ed-9c29-cbabbbfa7bcb
No advice for the moment. They just work...

Gilles.

[flolic]

I am using Rigol DP832 tripple output PSU, but that one is on the expensive side...

[pat30]

Thanks you Gilles, I did not know it. I've seen better (https://fr.aliexpress.com/item/DPS5015-LCD-Voltmeter-ammeter-0V-50V-0-15A-Constant-Voltage-Current-Step-down-Programmable-Power-Supply/32707926774.html?spm=2114.30011508. 3.2.CDTxA0&ws_ab_test=searchweb0_0,searchweb20 1602_3_10065_10068_10069_10017_1008 0_10082_10081_10060_10061_10062_100 56_10055_10037_10054_10059_10032_10 078_10079_10077_10073_10072_10070_4 21_420_10052_10053_10050_10051,sear chweb201603_4&btsid=81b5e858-5bf0-4ee4-b65b-2402d124135a), 15 Amp, I think order soon :)

[wasyoungonce]

Thanks Gents all good info for me to look up.

Yes Rigol, Instek are well known good ones. I read many of the knock off designs copy from Instek. Emona (http://www.emona.com.au/products/electronic-test-measure/power-supplies) here in Australia sell some nice units (Rigol, Instek)...decent pricing...errr not quite but still nice. My guess is 5A unit is plenty but if I buy cheap..maybe a 10A unit would better suit.

Those Korad look a damn nice and nice price. Also those smaller switchmode units, but I'd need a Transformer rectifier unit. I kinda like the idea of a Switchmode units as they are way more efficient which means less heat. They are cheap as well!

I'm kinda a cheap a** and like the idea of building my own as have always done, but this project has poked holes in my equipment availability.

Once again many thanks lots to look at...but I'll be getting something ASAP.

Brendan

edit:
There is a nice BWD246A (http://www.iceinspace.com.au/forum/showthread.php?t=148957) for sale on IIS and it has price dropped...but I feel even though they are very good, even if I offered $300, that's almost the price of new PSUs.

[flolic]

Regarding TECs (even that I already solved that problem on my camera), just for a experiment I will order two units, TEC1-07102 and TEC1-03506.
I will stack them together and connect them in series, I think that should provide quite big temperature differerence with low supply power.

From my limited experience, lower powered TECs (for the physical size) tend to perform better for this application. Also, thicker units with lower number of thermocouples perform better.

[luka]

It looks like delaying the PCB manufacture paid off. A recent comment from the Ukrainian forums:

[perdrix]

Bren and others

I was at Electronic Design Show in Coventry UK today and came across some nice IP68 circular size 00 USB connectors by a company called ODU who are based in Germany. The connectors are similar to Lemo size 0 but cheaper :).

The come in "break away" or locking types. for the free plugs, and the sockets can be PCB mount or panel mount with a flex PCB cable to a connector on the PCB. Pricing isn't bad either (awaiting info). The also do size 00 power connectors rated to 3A if you want to bring in 12V as as well as the USB five volts.

They do prebuilt cables from their connector to a standard USB (2.0 or 3.0 depennding) in 0.2mtr or 2mtr lenghts.

HTH
Dave

[flolic]

I don't see a point finding special sealed USB or power connector. On a few of my cameras I just covered outside surfaces and pins of those connectors with a RTV silicone and that works extremely well.

Of course, that should be done after you connect wires to them and mount them to the camera
enclosure.

[wasyoungonce]

Anyone has issues with VT1 peltier FET? Grim said he destroyed a few but of course he didn't say what sort of current he was passing. Current VT1 is marginal in specs in any-case.

Looking at using IRF7832 in S08 package as the specs at Vgs ~3~3.3V (gate signal from ATMEga328) is at this level. Doing drawings up for this change now, have to put it on base of PCB but that's not an issue with it's higher current paths.

Brendan

[luka]

Did you see faddy's post on the Ukrainian forums, he/she got almost zero humidity inside a sealed enclosure. The enclosure has desiccant and was also filled with kryptonm (dry gas). Dew point is <50 degrees C below ambient resulting in absolutely no condensation anywhere.

Faddy is using a separate humidity sensor inside the enclosure with a microcontroller outside. This could be a good indicator for when camera needs purging if we have spare pins.

[wasyoungonce]

Hi Luka yes I saw this. Pretty good work. He did have trouble sealing the camera with his wires thru the case.

[luka]

Yes, is is difficult to bond to (diecast) aluminium. Silicone does not work too well. Epoxy would have done the job but it is more-or-less a permanent solution. Cyanoacrylate (superglue) also seems to bond well.

[flolic]

Guys, I use this stuff (ebay item no. 121155304226) for all kind of sealing, potting, fixing components... It is some kind of thermally conductive RTV silicone. I don't know about it's thermal properties, but all other are great :)

It's cheap, it's thick but flow nicely, it stick to all kind of materials but can easily be removed if needed.

Two pictures showing how I sealed wires to a CCD chamber:

[Gary47]

Hi Luca,
Cyano will not bond to some plastics. Also how do you get the cyano to set in an unsealed area? I have used accelerant to set it in gap fill situations but the accelerant that I had makes it brittle and porous.
Gary

[Gary47]

Is this stuff neutral cure?
Gary

[flolic]

As far as I can see, yes. It doesn't have typical acetic acid odor of "bathroom" silicone.

[gehelem]

Hi,
there's a new article on ukrainian site http://astroccd.org/2016/10/cam86/
what do you think ?

BTW : my indilib driver for Cam84 is now working !

Gilles.

[luka]

It is great news that Cam86 was added to the astroccd site. This confirms that it had enough testing to join the other "finished" Cam versions.

And Gilles, great work with the indi driver for Cam86 :thumbsup:

[gehelem]

Please note that my driver is for Cam84, not Cam86 !
For sure il will also try to make another one for Cam86, but only when i'll have a working board, wich might be long (children, wife, work...)

Gilles.

[luka]

I have a D40 sensor that is still in the carrier. I was planing to debayer it and have removed the cover glass. But I broke the DSLR so I cannot track the progress of the debayering.

I could remove it from the carrier with the DMSO and use the future Cam86 for checking of the progress of debayering but I am not sure if it will survive the DMSO bath without the cover glass.

How deep does the DMSO bath need to be? Does the whole sensor get submerged?

[gehelem]

i think whatever you do, it will be difficult to avoid the DMSO enters the sensor: for all the glue to disappear you must stir and scrape
Gilles.

[gehelem]

Have you seen last posts on the Ukrainian forum ?
These guys are awesome !
Where do they find time for all these investigations ? really ? Where ?
unless it's not a question of time, but a question of knowledge...
No seller for that on ebay...

[gehelem]

you can also have a look at this video :
https://www.youtube.com/watch?v=_TBC5_qM-hs

[pat30]

I had put it just above the aluminum, not immersed whole that was enough.
Normally DMSO does not attack the CCD or microlens.

Yes Gilles, they do a great job!

[pat30]

hi,
I put the copy of my message sent to the Ukrainian for a few tests:

I have not yet received my cxd1267 but I mounted the Cam to do the cooling test.

For me the PMW is not good:
Test 1
Cam, fan and TEC powered with 7.5 V, PWM CAM86= -12 ° C under ambient temperature in 6 minutes, and is very long to go up after the capture
Test 2
Cam, fan 12 V and TEC 5.7 V, PWM CAM86 = -15 ° C under ambient temperature in 4 minutes and is very long to go up after the capture
Test 3
Cam, fan 12 V and TEC 5.7 V, no PWM (bang bang per switch) = -22 ° C under ambient temperature in 2 minutes and goes up very quickly after stop of 3 seconds

I think we should try for assembly as me:
1 / PWM 100% until good temperature and 60% after and during capture

if not good:

2 / BANG BANG (1-0)

[wasyoungonce]

Hi Luka...just enough to cover the Lens or pins. No need to agitate solution or assy. Just leave for a few days and away it will come away.

Sorry been sick for last 1.5 weeks damn flu or something. Got infected...worst man flu ever! Just received dual stage TEC TES2-19003. Will test and also try new VT1 switch IRF7832.

[pat30]

Hello, a member of the French forum gave me that, super explanation for the Peltier (http://www.physics.utoronto.ca/~phy326/xrf/APPENDIX%20E.pdf)

[luka]

Thank you Patrice for the link, it is a very good reading.

