#881  
Old 31-03-2018, 12:35 AM
redriver
Registered User

redriver is offline
 
Join Date: Jan 2018
Posts: 6
Quote:
Originally Posted by luka View Post
A summary about the horizontal stripes:

1. They are not noise caused by the TEC switching. If I set the switching to be very slow (once a second) the stripes stay at the same intensity for the whole 1 second. Any electrical noise caused by the switching would not last that long.

2. They feel more like the gain or offset or some voltage is slightly changing, hence the darker/brighter stripes. Or some capacitive effect???

3. I measured voltages with oscilloscope to try to find any variations that happen at the same frequency as the stripes. I did not find anything but it is very difficult to detect small voltage variations on large base voltages.

4. If we use higher frequency the stripes start losing their sharp transition edges. Unfortunately we don't have proper PWM on the pin controlling the TEC and we cannot manually change the TEC pin fast enough during the frame reads. The stripes are clearly visible even if the TEC is switched on and off at every line change (1000 lines in 2 seconds).

5. The intensity of the stripes can be reduced if the cold finger is electrically connected to the back of the sensor by using an electrically conductive thermal paste, like Arctic Silver 5. However, they are still present.

6. If the cold finger and the back of the sensor is grounded the camera will stop working.

So, unless someone figures out what is causing the stripes I don't think it is possible to use the current TEC pin to have manual PWM during the frame reading.

Note that Rim said:

Not sure how that would help us.

Anyway, Cam90 uses a different pin for the TEC control and that pin has PWM. TEC is switched at about 500Hz and the stripes are not visible. So the only option for us would be to use the humidity sensor pin (which supports PWM), cut traces on the board and wire it to the MOSFET that switches the TEC. Or alternatively wire the humidity sensor pin to a separate MOSFET that is not mounted on the board - this would mean that no traces need to be cut.
If the horizontal stripes are caused by PWM of contolling TEC, then is it possible to use a separate 12V power supply? Or use isolated DC-DC to power the TEC and then a opticalcoupler to control the TEC? My CAM86 is under construction, TEC part is still not finished, I use a 40mm*40mm TEC1-12706 and it can only reach -12.8 with a huge powerful GPU cooler. I am considering with two stage TEC...
Reply With Quote
  #882  
Old 31-03-2018, 12:53 AM
luka's Avatar
luka
Unregistered User

luka is offline
 
Join Date: Apr 2007
Location: Perth, Australia
Posts: 1,164
It is definitely caused by the power supply and some sort of capacitive coupling. I have not posted the results yet but I am on the verge of solving the issue, at least with my camera.

1. I swapped the humidity and TEC control pins by some soldering on the board.
The humidity pin can use hardware PWM but the original TEC pin cannot. This will allow us to do faster TEC switching.
2. At 8KHz switching frequency I could not notice any "noise". The frequency is higher than the line reading frequency.
3. Another improvement has been achieved by using electrically conductive thermal paste between the cold finger and the back of the sensor. This puts the cold finger at the same potential as the back of the sensor and it helps a lot.
DO NOT GROUND THE COLD FINGER.

I am still doing tests but it looks like this has solved the problem for me. Few other people will be doing tests soon and I will post the results then.

If needed further options would be to connect ground directly from the power connector to the VT1 MOSFET. Possibly even move the VT1 from the main PCB. Optoisolator would probably help as well.

Do not use a single stage TEC, you won't get very high temperature differential. Quite a few of us used TEC2-19003 from ebay and it is possible to get 40 degrees C temperature difference.
Reply With Quote
  #883  
Old 04-04-2018, 11:34 PM
redriver
Registered User

redriver is offline
 
Join Date: Jan 2018
Posts: 6
Quote:
Originally Posted by luka View Post
Do not use a single stage TEC, you won't get very high temperature differential. Quite a few of us used TEC2-19003 from ebay and it is possible to get 40 degrees C temperature difference.
Hi Luka,

I will practice to use a optoisolator and a separate power for peltier. And I have a issue of peltier placement, does the peltier reside inside the case or outside the case? If the peltier is outside the case, then the cooling if much better, as it has larger heat dissipation area, but the cold finger copper is more long, which brings bad for colding.

Thanks!
Attached Thumbnails
Click for full-size image (制冷方案1.jpg)
43.3 KB37 views
Click for full-size image (制冷方案2.jpg)
39.5 KB34 views
Reply With Quote
  #884  
Old 05-04-2018, 12:53 AM
luka's Avatar
luka
Unregistered User

luka is offline
 
Join Date: Apr 2007
Location: Perth, Australia
Posts: 1,164
Optoisolator should help as it will keep the board power and the TEC power completely separate. It will complicate the electronics though, if you are using the original G106 housing things may not fit inside any more. By the way, there is a deeper version of G106 by Hammond, part number 1550Z111.

Keep in mind that we will most likely have to use the humidity pin for the TEC as the current TEC pin does not support PWM. Still doing tests here so don't do any changes yet.

While you are at the design stage here is another thought. The 3.3V regulator gets quite hot as it powered by the input voltage (12V). The camera actually only needs about 6.5V (say 7V to be safe) to run. If you are using separate power supplies for the TEC and the camera electronics you can power the camera with a lower voltage to reduce the heat production.
In my current design I use one 12V power supply but the camera electronics is powered by an external 8V regulator while the TEC gets the full 12V. This way some of the heat is "moved" from the PCB to the breakout box outside the main case.

Regarding the TEC designs, both are possible but I am not sure about the 2nd drawing where only a part of the hot side of the TEC is being cooled. If possible you should shape the cold finger to look something like a letter T in side view. The top of T should be square and fully cover the TEC while the bottom "leg" of T should be round and go to the cold finger through a hole in the PCB. Hope this description makes sense, if not I can do a drawing.

My current design is like your first drawing and I can get down to -40 degrees C below the ambient temperature with the TEC I mentioned before. The final temperature, of course, also depends on the size of the heatsink and the fan.
Reply With Quote
  #885  
Old 05-04-2018, 09:14 PM
redriver
Registered User

redriver is offline
 
Join Date: Jan 2018
Posts: 6
Thanks for your support

For the hot regulator problem, right I now I am using a 9V DC supply to power the board. VT1 is not soldered yet as I have not using that to power the TEC. I am in early stage, TEC is testing with a stand alone DC power.

I have understand your description of the "T" colder finger. I have also see some similar designs from CN forum, see below pictures. I will use design 1 to make the TEC outside of the case and also a "T cold finger
Attached Thumbnails
Click for full-size image (30683-1484839655.jpg)
32.4 KB70 views
Click for full-size image (30683-1484839572.jpg)
35.7 KB48 views
Click for full-size image (17944-1488269752.jpg)
50.3 KB46 views
Reply With Quote
  #886  
Old 05-04-2018, 09:40 PM
luka's Avatar
luka
Unregistered User

luka is offline
 
Join Date: Apr 2007
Location: Perth, Australia
Posts: 1,164
Killed my camera.
The ground leg of the temperature sensor broke off when I was assembling the camera and shorted something inside when I connected it to power.
Reply With Quote
  #887  
Old 05-04-2018, 10:33 PM
redriver
Registered User

redriver is offline
 
Join Date: Jan 2018
Posts: 6
...
Is the 5V, 3.3V, 15, -8V and 6V are shortened to GND? I think you should check these power supply to GND resistance without power. One of my friends killed his CAM86 as the VT1(AO3400 or AO3401 I can't remember) is shortened with TEC current >3.5A, FT2232 is damaged (no any device when connected to PC), maybe also Mega328P, no idea whether CCD is killed or not.
Reply With Quote
  #888  
Old 06-04-2018, 12:03 AM
luka's Avatar
luka
Unregistered User

luka is offline
 
Join Date: Apr 2007
Location: Perth, Australia
Posts: 1,164
3.3V is grounded. The other voltages seem OK.

I already tested the regulator and DD2 EEPROM (by cutting tracks on the PCB) and they are fine. I also left the camera plugged in hoping that something will get hot... no luck, only the 3.3V regulator got hot. I have to stop now as I need to get a new regulator tomorrow, I totally butchered this one.