By the way, I picked up 2 more semi-broken D70 on ebay for about $35 each (shipped).

[wasyoungonce]

Ok just a quick post...reaching -21ºC with TES2-19003, 10V, 2A in 10mins (-15ºC in 5 mins). So it's promising indeed.

Need to find optimum volts and current and test it with the new VT1.

Brendan

[wasyoungonce]

Sorry stuck atm talking to a man about a PSU. My DIY is not working I need a new PSU now. Hopefully soon within a few days.

I tested the TES2-19003 @12V DC....managed down to -22ºC in 10 mins (was -17ºC in 5 mins, ambient 20ºC) @2.2A, a delta of 42ºC!:thumbsup:

That said we need something constant not something minimum. However I get the impression this TEC would easily run at -15ºC ~-18ºC in camera but I need to try different try different combinations of power first.

My new thermoelectric controller display changes from points of a ºC to whole ºC when it reaches -10ºC. Limited digits....bummer!

[luka]

Excellent work Brendan :thumbsup::thumbsup::thumbsup:

That is very good news about the Peltier. I had a good feeling about this one from the beginning.

[pat30]

Hello everyone
Just a word to tell you that my Cam86 works :) but I have a problem with the cooling which once all mounted does not give good result 10 ° less than at room temperature. Stay tuned .....

http://www.astroclub.kiev.ua/forum/index.php?action=dlattach;topic=289 29.0;attach=114582;image

[pat30]

Hello everyone
Just a word to tell you that my Cam86 works :) but I have a problem with the cooling which once all mounted does not give good result 10 ° less than at room temperature. Stay tuned .....


http://www.astroclub.kiev.ua/forum/index.php?action=dlattach;topic=289 29.0;attach=114582;image

[pat30]

Hello everyone
Just a word to tell you that my Cam86 works :) but I have a problem with the cooling which once all mounted does not give good result 10 ° less than at room temperature :shrug:. Stay tuned .....


http://www.astroclub.kiev.ua/forum/index.php?action=dlattach;topic=289 29.0;attach=114582;image

[wasyoungonce]

Hi Pat.....excellent work...but hey your hair looks....fuzzzy!;)

Been a little stuck atm...I had small bug that got worse then the damn Melbourne Asthma storm of the century hit. This buggered me quite a bit. Finally on steroids...should be good to go soon.

But I'm still suffering from "bench PSU fixation". I need one to finalise my TEC tests as my PSUs cut out between 10V DC to ~8V DC...right where I need to test!....mumble mumble.

I missed out on some neat HP 224B's on fleabay now looking at Korad KA-3005D (https://www.amazon.com/KORAD-KA3005D-Precision-Adjustable-Regulated/dp/B0084JFWNY) (or 3005P (https://www.amazon.com/gp/offer-listing/B0085QLNFM?tag=eevblogstore-20&linkCode=sb1&camp=212353&creative=380553))to pickup from wavecom....hopefully on Monday. Then I had to fix my CRO.....urgggh!

Cool story though...:lol:

[rcheshire]

Wish I had the time. Great project and I have D70 CCD. Project burn out.

Is this of any help for your TEC calculations.

TEC hot side (TH) formula; where

TA = air temperature
Q = heatload - active + passive

Calculate Heatsink RQ

TH = TA + (V*I + Q)RQ

Transposing;

Heatsink Rating C/W

RQ = (TH - TA)/(V*I +Q)

or, substituting W for V*I

RQ = (TH – TA)/(W + Q)

Calculate copper plate passive heat load

Passive heat load formula Qp = kAD/L where;

k = the conductivity value of the material
A = the area of the conduction source
D = the differential between air temperature and component temperature.
L = the length of the conductive medium
k (kappa) of copper = 348 - steel k = 56 - fibreglass or plastic attributed k = 0.04.

Sensor active heat load
~250mW / MP - just an estimate.

I've attached an RQ calculator which is interesting to play with. Based on the TH formula above.

[pat30]

Not the hair, I do not have :lol: , sewing thread !

[wasyoungonce]

Thanks Rowland that's excellent.

Brendan

[wasyoungonce]

Finally done the TEC tests. Pretty much TES2-19003 run at 12V, ~2.1A get a delta of 50.2ºC (-20ºC and lower) regulated. So running it at -15ºC (within 10 mins time) should be fine.

Tested TEC with IRF7832...no problems as a switch device. Had to add resistors to gate to lower capacitance but this was not an issue running TEC Vgs @3.0V Pd across TEC was .273V, well within the devices limits.

I did hear a "crack" when re-mounting the TEC...although inspection showed no cracks but I did notice it was a little slower in cooling than my original runs...by ~2ºC. Ordered another TEC....:ashamed:

Funny thing increasing Vgs did not improve conduction or cooling! Rds on is good for low Vgs on this device.

So final hurdle, need to add FET gate resistors to schematic and its ready to roll!....finally:D

Yeah baby!

[wasyoungonce]

PCBs done...just not need to double check design rules for fab house and gerbers.

Finally!

[flolic]

Brendan, can you share your latest PCB design? I've done some work on my own, modifying design that you sent me few months ago...

I have ordered PCBs from seeedstudio a month ago (but they scre*ed something with shipping), so I reordered boards this time from OSH Park and they are in production right now. I expect them to arrive before Christmas.
I gathered all other parts, only boards are missing.

PCB preview from OSH Park:

https://dl.dropboxusercontent.com/u/21387397/Astroforum/Cam86/cam86_top.png

https://dl.dropboxusercontent.com/u/21387397/Astroforum/Cam86/cam86_bottom.png

[wasyoungonce]

Hi Flip...no problems..I'll send a link to our Google Drive.

Basically we changed the VT1 to a IRF7832 and tested this using TECs. The reason was the original VT1 as not good enough Rds on (too high) with low gate Vgs from the ATMEGA328 (~3.0~3.3V).

You look like you have already done something like this with Q2? Oh we added 2 extra 100uf ceramic capacitors on DD11 output as per Ukrainian forum.

Anyway you boards look damn fine. Both our layouts are very similar. We are just about to order, probably from PCBWay. Luka is doing this as he's dealt with them before. I've only used OSHpark and futurlec (pls don't use futurlec anyone they have gone off!)

Brendan

edit:
Forgot to ask...have you made any other changes ...kinda something we might have missed?
Brendan

[flolic]

Brendan, thanks for the GD link ;)
In short, layout of my board is just modified version of one of your earlier designs (which is based on original Ukrainian PCB). So yes, all of our boards are very similar.
As you noticed, I replaced VT1 with some of more robust mosfets in SO8 package. I have bunch of IRL6342 transistors, that one should be ok.
I also increased cutout area under the sensor, there is no reason to limit yourself without any valid reason.

I had excellent experience with seeedstudio in the past, but now they screwed something and not responding to my emails..:confused2:
So I tried OSHPark for the first time, they are more expensive but I will get ENIG finish. They charged me 35USD for three boards.

[wasyoungonce]

Hi Filip.
I uploaded my design to OSHPark just to see if it was ok even though my gerbers look ok in Gerbv. I note they are quite a bit more expensive but they do nice make boards.

The IRL6342 looks to be a good choice for TEC switch. Though I cannot see any gate resistor or gate bleed resistor to gnd? Look it shouldn't matter as it's driven from the ATMEGA328 out but its just good practice to include such resistors to lower stray capacitance, improve switching time and help dampen any oscillations.

But in reality if the PWM is 1Hz....I don't think its going to be an issue.

I actually tried permutations of switching our FET without gate resistors...there was no change to conductance nor package Pd. So take the above 2cents advice with the worth of a :Dgrain of salt!:thumbsup:

Keep us up to date. Good to see others and their ideas coming together.:thumbsup:

[wasyoungonce]

Filip gave me some layout suggestions and shared his layout. His layout is very close to ours and I agree I probably needed to spend a little more time with mixed signal gnds and routing paths for bypass caps...as well as placing fiducials closer to the GND and these caps.

This allows low impedance between the layers and reduces noise loops. I kinda placed gnd fiducials in geometric patterns but I did place more near low sensitive signal areas.

Also my GND planes could have been a little more defined in encompassing the cold finger cutout.

So...we are going with what we have and maybe next iteration I'll get better.

Ok Filip...Luka said he can throw your PCB design onto our batch if you want. It'll be cheaper! Just a thought and the Christmas goodwill is flowing like a beer in a duststorm!

Just PM me or Luka if you want.

Brendan

[flolic]

Hi Brendan,
Thanks for your offer, but I already have open (and paid) orders on a two different PCB houses :rofl:

[wasyoungonce]

For the record...PCB orders placed by Luka. Gary is hunting optical glass windows.

We are starting the ball rolling now!

Its taken ;longer than expected but we have done our own PCBs and made improvements suggested along the way.

Finally, like Christmas...its happening! Jeez I hope my soldering is up to scratch!;)

Brendan

[wasyoungonce]

Also looks to be some new iterations of cam86 already being made by Faddy. A round stackable PCB (http://www.astroclub.kiev.ua/forum/index.php?topic=28929.2420). Makes it easier to seal the sensor area. Looks very nice but needs a few machined parts.

Cam86 firmware and drivers (http://astroccd.org/category/cam86/)now appear quite mature and even usable in indi as well as ASCOM.

Also a Cam90 (http://www.astroclub.kiev.ua/forum/index.php?topic=28929.2420) in progress using ICX493A (10M) and ICX613 (14M) sensors.

Damn this really have taken on a life of it's own! Well done to Ukrainians. Now to build it...:lol:

[pat30]

My friends, I am happy to present my first picture with Cam86, there is still a little work to do but it is a first.

A very good thing to have increased the capacitors of DD11.


http://www.webastro.net/upload/images/8654-1481443054.jpg

[wasyoungonce]

Hi Pat...looking good.

We added qty 3 places for DD11 ceramic caps on advice and changed VT1 to IRF7832.

We are all looking at getting this working and you are providing inspiration. Our PCBs shouldn't be too long now. Many thanks

Brendan

[luka]

Just a quick update, the boards are finished and are waiting to be collected by DHL.
Brendan, I scored us some freebies as well :D

[wasyoungonce]

Excellent...Santa is early this year?

Brendan

[rcheshire]

Exciting..! This is (https://www.google.com.au/url?q=http://www.infineon.com/dgdl/irfs7430-7ppbf.pdf%3FfileId%3D5546d462533600 a40153563a69b721ca&sa=U&ved=0ahUKEwio3NSgm4DRAhUIQLwKHf9nCr 4QFggLMAA&usg=AFQjCNG75MR56nxcIBiLSeHOgsktdDV JcQ) a good little mosfet for cooling. I'm using the DPAK2-7 smd version. Very low Rds(on). Runs barely warm to the touch.

[wasyoungonce]

Rowland I think your link is broken.

Luka has received our PCBs.....but...I stuffed up the top and bottom silkscreen layers. I have these files included and they had bytes of data...but no layers were drawn.

Looking at all my other Gerbers..even others I generated that day....they are ok...no idea how this happened...operator error I suspect!:sadeyes:

[luka]

The boards are here and I spent most of the day soldering.

Step 1: solder the FTDI chip and the 3.3V power
The chip was detected on PC, everything worked fine. The 3.3V regulator is getting quite hot as the voltage is getting dropped from 12V to 3.3V. Bad design but there are several workarounds.

Step 2: solder the rest of the voltage converters
All voltages tested OK.

Step 3: solder AtMega328P, ADCs etc.
I soldered most of the other components but did not finish everything. Also I noticed that I had incorrect DD6 and VT1 so I ordered new ones. Probably won't have them before Monday.
VT1 is used to run the TEC so it is not critical. But DD6 is a show stopper. I suppose I can still flash the AtMega firmware without it and test that.

Also my DMSO finally arrived and one sensor is having a bath. The bottle leaked during the transport and I got the stuff all over my hands while unpacking. I suppose I am going for the "garlic taste ride" :(.

Brendan, what did you use to clean the sensor after the DMSO application?

[luka]

Also Rowland's link did not work for me when I clicked on it but when I copied/pasted it, it worked fine :shrug:

It is for IRFP7430PbF rectifier.

[pat30]

Acetone

[flolic]

My boards from Seeed also arrived today! :)
They look great and tomorrow I will start soldering.

[wasyoungonce]

I've just left them till I need them but linked all the pins together with fine copper wire. Kinda just storing them.

But you'll use them soon so no need for that.
Cleaning, Isopropyl, Acetone, Chloroform (yes don't laugh it works). The surface cover glass is quite hardy. You can even use warm vapour like from an ultrasonic bath. As it forms and evaporates it takes away contaminates.

Hi Flip....you're PCB looks great I look at it with envy:ashamed:...you and Luka were to kind not to mention my little screw up!:whistle: Also you're PCB has some tented via and some not! How did you do that?

[flolic]

Honestly, I don't know :rofl:
But after closer examination I can see slight variations between different boards. It looks like inconsistency in manufacturing process, some boards have thicker solder mask around vias, some not. Also there are variations on the same board. All vias should be covered and they are, just on some solder mask is thinner and looks like it's missing.
I am wondering how boards from OSH Park will be...

[luka]

I just had an interesting thought... we ordered 2x 10 boards and not 20 boards, i.e. there were two separate orders of identical 10 boards done at different times. It was cheaper that way.

Anyway, it would be interesting to compare the boards from two different batches... Unfortunately you have one batch Brendan while I have the other :) I was planing adding a few more for you but in a rush left them on my desk. Compare photos?

[luka]

Thank you guys for the info. The DMSO worked very fast, I checked 20h after I put the first sensor in and it was free. I thought it would take a few days???

Now my test debayered one is in, to see if DMSO affects the sensor surface/epoxy/golden wires. No cover glass.

Will finish soldering the first board later tonight...

[luka]

Great news, the Cam86 is almost alive :party:

I finished soldering all the components I had. Then I flashed the FTDI chip (my previous information that this was not required was not correct) and also the Atmega328P.

I even managed to get a first image... now I am missing DD6 so the Atmega328P cannot talk to the ADC and the sensor, hence the image is black.

Almost there... so far so good. Just waiting for the DD6 which will probably be here on Monday.

I am compiling detailed instructions, the Ukrainian translations by Google are not very clear. Also need to update the parts list.

And finally my 2nd sensor in DMSO got free from the carrier within less than 5h. Damn that stuff is working fast.

[wasyoungonce]

Jeez Luka you are stormin ahead. Breaking new ground...IIS's first cam86 (with a bucketload of work by Grim etc).

Looking fwd to seeing this baby opens it's eyes to the public! Well done.

Brendan

[wasyoungonce]

Pat and Flip...Luka asked me the value of C43. Unfortunately I have it as 10nf but my parts list says 1nf. If I make a change I usually put a hidden pop note in xls for the reasons...on this I have none. Just for interest sake...I went thru all my drawing revisions and I have always had C43 as 10nf in schematic and 1nf on parts list so I think the schematic was a typo.

Can you please advise.

Thanks

Brendan

[luka]

Than you Brendan but credits should go where they are due, i.e to you. It was your hard work to get the PCBs redesigned and check every component.

Excellent work :thumbsup:

[luka]

Here is detailed info how to flash firmware to the Cam86. Hopefully it will be useful to someone:

1. Connect Cam86 and let Windows install the FTDI drivers.

2. Update firmware on the FTDI device:
- Run the MPROG utility
- In Mprog, Click on the magnifying glass icon and check that it shows number of programmed devices = 1. If it shows more, unplug any other USB devices that may contain FTDI chips. if less, something has not installed properly.
- Open the template file cam86.ept, and click on the 'program existing devices' button.