Knowing my luck it will be the FTDI chip, swapping an LQFP64 part will be "interesting".

By the way, I took apart the camera LOTS AND LOTS of times (>50) as I was experimenting with different methods of cooling. Every time the temperature sensor legs were bent and then bent back, no wonder one of them snapped.
Reply With Quote
  #889  
Old 06-04-2018, 11:12 PM
luka's Avatar
luka
Unregistered User

luka is offline
 
Join Date: Apr 2007
Location: Perth, Australia
Posts: 1,164
Removed the FTDI chip with the help of my good old friend, ChipQuik.
No short any more. The replacement chip is coming on Monday.
Reply With Quote
  #890  
Old 08-04-2018, 10:09 PM
redriver
Registered User

redriver is offline
 
Join Date: Jan 2018
Posts: 6
It seems FTD2232 is weak... Hope all other chips are OK.
Remove LQFP is easy if you have a heat gun
By the way, how do you remove the CCD from D70 board? with heat gun or electric iron&desoldering pump? From ICX413 datasheet the soldering temperature should not exceeds 80 degree, we use electric iron&desoldering to remove it from D70 board, and then someone uses heat gun to isolate the CCD from the aluminum board (heat the back of aluminum with about 350 degree and then the glue melts)
Reply With Quote
  #891  
Old 08-04-2018, 10:50 PM
luka's Avatar
luka
Unregistered User

luka is offline
 
Join Date: Apr 2007
Location: Perth, Australia
Posts: 1,164
Nothing else seems shorting to ground. But we will see tomorrow after I swap the FTDI chip.
(I used ChipQuik to remove the FTDI chip instead of a heatgun. Quick and easy.)

Removal of the CCD... two steps (I can't remember in which order).

1. The sensor is glued (epoxied) to the metal carrier so you need to "melt" the epoxy. The usual procedure is to use an DMSO bath for a few days. It will completely melt the epoxy and the sensor will simply come off. It won't damage the sensor. Make sure you bathe the sensor long enough so that no epoxy is left on its back or it can affect the cooling performance later on.
Make sure you read the safety precautions for DMSO.
The usual supplier is ebay or chemists.

2. For desoldering of the sensor Brendan used the desoldering pump (watch for static electricity) on 4 sensors and I used ChipQuik on 3 sensors. They all survived. The recommendation is to keep the sensor as cool as possible so try to be as quick as possible.

Unlike D70, the D40 has a flexible PCB and that one can be removed quite easily, pin-by-pin.
Reply With Quote
  #892  
Old 10-04-2018, 01:15 AM
luka's Avatar
luka
Unregistered User

luka is offline
 
Join Date: Apr 2007
Location: Perth, Australia
Posts: 1,164
Good news, I swapped the FTDI chip and the camera is alive again.
Reply With Quote
  #893  
Old 10-04-2018, 02:18 PM
N_DD
Registered User

N_DD is offline
 
Join Date: Oct 2016
Posts: 33
Luka, happy for you!!!
FTDI chips are my nightmare: I don't remember how many of them I have burned befire learning how to handle them properly...
Reply With Quote
  #894  
Old 10-04-2018, 02:51 PM
luka's Avatar
luka
Unregistered User

luka is offline
 
Join Date: Apr 2007
Location: Perth, Australia
Posts: 1,164
It is a great feeling to have it working again... of course we have clouds and rain now.

Just a quick update about the hardware PWM I have been working on. On one of my cameras it is working perfectly but the other camera it still has some artifacts in the bias images, see the attached image. I have added the black line to show the direction of the artifacts.

My next step will be to take the MOSFET off the PCB and ground it directly from the power plug.