3. Remove power to Cam86, wait few seconds, plug power back in

4. Setup FTDI device in device manager
- In device manager under "Universal Serial Bus Controller" find "USB Serial Controller A" and "USB Serial Controller B", right click on one of them and select properties.
- Under tab Advanced unclick loadVCP and click OK
- Do the same of the other device

5. Flash firmware to the Atmega328P chip
- Run Cam86 Programmer
- Click "Open", you should see something like this:
Number of devices : 2

device ¹0 :
SerialNumber :CAM86A
Description :CAM86 A

device ¹1 :
SerialNumber :CAM86B
Description :CAM86 B
Open :CAM86 B
BitMode Ok
SetBaudRate : 60000

- Click "Prog Enable", you should see something like this:
signature: 1E0F95
ATmega328P

- Click "Read Fuses", the fuses fields will be filled with binary values (1s and 0s)

- EXTREME CARE NEEDED HERE or you can disable the internal oscillator in the ATmega328P chip
Type in new fuses values as follows:
Fuse high: 11011001 (this is probably identical to what was read from the chip)
Fuse low: 11100010
Fuse EX: 11111111

- DOUBLE CHECK WHAT YOU ENTERED

- CHECK AGAIN

- Click "Write Fuses", you should see something like this:
E2
FF


- Click "Chip Erase", you should see something like this:
FF AC 80 00


- Click "Write PROGR", select the firmware in the cam86-firmware folder.
The flashing process will output
Page 1
Page 2
...
Then the program got stuck at around page 30. I waited about 30 seconds and eventually it finished with Page 219. Be patient.
When it finishes it will show over than 200 pages.

- Unplug power, wait a few seconds and you should have a working Cam86

- Install the ASCOM driver and test with CamView utility.

[pat30]

Hi,
C43 is 1000 pico farad.
sorry to do short, I am with gsm

[luka]

Thank you Patrice, that is the answer I like as I already have those capacitors here and don't have to order the other ones :-)

[pat30]

No problem, Congratulations Luka for Cam :thumbsup:

[flolic]

It's alive! :D

Cam86; exposure time 0.1s, gain 38, offset -15, debayered (monochrome) sensor
.fit here: https://dl.dropboxusercontent.com/u/21387397/Astroforum/Cam86/test0.1s_gain38_offset-15__001.fit

[wasyoungonce]

Well done indeed Filip, camera working and debayered.... now you're cooking!:thumbsup::thumbsup:

Pat thanks for that info

Brendan

[flolic]

Thanks Brendan!

Now, to proceed any further, first I have to make new enclosure and solve CCD cooling. That will have to wait a little...

But from initial tests, camera produces very clean images with low noise free of any electrical interference.
I've put camera in freezer (-18C) and took few bias frames, resulting images show very low noise with perfect Gaussian distribution. I am extremely happy so far :)

If someone wants to examine bias frames, they are here (rar):
https://dl.dropboxusercontent.com/u/21387397/Astroforum/Cam86/bias_gain38_offset-15_temp-18C.rar

[luka]

Great work Filip :thumbsup::thumbsup::thumbsup:

Few questions about debayering the sensor? I had a practice go at a broken one (from a broken camera) and while I debayered it I am not sure about the outcome as the sensor (and my DSLR) was already broken.

Did you just scrape the CFA or did you use something like dichloromethane (paintstripper) to soften the CFA?
I tried the dichloromethae method and it worked well but I had issues cleaning the gel-like leftovers. The sensor cleaned OK but the surroundings are covered by it and wiping is out of question because of the gold wires. I covered them by epoxy but some of the ends where it attaches to the sensor surface are still exposed as I did not want to risk getting epoxy on the sensor itself.

Also did you use epoxy to protect the gold wires and did you seal the sensor with the original glass again or did you just leave it open?

Thank you
Luka

[pat30]

Well done Flolic, as Luka says, give us more information for the converse of the sensor in B / W, I am greatly interested in the method!

[wasyoungonce]

So am I;)

Damn impressive I'd say!:thumbsup:

[flolic]

Thanks Luka and others!
I soaked sensor surface with Ethyl acetate, that thing will not dissolve the Bayer mask and microlenses, but it will make them much softer. Than I precisely sharpened end of wooden handle of the small art brush and scraped surface under stereo microscope with it (if I remember correctly, under 20x magnification).
Ethyl acetate evaporate very fast, so I was constantly adding it and never let the surface to dry out. That is very important, because as you scrape the surface, small fragments of Bayer mask will migrate through all the liquid, but will not stuck on some other area of the sensor.
I was working slowly from the middle to the edges. Edges are particularly sensitive because there are some very sensitive circuitry right next to the active pixels region that you can (will) kill if you scrape it. Those areas are under blue area, but there are also some narrow strips between active pixels and blue area that you should avoid touching.
The safest bet is to leave some Bayer mask near the sensor edges, but I decided (and succeeded) to clean it to the very last pixel :) But you absolutely need steady hand, good microscope and adequately sharpened wooden tool for that.

After the whole surface was scraped off I rinsed it with Ethyl acetate and while it was still wet with isopropyl alcohol. Than I blow the alcohol off with compressed air and examined surface under microscope (with highest magnification). Any remaining speck was then cleaned off with sharpened toothpick.
You can then install the sensor in the camera and check is it still working and is it clean.


For me, the hardest part was to remove original cover glass without damaging the sensor. My glass shattered but did not damage surface or any of the bonding wires.
I did not covered the bonding wires in epoxy. Working under the microscope , they are far away and of no concern to me ( but I did cover them few years ago when I was debayering my 450D without microscope).
If you are using regular epoxy there is a risk that it will break wires because it shrinks a little while setting.

I glued new glass on my sensor using UV curable resin.

You will not blow off bonding wires using regular air compressor (cca. 6 bar pressure) :D At least that's my experience on dozen of different sensors...

[luka]

Thank you Filip. The tip about the ethyl acetate is greatly appreciated.

[flolic]

Ethyl acetate is just something that I had lying around. I suppose there are other solvents more suitable for this job, but I have no experience beside this...

[pat30]

THANK YOU flolic for sharing the technique.
At the moment I test a product on a matrix, the micro lenses were dissolved in a few seconds without rubbing, now it is on the matrix. Unfortunately the electronics of this camera is dead so I could not test.

[wasyoungonce]

Our PCBs.....Luka sent me. He also included a lot of spare parts he had...damn decent of him!

The PCBs are ok, I made thge via holes a little too large but this isn't effecting anything. Just don't mention the missing silkscreen's!:sadeyes:

Hopefully in new year...they will be converting photons to electron charges!:lol:

Brendan

[luka]

Great to hear that the parts finally arrived. Australia Post express took "only" Thursday -> Monday to deliver. I though they were delivering parcels on the weekends during the Christmas period?

I ran my partial Cam86 with a 9V power supply. 3.3V regulator runs at about 60C while 5.5V regulator runs at about 50C. The 6V? regulator is at about 40C. I am not sure if using 6V to feed the 3.3V regulator would work as it may just overheat the 6V regulator.

Also found some 7.5V power supplies here (https://www.jaycar.com.au/27w-3-12vdc-switchmode-plugpack-with-usb-outlet/p/MP3316). May not be enough power for the TEC.

Someone correct me if I am wrong but I could not find the old posts, is the minimum voltage to run Cam86 7.2V?

[wasyoungonce]

I am guessing the 6V, SOT89 LDO can dissipate more heat more easily than the T092, that's why I thought of the 3.3V T092 from this. Yes its just putting more thru the SOT89 but It should handle it. Just a guess. Also we would really need to do the thermal calcs to determine the 3.3V die temp. My guess is thats it's ok as these regulators thermally shutdown themselves.

Most of the SOT89 LDOs mention 1.7V to 2.0V head room voltage above Vout but DD11 (3.3V LDO) T092 mentions as little as 1.0V above that is Vout + 1.0V to attain Vregulated. I didn't know that!

Same with DD13 the +5V LDO.

So the SOT89 LDOs can generally get away with 1.7V about output and the other 1.0V. So in theory you minimum in could be 7.7V.

All you can do is try! I also read the reference to 7.2V but have no idea where I read it.

Brendan

[wasyoungonce]

Looking thru the parts list....

Jeez those: C17,C30, C33,C38, C46, C47; 100uf 10V (X5Rs) ceramics in 1206 (3.2mm x 1.6mm) formfactor are a bit costly. I've found RS sell them in Pkts of 5 for $12.68 for 25+ units (as cheap as I can find them). We need 6 per PCB so 4 PCBs is 24 units ~ $50 for these alone.

Damn...that's expensive. I think I should have been looking at this more closely as I see I can get 3225 (http://au.rs-online.com/web/p/ceramic-multilayer-capacitors/8851960/) (3.2 x 2.5mm) and 1210 (http://au.rs-online.com/web/p/ceramic-multilayer-capacitors/1034219/) (3.2 x 2.5mm) for ~$18 for a pkt 20 and $15 respectively.

I did not choose wisely. The devil is always in the details. I really need to pay more attention to things like this...it would be easy to place these sizes on the PCB. Sorry gents.

Next iteration.....I'll be on to this more so. Still...I'm looking fwd to getting these running!

Brendan

[pat30]

Luka, yes 7.2V mini.
My CAM work with 7.5V and not problem.

[luka]

Pat and Filip, I got the final parts for the Cam86 and I finished the assembly. I was looking to start testing it using cam86-view program like they did on the Ukrainian forums.

Without sensor, do I need to do anything with the sensor pin 19 (output)? Should I ground it or should I ground anything else? I can imagine that leaving it floating won't give me meaningful results.

Thanks

[pat30]

HI,
With Camview, gain 0, shift 0, speed 0, without CCD this allows to evaluate the electronic noise of the board.
I did it without changing the PIN 19.

[pat30]

I did a debayering test on a dead CCD with "cleaner paint", I do not know if this is the right word in English (paint cleaning product on wall of city).
The microlenses are dissolved without rubbing in a few seconds but harmless on the matrix so I put product to unclog the pipes ( WC and kitchen room) and and it seems that the matrix is gone without rubbing, the electronics of this camera is dead so I can not testing, but Under the microscope (toy of my daughter) one sees well that it is no longer there.

I do not have any CCD to try again but may be an idea to consider if someone wants to do it, we never know!

http://www.webastro.net/upload/images/8654-1482089511.jpg

[luka]

Patrice, thank you for the info about the Cam86 settings. I will post some results later, still testing.

Do you know what is the main ingredient of your paint remover?

I read somewhere that dichloromethane will dissolve the CFA and dichloromethane is used in paint removers. Unfortunately the only paint remover I could find also had some gel-like substance which is then almost impossible to remove from the sensor surroundings (golden wires). I removed the CFA relatively easily from a dead sensor but the gel is still on it.

[luka]

OK, all assembled and either I don't know what I am doing or it is not working properly.

Without the sensor I get the noise image (attached). Apart from the vertical line it looks like the ones on the Ukrainian forums.

When the sensor is connected I get the 2nd image. It is just noise and does not change if I expose it to more light. The min/max values do not change even if I change the exposure time.

Any ideas? Is my sensor dead or should I look elsewhere in the circuit?

[pat30]

Try 10ms and post the picture

[pat30]

What voltage DD11 ?

[luka]

DD11 = 6V.
Other voltages (-8V and 15V) are fine as well.
10ms and 10sec images are attached.
Thank you for any ideas.

[pat30]

I don't think problem CCD.
check the circuit.
How much Amp ?

[luka]

Total current = about 200mA. It goes up to 250mA during the image acquisition.

[luka]

Another capture. This time with no sensor and I grounded the pin 19, the sensor output going to the opamp.

It looks quite different. Would this indicate that the ADC circuitry works?

[luka]

And another one without the sensor. As the reading of the sensor is slow (it takes about 2 seconds), during this time I disconnected the P19 (input of the opamp), then grounded it and then disconnected it again.

This simulates different voltages going into the ADC circuit. The pattern is obvious in the image.

Any thoughts?

[pat30]

I am at home I see the pictures better than on the gsm.

It seems even problem I had in the first photo

What value is C40? Make test with greater value.

Well wait between the images 5 seconds mini and turn off all the software on the PC. But I find it unusual for the CCD to send anything.

Send a message to Rome to see if he has an idea.

[flolic]

Luka, do you have an oscilloscope? If all supply voltages are ok, next thing to check are waveforms as per timing diagrams here: http://astroccd.org/2016/10/cam86/#more-1345

It looks like your ADC works as it should.

I made two mistakes while assembling my camera. First, I installed C15 and C16 capacitors because I was looking at schematic and not part list. Those capacitors are not needed , in fact camera was not working with them installed (no H1 output from horizontal driver).
Second mistake was C12, C13 and C14 where I mistakenly put capacitors with 10x needed value. Again, those capacitors control timing and camera produced only garbage.

[wasyoungonce]

Great pick-up Filip.....Hey Luka...guess what...I also have them on the schematic and brd. Its possible you put these on the PCB? I would have also!:doh:

Brendan

[luka]

It is alive!!!
Photos and tests later tonight, no time right now.

Filip, you were right, removing C15 and C16 solved the problems. Thank you and everybody else very much!!! And special thanks to the people from the Ukrainian forums who designed it in the first place.

[wasyoungonce]

Ugggghhh...another screw up by me!:mad2:

FWIW...I was worrying so went thru the entire schematics this morning...suffice to say they were alright!

Congrats Luka! When you get some pics....post up in the Ukrainian forum, thanks Grim...them all. I know he was watching Filip's debayering with interest! Filip you're famous! (https://translate.google.com.au/translate?sl=auto&tl=en&js=y&prev=_t&hl=en&ie=UTF-8&u=http%3A%2F%2Fwww.astroclub.kiev.u a%2Fforum%2Findex.php%3Ftopic%3D289 29.2520&edit-text=):thumbsup: (edit sorry hyper link does not work with language translation, see page 126).

[pat30]

YESSSS,
Congratulation Flolic I forgot C15 and C16

[pat30]

After the good news of luka, I give you a little of mine, finally if you are interested, if not, tell me ;) !

I tried to make the best photo of the matrix, it seems to be leave :)

http://www.webastro.net/upload/images/8654-1482257349.jpg

http://www.webastro.net/upload/images/8654-1482257385.jpg

[flolic]

Congratulations to Luka for successful camera build! :thumbsup:

@pat30, great images! Have you scraped Bayer mask off, or is that just result of chemicals?

[flolic]

Another observation. In my camera I used AD811 opamp from China (ebay). Based on price (US$2.99) and how it looks, I am quite sure it is not a genuine part.
Today I tried another opamp that I had lying around, THS4031 that came from Farnell few years ago and is for sure genuine. Comparing bias frames between the two, and without sensor, standard deviation for AD811 is 19ADU, and for THS4031 is 10ADU. That is clear and massive improvement.
I was in a hurry doing the tests, so for now take this with a grain of salt.

Those two opamps should be comparable in performance, with slightly lower noise on THS part. But who knows what is really encapsulated in a AD package...

More tests will follow :)

[pat30]

Just chemical, but one would have to try on another sensor to be able to validate the method, I would give the products tonight if you want.

[luka]

Hi guys, thank you. Sorry it took so long to do some testing but I was not feeling well...

My cam86 works well but there are few "things" I noticed. Initially I thought it had problems but I just spent 2h reading 50-ish pages on the Ukrainian forums and collecting notes and I believe I have found lots of answers. I have included the comments below as Brendan and Gary may have the same questions:

Note that all my tests were done without any cooling at ambient temperature of about 25C.

1. There is a strong vertical line on the left hand side. Actually there are other weaker vertical lines throughout the images. See attached image 1. On the Ukrainian forums it was noted that the vertical lines should go away once the sensor gets cooled. I am not sure about the big line. Pat, from memory you fixed this by using a larger capacitor C30=330uF, 6.3V. We have used 3x100uF ceramic capacitors so hopefully we are in the clear. If needed we could even boost this to 3x200uF ceramic (E14 has stock of those).

2. If you run the camera for a while you will notice that the standard deviation for the frames goes up. This is because the sensor warms up and the thermal noise goes up.

3. The 0ms exposure is not dark if there is any light shining on the sensor (image 2). I found this comment on the Ukrainian forum explaining what is going on:
4. Every now and then the camera goes nuts and the reading takes 3.8 seconds instead of the usual 2 seconds. It will do this for several images and then go back to normal. See attached image 3.
There were few comments on the Ukrainian forums to change the USB cable, to check feet 8 of DD6, to make sure the CPU on the computer is not overloaded or to change the resistor R12.
My problem is likely my temporarily USB cable which has a large section of unshielded wires.

And my final image has been taken through hand-held Nikkor 50mm f/1.8D @f/5.6 :D

[pat30]

I am surprised that with C30 300μf there is the band on the left !!!