But the cooling works brilliantly with the hardware PWM!!!
Attached Thumbnails
Click for full-size image (35pc power.jpg)
138.4 KB64 views
Reply With Quote
  #895  
Old 26-04-2018, 10:10 PM
redriver
Registered User

redriver is offline
 
Join Date: Jan 2018
Posts: 6
Hi Luka,
Happy to know your camera is alive
Here I have a note that someone in our DIY group said that disassembling the CCD out of its aluminum back with heat gun causes CCD performance degradation, the noise increase. Currently it is better to use chemical method. Also some guy are trying to remove the bayer filter.
Reply With Quote
  #896  
Old 26-04-2018, 10:17 PM
luka's Avatar
luka
Unregistered User

luka is offline
 
Join Date: Apr 2007
Location: Perth, Australia
Posts: 1,164
The chemical method (DMSO) works very well, it just requires patience (a day or two).

Regarding removal of the Bayer filter, few of us have done it. See here or here for descriptions.
Reply With Quote
  #897  
Old 26-05-2018, 02:17 AM
pat30 (Patrice)
Registered User

pat30 is offline
 
Join Date: Oct 2016
Location: France
Posts: 120
Hello my friends,
a quick word to tell you that even if I'm not very talkative for some time, I read your messages.
I have a lot of work now so very busy.

Redriver, the few tests that were made with a black and white sensor are not great, the only advantage is to use filters like to make the false color ex SHO, if not as much keep it original unless you do 2 camera.
Reply With Quote
  #898  
Old 26-05-2018, 11:08 PM
wasyoungonce's Avatar
wasyoungonce (Brendan)
Certified Village Idiot

wasyoungonce is offline
 
Join Date: Jul 2006
Location: Mexico city (Melb), Australia
Posts: 2,336
Oh jeez it’s been ages since I’ve posted or. Done anything I’m so sorry. Been in for 3rd spinal fusion ( 2nd lumber) and at rehab atm . I also fell over the wife’s new blanket box at the end of the bed fractured C5 and prolapsed C5 disc ( all in between surgeries). Had no ide I did damage which of course got worst neck pain hand tingling numb loss of strength

So it’s been a hell ride and more to come. Last lumber fixed my lower issues really well the I dun buggered my neck! Sheee’s talk about bad luck.

Anyway atm things are looking up lower back is great I’ll be home soon then we’ll decide what to do with the neck if anything. Me thinks they have to way too much pain etc.

Anyway some day I hope to return to my bench and resume work on all this ... but it’ll take me ages just to catch up! Oh forgive typos using iPhone at midnight!

Brendan
Reply With Quote
  #899  
Old 26-05-2018, 11:35 PM
luka's Avatar
luka
Unregistered User

luka is offline
 
Join Date: Apr 2007
Location: Perth, Australia
Posts: 1,164
Brendan, welcome back. It is great to hear from you again, we have been worried. At least your lower back has been fixed, your luck finally may be changing. Best wishes for the neck.


I have been testing a new idea for the PWM cooling that should solve the "stripes" issue. A thermally conductive (but not electrically conductive) thin film between the cold finger and the back of the sensor with Arctic Silver 5 on both sides for improved thermal contact. The electrical separation allowed me to ground the cold finger without grounding the back of the sensor and this removed the stripes completely.

I have been testing this solution by imaging over multiple nights and it seems to work well. The sensor temperature is rock-stable.

I still have to quantify the loss in the thermal transfer due to the film before making final conclusions but I don't think it will be very large... hopefully.
Reply With Quote
  #900  
Old 27-05-2018, 12:10 AM
wasyoungonce's Avatar
wasyoungonce (Brendan)
Certified Village Idiot

wasyoungonce is offline
 
Join Date: Jul 2006
Location: Mexico city (Melb), Australia
Posts: 2,336
Stupid question Luka.. did someone try straight DC source (non PWM) to act as cooling supply...ads did the “stripes go away”?

Just guessing if PWM is inducing noise then constant cooling current voltage may fix it.

Really shooting st hip I’ll have to read all posts and get s handle first. Hopefully I’ll be home mid week and be able to devote time and effort.
Brendan...urge sleep time it’s 1am I’m still in hospital!
Reply With Quote
Reply

Bookmarks

Thread Tools
Rate This Thread
Rate This Thread:

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +10. The time is now 06:05 AM.

Powered by vBulletin Version 3.8.7 | Copyright ©2000 - 2024, Jelsoft Enterprises Ltd.
Advertisement
Testar
Advertisement
Bintel
Advertisement