For the Horizontal band, I had to turn off Windows defender and service "Superfetch" it seems resolved the problem, but I did not try an evening because of bad weather :(

[flolic]

Horizontal driver (DD8) is not routed correctly on your board(s). What is wrong is component placement. Decoupling capacitors should be placed the shortest electrical way to supply pins (pin 9 and pin 16). On your board the ground is totally messed up.
You can put ten more capacitors parallel to C30, C46 and C47, and that will not help anything. Those capacitors must be placed where they need to be.
Try to solder 100uF ceramic directly to DD8 supply pins with shortest wires possible.


Weaker vertical lines will disappear when sensor is cooled.

[luka]

Problem is fixed.

I added 100uF capacitor next to the pin 9 of DD8, going between the pin 16 and 12 track and GND. Image 1 shows the location of the capacitor.

Interestingly adding a capacitor next to pin 16 on the top of R16 to GND did not have any effect. In this position it is even closer to pin 16 as in the example above. Image 2 shows the location of the capacitor.

Also it should be noted that the quality of the power supply plays an important role. I tried several and some had vertical stripes which were not visible with other power supplies.

Finally, image 3 shows noise for 0ms exposure of the improved Cam86. It was kept in the fridge for few minutes to cool down but not for too long as I didn't want to have condensation on it. My guess is that the sensor was at 10-15 degrees C. Also the sensor was not in complete darkness (light leaking from the left hand side on the image).

[luka]

And one more image, M42 through 10" dob, 1s exposure. Camera was not cooled, ambient temperature was about 20 degrees C. Custom made 3D printed plastic housing.

Still have to figure out what to do with the gain and offset :shrug:

[pat30]

I had difficulty understanding "GAIN and OFFSET".

In France we say "EXPOSURE and CONTRAST" to regulate the gain it is necessary to make several picture with different times of poses, to make simple to regulate the GAIN to the highest being careful that the stars is not saturated, if not to lower a little and for OFFSET I do not touch.

We understand better using real acquisition software; I chose ASTROPHOTOGRAPHYTOOL (http://www.ideiki.com/astro/Default.aspx), it is really good and free.

The top stars seem oval, right ?

[wasyoungonce]

Excellent...but its counter intuitive. If it was a noise issue then right at the pin would fix it. I wonder if it's acoustic oscillations. I also wonder if the data sheet capacitor values should have been used.

We may have to "fine tune" each camera?

Well done, 1s exposure M42...well done indeed. You have been good for Christmas...Santa will be pleased.:thumbsup:

Brendan

edit:
oh meant to say...making notes for next iteration already especially wrt bypass caps and maybe even leaving a few extra pads for extra bypass caps. And I'll try my PCBs with the data spec caps, an electro and ceramic just to see what happens. It appears Filip was pretty spot on...curse him I was muttering under my breath but have to acknowledge he was pretty spot on and I salute that! Well done Filip!

Also...the issue could relate back to the 6V supply, maybe this needs a chip inductor on the output! I can try this on my boards! Further to this Intersil AN1108 mosfet drivers talks of mosft driving high freq capacitive loads (such as the CMOS sensor) and talks of using ferrite bead on Vs+ in parallel with a small ohm resistor as well as the 4.7uf electro and ceramic bypass caps. maybe I can do some experimenting...suffice to say...try to take the easiest route.

Lots of work ahead!

Looking fwd to this. And indeed glad its ok ATM and well we have achieved something that works!;)

[luka]

Pat, I have been using APT with my DSLR but didn't even think to try using it with the new Cam86... good thing to know that it will work, thanks. The test image of M42 was a very quick one. The Dobsonian was not properly collimated or cooled (and there was no coma corrector), the focusing was very rough etc. I just wanted to get my first DSO image with the Cam86 :D

Brendan, I did lots of testing of the effect of capacitors.
1. The extra capacitor over R16 did not make any difference. This is weird as it sits directly on pin 16 and it has almost identical distance to pin 12 as compared when it is placed next to pin 9.
2. When I removed one capacitor from our three I did not notice any difference (200uF total capacitance).
3. When I removed another capacitor there was obvious ringing of the vertical stripe (100uF total capacitance).
4. Adding capacitor next to pin 9 solved the problem.
5. Adding back two removed capacitors did not improve anything. I think I will need to cool the sensor to reduce thermal noise before making any conclusions here.

Again the proximity of the capacitor does not clear up the picture completely, as in my point 1 above. The "sweet spot" for C30 seems to be next to pin 9, just like in Grim's and Filip's circuits. Actually placing the capacitor "above" the DD9 and not "to the right of it" would probably make it even better as it is closer to the power pins.

Furthermore, once the stripe was there I could see a definite effect of different power supplies. Perhaps increasing C29 before DD11 (6V regulator) would help stabilise the power even more.

Also, if I was doing the schematics I would place the bypass capacitors next to the actual chip that they are used with instead of next to the power regulators. This will make their function immediately obvious. For example, I think it is slightly misleading to have the bypass capacitor C30 next to power regulator and not next to the actual chip DD11 (in the schematic).

[flolic]

Luka, I'm happy that you've solved the issue :)

About capacitor placement and their effect on circuit behavior, that's important. But when you are designing high speed/fast rise time circuit like this, you have to look wider. You have to look at power supply (DD11), driver (DD8) and image sensor (DD14) as a single circuit. And that mean optimizing current paths from the supply through the switch to the load. The most important thing here is ground that connect all these parts.
If you look at the layout of my board, you will notice that I tried to place components and tracks in a way that there is a shortest possible distance between them. And for a ground, I used ground fill everywhere, on a both layers, and with strategically (!) placed vias that connect top and bottom grounds.

I must confess that even my board is NOT ideally routed. The most important reason for this is that I used existing design and tried to improve it. In a process you can (and will) overlook something...
As in your example, only real life tests will show all the problems with the design. So, standard thing to do next is to redesign the circuit and try how that works. Or you can bodge few parts here and there and call it done :)

[luka]

Pat, did you get the name of the chemical used for debayering of the sensor?

Filip, did you do any more tests regarding the use of THS4031 instead of AD811?

[pat30]

Yes, sorry for the time .
I put the photos, the blue bottle for microlens and the yellow for matrix.
For the blue bottle there are no written components.


I still have the horizontal strip problem, I do not know what to do, I have to watch R12 worm now :shrug:


edit: Luka, looks on the Russian forum, Rome gave me an information for the horizontal lines.
Unfortunately I do not have oscilloscopes so I can not see now, I have to find someone !!

[flolic]

I have no time for that now...

[flolic]

pat30, on Ukrainian forum you asked about AD9826. All of AD9826s are the same (3 channel input), additional lettering are for various options (lead free, packaging...)
RL at the KRSZRL code means reel of 1500 pieces.

There is error on Mouser site stating that device have only 1 channel ;)

[luka]

Pat, if you still have problems after checking R12 and C14, as Grim suggested, can I suggest to try checking the values of various capacitors around the relevant components. When I was assembling my board I noticed how easy it was to mix up components, especially the capacitors. Apart from the size they all look the same and none have labels. Mixing them up is just too easy.

Also I think I solved the mystery why I was having reads that last 3.8 seconds... clicking too quickly/often in the software. Firstly the cam does not like taking photos immediately after a power up and also the software allows to start an exposure while the previous exposure is still going (continuous button).

[pat30]

Thank you Flolic, I found this, weird too :question:

[pat30]

I ordered all the components with separate packaging and reference on paper,
So I think I did not make a mistake but it is true that it is possible I did it with R14 and R15

[luka]

Hopefully you find the issue Patrice. Also there was lots of talk on the Ukrainian forums about bad components, even from reputable suppliers like Farnell.

Also, probably not relevant for us in Australia, there was an interesting bit of information on the Ukrainian forums:
Basically the opamp AD811 in SOP package does not works under 0 degrees C. The same opamp in other packages work fine below zero. Interesting...

[pat30]

I go a bit in the search for horizontal bands, the Cam is not the problem, it is the PC and there I need help:

The PC is in high performance mode about 1 photos out of 3 has tapes, I put it in energy saving on 40 pictures just 1 had tapes, so I think there is a relationship.

In this mode the Cam has more time to connect and the pictures are longer to download.

With the liveview function of the APT, no band so assume is at the time of recording that everything happens!

On my small notebook no problem on all the photos taken!

So if a computer gods little help me this would be nice.

[luka]

Pat, it is good news that the stripes are not caused by the cam itself.

Regarding the computer, there are few things to check:
1. Drivers - are you using the latest ones? Both the cam drivers and the motherboard drivers (check the motherboard manufacturer's website).
2. USB port - try a different port. Perhaps try the ones at the back of the computer.
3. USB 3 vs USB 2 - new computers come with most of USB 3 ports and few USB 2 ports. They are often colour coded. If you are using USB 3 port, try an USB 2 port. And other way around.

Good luck and let us know what happened.

[luka]

Another thought, I had issues (random crashing, not responding, bad images) with Cam86 running on Windows on the VirtualBox on a linux host. Moving it to a laptop running Windows solved the problems.

[pat30]

It is good, I found the solution for the horizontal bands; I changed the PC :bashcomp:.

Radical as a solution but it works :rofl:

[luka]

Great work Pat, problem solved one way or the other :lol:

Filip, when you removed the glass from your sensor for debayering, what temperature setting did you use on your heat gun? I just started working on mine and the glass came off but in 2 pieces. Sensor still works. My heat gun (858D clone) was set to 340 degrees C.

The reason I am asking is that previously I took glass from a broken D40 sensor in one piece. I used maximum temperature setting at that time.

[luka]

Another image from uncooled cam86, 78x 1s exposures through SW 10" dob. I did 100 images but lots were blurred as I had to nudge the dob several times during the acquisition to keep the target in sight.

Ignore the trailing, 1s is clearly too long exposure without any tracking.
0 gain and 0 offset, no binning.
Ambient temperature was about 20 degrees C.
3D-printed camera housing with several light-leaking holes at the back for cables.
The cam worked perfectly with APT (Pat thanks for the tip).

I noticed that 3 images out of the lot had higher noise than the rest. They were consecutive images in the middle of the sequence. Not sure what happened there.

Also, Pat and Filip, I vaguely remember reading somewhere that we should use several (4?) seconds pause between exposures. Is this right or is my memory playing tricks on me? I used 1s between exposures for this photo and everything seemed OK.

[pat30]

APT sends the command after the download of the image so no problem with 1 second.
I think the 4 seconds are for CamView because with it it is possible to ask for an image while the first is not registered!

[luka]

Debayering the sensor... need to stop myself before going too far...

[luka]

I think I will stop debayering here, at least for tonight. My hand hurts and can't even type properly so details will need to wait until tomorrow. Tools used are toothpick and patience.

Some of the CFA (on the top of image) just won't budge, no matter how hard I scrape... a bit annoyed about this.

[rcheshire]

It's a logic level MOSFET - very low rds on < 1mOhm - very cool running... on second thoughts the footprint is quite big.

[flolic]

Luka, IIRC I used around 360C to remove glass. I cooled back of the sensor with damp cloth, not allowing it to overheat.
Glass on my sensor shattered in few pieces, but fortunately did not damaged the sensor.

Good job on debayering btw ;)

[luka]

Funny, I probably got rectifier stuck in my head instead of MOSFET as there was a big logo on the top of the datasheet with the name of the manufacturer, International Rectifier :lol:

Anyway, apart from the too large footprint that MOSFET only works with 5V logic and we are using 3.3V. RDS climbs very steeply with VGS under 6V. From the datasheet I would guess it is 6mΩ for 5V.


Thank you Filip. It was actually easier than I thought, apart from few tough spots that just would not go away. Perhaps a key to getting the glass off in one piece is higher temperature of the hot-air gun. The glass will heat up quicker and there will be larger temperature differential to the resin. Hence the glass may come off quicker... but the strain on the glass would be larger as well :shrug:
Another thing I did differently with the sensor that the glass survived the removal was to start heating all corners instead of starting at one corner and chasing the opaque, separated bond.

[rcheshire]

OK. Thanks. 3.3V - just curious and must have misread Brendan's post on this - thinking 10V, which didn't add up anyway. Can you point me to the data sheet for the device and a brief description of how the gate is being driven and the method of temperature control, if it's not too much trouble.

EDIT: Found it. Just finished reading the entire thread.

[gehelem]

there is long time i've not posted, but here it is :
my cam86 seems to work, and i'm playing with my linux indi driver, here is the very first image :
http://www.webastro.net/upload/mediumimages/30193-1483201176.jpg

i will share it once it is better tested/cleaned !

:rofl:
http://www.webastro.net/upload/images/30193-1483201176.jpg][img]http://www.webastro.net/upload/minimages/30193-1483201176.jpg

[luka]

Rowland, I have attached the PDF of the design. Look at VT1 and DD5 in the middle of the PDF. Nothing very exciting to see, the ATMega328 is PWM-ing the MOSFET and hence controlling the TEC. The TEC cannot be controlled during the sensor frame read (which takes about 2 seconds) so the TEC can be left on or off during the read. This will affect the sensor temperature a bit.

Gilles, great work on the indi drivers :thumbsup:
I will have to try them out when ready.

Patrice, when you were getting horizontal stripes, did they look like in the attached image?

[luka]

The simple things we learn the hard way...
I was playing with the new Cam86. APT was set to use 2x2 binning with the sensor that still has the Bayer matrix and I kept wondering why I could not get any colour... D'oh

(Image attached, ambient was about 22 degrees C, not cooled Cam86, 112x 60s lights, 5 darks, 20 flats)

[gehelem]

Yes ! I've had the question with my driver a few weeks ago
See Jasem's answer here :
http://indilib.org/forum/ccds-dslrs/1724-bayer-matrix-and-binning.html
:)

[gehelem]

Btw : congratulations for the debayer stuff !!! Amazing

[luka]

Thank you. I still want to try debayering the corners of the sensor but just don't feel brave enough to do it :P

For 2x2 binning the sensor adds the 4 adjacent pixels together into a super-pixel. This means that the RGGB pixels defined by the Bayer matrix get added together and hence the colour information gets lost in the process. Basically I got 1/4 intensity and 1/4 resolution :(
At least the read noise was 1/4 compared to unbinned image :)

[gehelem]

Sorry, but isn't it 4x intensity ?

[luka]

Excellent point and you are absolutely right. Actually we are probably both right, it all depends what it is being compared. I should have expressed myself better.

On one hand binning reduces the resolution but we get 4x the intensity for each super-pixel as 4 individual pixel get added together, just as you said. The well depth is increased 4x as well and the read noise is 1/4 (am I right about the read noise?).

On the other hand the CFA of the Bayer matrix reduces the intensity of light each pixel sees when compared to a mono sensor. Red pixel will only see the red light, the blue pixel will only see the blue light and the two green pixels will only see the green light. Note that this is simplification, each "colour" lets a "band" of wavelengths through, see the chart for example here (http://iopscience.iop.org/article/10.1088/1742-6596/536/1/012021/pdf).

Anyway, the point is that each pixel behind CFA will see less light than its equivalent mono pixel. And I had a mono sensor next to me and could have easily swapped to it :(
How much less light? That depends on a particular CFA for a particular sensor but I guessed that 1/4 of the intensity will go through (compared to a mono sensor without the CFA). We are not doing the usual Bayer interpolation to extract the colour as we are binning the pixels on the sensor.


Ah, I am even managing to confuse myself now... do you see what I mean? Of course I could be wrong with what I said. Especially the factor of 1/4 between the mono and colour sensor is a guess anyway.

[pat30]

Yes, the bands were the ones there !

[gehelem]

This is exactly where i think all of this is really interesting : This way whe think, we understand, we learn.
None of this happens when you buy a 1000€ camera :lol:

BTW (2) :
Here is my linux driver :
https://github.com/gehelem/indi_cam86_ccd
Really beta test, please don't blame

[gehelem]

I've noticed that these bands appears when there is a problem with usb bandwidth or something like that :
with my driver, i can adjust timings and baudrates, and with "some" parameters they appear, with "others" they disappear.
Just my 2 cents...

[luka]

I did some more investigation of the horizontal lines and really could not figure out what is causing it. My guess is that the FTDI chip is not buffering the frame data but passing it straight to the PC and the data gets lost in this transfer.

My results:
Laptop 1 (win7) consistently shows them (one in 10-20 images).
Laptop 2 (win10) never does. This one has USB3 but the cam86 is USB2 so it should not make a difference, unless the problem is caused by the chipset/drivers itself.
The CPU load does not seem to affect anything.

Gilles, which operating system are you using with your Indi drivers? I didn't get a chance to try them out on my main PC (linux) yet, have to figure out how to use Indi first :)

Pat, which operating system were you using with your old computer (that was showing lines) and which one are you using now (computer not showing lines).

[gehelem]

May i suggest you to check if your laptops ar using different versions of D2XX libraries ?
For sure, FT2232 doesn't buffer anything, it sends data with the rythm given by the atmega. It's the driver's job to catch data.

My driver is running on Odroid XU4 for the server side, and kstars runs as a client either on windows 7 PC ( YES : kstars/ekos for windows !!!), or 2 or 3 classical Ubuntu distros (better with lastest 16.10 as indi is moving very fast).
Will soon try on Raspbery pi 3 as server, freshly arrived under christmas tree.
I added yesterday to my driver the ability to play with different baudrates for the 2 channels of FT2232, just to see...
If you need help to try, do not hesitate !

[luka]

Gilles, I tried compiling the driver without much luck. cmake complains about missing libftdi1 config files:
Any hints? Using debian stable. I have all the ftdi packages installed (libftdi-dev, libftd1, libftdipp-dev, libftdipp1) that I think are of relevance. I have also updated those packages from the latest debian unstable without any success.
I have noticed that LibFTDI1Config.cmake is not in the folder cmake_modules. Should it be there? I tried downloading LibFTDI1Config.cmake from the libftdi source and the error message was identical.

Thanks
Luka

[gehelem]

Did you try to build and install libftdi directly from Intra2net ?
(i don't remember, but i think it's the way i've done)
http://developer.intra2net.com/git/?p=libftdi;a=summary
This way i think necessary .cmake files should be installed properly on your system...

EDIT : happy to see you going this way :-)

[gehelem]

have a look here :
/usr/lib/cmake/libftdi1
this is where required modules where installed for me

[luka]

Installed cmake from Intra2Net and got cmake to work. Thanks for the tip. Interesting how the packaged libraries do not work.

However, now I get compilation errors which I traced to the old version of indi that comes with Debian stable. And installing indi packages from unstable will break half of the system. I think I will need to compile indi... probably tomorrow, it is time for :zzz: now.

Thanks for your help.

[gehelem]

Have a good night
Another idea : if your purpose is to play with baudrates, you can try to build the library your self, with delphi, and tweak with baudrates.
i'm about to do it, just to understand ...well .. how it works !

[luka]

Cough, cough
https://landingfield.wordpress.com/2016/02/18/the-making-of-a-cooled-cmos-camera-p1/

[gehelem]

In french : "La vache !"
no, no, no, we have to finish this one first... :)

[luka]

Gilles, I got much further today trying to get the driver working under indi:

1. It is not possible to use the driver with the current debian stable (jessie) as the other packages are just too old. I managed to compile the ftdi library and then to compile indi and then the driver and then I stumbled... ekos and the newer kstars need qt5 which debian jessie does not have. So I had to abandon this approach as it was getting too messy.

2. I got new system running debian testing (stretch). Your driver compiled with stock FTDI and indi packages and I copied the executable indi_cam86_ccd to /usr/bin. I also copied the indi_cam86_ccd.xml file to /usr/share/indi. Was this the correct?

Note: Can you please update the indi_cam86_ccd.xml file on github to say Cam86 instead of Cam84 in two places (name + binary).

3. kstars recognised and "connected" to the driver. Note that it "connects" to the driver even when the camera is not connected so I don't really believe that is connected.

This is where my success ends. The indi control panel does not work/open for Cam86. It refuses to do anything with it. I am not sure what to do next. I never used indi before so that may be my mistake. But the simulated CCD works.

Any hints?

Thanks
Luka

[gehelem]

=> xml file is ok now

you don't have to install everything, just run the driver localy in your build directory in a console, e.g. :
odroid@odroid:~/projects/indi_cam86_ccd/build$ indiserver -v -m 100 ./indi_cam86_ccdThen, select "client" in device maganer of Kstars
Add a host (localhost port 7624), and connect to it
If you want to run it with other drivers, it's this way :
odroid@odroid:~/projects/indi_cam86_ccd/build$ indiserver -v -m 100 ./indi_cam86_ccd indi_asi_ccd indi_eqmod_telescope indi_moonlite_focus(here : cam86, eqmod mount, asi ccd, and a focuser)

This is where indi is a killer app : run your drivers on a raspberry installed under your mount, and stay in your living room using Kstars (even a android versions exists, it works quite well). The goal is to run a least a mount, a ccd and a focuser (even a fake one) in order to open "Ekos" : killer of the killer app !!! May i also suggest to try Astrometry.net, hard to setup, but once it works, OMG...

In my driver, error handling is not yet propagated to indi : running the command line driver you will be able to see the console log, and most probably connections atempts failing.
Did you install the udev file ? this is very important to tell kernel to unmap ft2232 from 2 serial USB (in fact exactly like in Windows, where you tell to unload serial COM ports)
You might also have permissions problems, try to run the driver as root to check.
If so, you may add your user to "dialout" group

[luka]

Thank you Gilles.

I did not see the udev file, copying that file solved the problem. And I kept wondering before how the driver was detecting which serial ports the cam was using :D

Manually running the server like in your instructions works. I can connect to it from kstars and I get into the CCD indi control panel. I got an image as well :)
I still cannot start the server from kstars but that could be me not knowing what I am doing :(

Funny thing, I kept getting white saturated images. I tried switching the light off and so on but the images were always saturated. After eventually giving up I went to disconnect the camera and noticed that it was not even connected :P

Anyway, it is working now and I will have a play with it tomorrow. Great work Gilles :thumbsup:
Thank you again for your help.

[luka]

By the way, I posted my debayering notes in this thread (http://www.iceinspace.com.au/forum/showthread.php?t=152057). Hope it is of use to someone.

[gehelem]

Thank you for the report, i'm very happy it's working for you !

i've added yesterday a big set of parameters (LibFTDI panel) to play with baudrates and timers, i'm still searching the good ones to make horizontal lines disapear...
Still need to add :


framing
temperature control
error handling (!)

Edit :

to launch directly from Kstars, you can try "sudo make install", it's working for me : Kstars shows a cam86 item under CCDs
I didn't even noticed you can also see console message from here...

[pat30]

Windows 7 (lenovo) with lines and now Windows Starter (toshiba) and still a windows 7 (Samsung) that does not lines.

I do not understand the problem :shrug:

[flolic]

I finished first part of the camera housing :)

[gehelem]

Very nice !
Are you trying to debayer your sensor ??

[flolic]

This one IS already debayered ;)

[luka]

New ATMega firmware and new viewer software are on the Ukrainian forums or on our Google Drive.

[flolic]

Thanks!
Do you know, does new firmware keep peltier ON during frame readout?

[luka]

Not sure what this firmware does but it fixes the temperature sensor reading bug.
(The first firmware came in two versions, one with TEC ON and one with TEC off during the frame readout.)

The viewer software update fixed the bug where exposures over 15min did not work. However, in this version the timer is not very accurate for very short exposures. Take a pick...

[pat30]

Hi,
Flolic, beautiful box ;)

[luka]

Great box Filip, can't wait to see it completed.

A question to the people who have their cam86 working. About 50% of my photos have the first few pixels in the first row set to maximum intensity. The attached image shows a crop of the top left corner of one such image. While this gets cropped out in the processing, these white pixels make it impossible to use the histogram stretching in the acquisition software (APT). Does anybody else have this problem?

[luka]

Interesting, no such line is visible for images captured with cam86-viewer or sharp-cap. It appears only in images captured with APT.

Patrice, you said that you were using APT. Did you notice such line?

[pat30]

Now that I see it, actually I had this problem on a few photo but I did not look too much, because priority problem of horizontal bands.

[gehelem]

=> might it be the same problem ?!

[pat30]

It's possible

[luka]

Not sure if it is the same problem but the short white line appears only when taking images with APT and not with any other software I tried. And it is always on the top left corner of the image.

On a similar matter, I also tried taking images with nebulosity and there is a pattern of horizontal lines throughout all images when acquired in series capture. However, doing preview images (i.e. one at a time) works fine. Not sure what is going there.

What other software are people using?

(Gilles, I didn't get any time to play with the indi driver yet).