as RF electronics engineer, only thing I can say is that manual routing is the ONLY way to properly route the layout
I guess I could lay signals I want laid then rely on autoroute...nah! I don't fancy RF...damn backplane on RFELATS was a rats next of solid shielded coax. I still shudder thinking about it.
1st Firmware hex released, many thanks grim, ASCOM handler later
I could not stop thinking about using the 4 PCB mounting screws for the PCB/sensor alignment. I may have come with a solution.
Here is back-of-envelope drawing of the possible solution.
We would need to:
1. Drill a hole all the way through the box and make it large enough to remove the existing thread in the mounting holes
2. Bore a hole from outside with diameter larger than the existing hole and deep enough to fit a rubber o-ring in it
3. Put a screw into the new hole so it sits on the o-ring
4. PCB sits on a spring
5. The screw goes through the spring, through the PCB and is screwed into a nut that is fixed so it cannot rotate. For example, it could be glued to the PCB.
The screw will compress the o-ring when tightened and create a seal. Tightening/loosening the screw (from outside) will change the spacing of that corner of the PCB. We have 4 corners instead of 3 which makes it a bit trickier to do the alignment but it should work.
Longer screws will even allow adjustment of the sensor distance from the glass.
Slow progress on routing. I'll get there...someday. Found some differences and errors between schematic and board. So sometimes laying tracks...ripping them up, moving components points of a mm then re-doing it, more than once, is the work flow order. But, over all a much better layout than cam85m.
Also, so far it's easier to lay these tracks, compared to cam85m, changing this board to a different layout shouldn't be a major hurdle in the future. All that said....I'm not a fan of placing IC on the bottom layer...a few resistors and caps are ok but it gets complicated with ICs etc on base layer. I feel the PCB size is about on the limit...for my meagre skills. Also the board size limits adding things like sensor orthogonality jackscrews.
Having a ulnar nerve decompress surgery on Wednesday...might be not able to do do more on this for a week from then. lets hope I can get most done prior.
Last edited by wasyoungonce; 23-07-2016 at 06:35 PM.
As you said IC on the bottom is not the optimum. But I did not realise that the board is so crowded. I was thinking if we every build a round board, the way to go would be to stack two smaller PCBs. If possible to physically separate power and signal boards...
All the best with the surgery and my best wishes for a quick recovery!!!
it's just minor surgery...already had right arm done and I lived...suspect I'll be back to havoc more destruction...
The board is a little crowded in areas. I was also thinking of splitting the PCBs into 2. One power, the other signal. Double stack them. Not difficult to do, ideal for a round board design. Could even slip in an EMI shield between them.
That's exactly what I was thinking. Build one as it is and then try to improve. We will also have something to compare to.
And if we manage to put all power circuitry on one side of one of the stacked boards we can use the backplane (copper pour) of the power PCB as the EMI shield. Just trying to be cheap
Managed to pick up a broken D70 on ebay for $12 (+ $13 shipping)
Hopefully the sensor is working. No way to test it as now I have D40 (working) and D70 (broken).
(And my daytime-use D90 + cold finger & IR modded Canon 450D + IR modded 1200D. Damn I got quite a collection )
Regarding the humidity absorption, have you seen these? Usually retailing at about $25. Not that they are any better (or cheaper) than silica gel but it may be easier to mount an anti-fog sheet that a bag with small balls inside the camera
Got some new 63/37 solder, .56mm from Mektronics, on sale. I'll need to do some solder practice boards to re-learn eutectic soldering again. I generally use 64/40 due to cost.
Also got some more equipment to get my Vacuum de-solder working 100% again. This will really help de-soldering the sensors. I did some trial vac de-soldering on an old cct board...around a second per leg "bingo" all solder gone. Sony does state (with their ICX413 sensor):
"To dismount an image sensor, do not use a solder suction equipment. When using an electric desoldering tool, use a thermal controller of the zero-cross On/Off type and connect it to ground."
Which I take to mean hand vacuum solder suckers are out proper vac de-solders are ok as long as zero point power switching and earthed tip. Gary asked me about this I said vac de-soldering should be ok. Mine is a proper vacuum de-solder (...a little old) but I'll interweave all the sensor pins with conductive wire (and probably connect to earth) before I attempt de-soldering.
Funny thing about the Sony anit-fog sheet...$5 HN...much more other places! Who'd have thunk it!
I was watching fleabay for Nikons, I saw one "damaged not working" they wanted ~ $60. I wasn't brave enough. Funny enough fleabay has Nikon sensors for sale reasonably cheap.
edit:
I'm also eyeing off some DMSO to "un-stick" the sensor from the carrier. Once used I should be able to send it on to others for them to use.
Reading up Onsemi handling notes on CMOS sensor atm, AN56521. Nothing I'm not aware of though.
Last edited by wasyoungonce; 25-07-2016 at 09:45 AM.
I remember reading somewhere (I read a lot so I don't remember where) that D40 and D50 use flex boards for the sensor mounting, i.e. it is possible to unsolder it pin by pin. D70 is more difficult as it is a normal PCB.
That is one of the reasons I got D40 first and was planing to get another one. But the D70 for $12 + postage was way too cheap to miss, broken or not. As you said usually they go for around $60. My highest bid was actually $40 but nobody else went that high
I doubt that the Ukrainians + people on CN had access to proper vacuum de-solderers. I don't have one either (apart from a hand vacuum solder sucker). I may try Chipquik (the sensor may be too big for it) or copper braid
If the sensor is on a flex then might be easiest to cut away extra flex material or use the flex to pry off the solder as you melt it. Chipquick would be ideal for this.
Ok rough completed PCB. Needs some clean up on layers "restrict" and "keepout"...well a lot, not just some! But finally getting there...where there is ...no idea, the damn image graphic hurts my eyes!
And...no DRC errors! Been fixing them as I go.
Had a lot of warning as I used "0V" in lieu of "GND" so Eagle PCB has a hissy on this but it's not an issue. Guess I should use proper conventions.
Excellent work, let me know what I can do to help. Are we building your or Ukrainian Gerbers?
Now we just have to wait for drivers. Do we have the latest BOM for Cam86 yet?
Just a thought, making the PCB square (no extra "wings" on sides) will let us fit some connectors on side of the box if needed, like grim did in his latest "upside-down" version.
I'd use our gerbers but it doesn't matter. There were some small issues I found with grims/faddy's layout, apart from small things like miss-naming or qty 2 R4's on base layer. The only major deviations were:
DS18B20 temp sensor is powered from 3.3V direct but the PCB had this powered from same 3.3V but after inductor DR6. Not a big issue really. This output from DR6 is also put to a pad "VDD", no reference about this on schematic, but I can do this.
VT1 schematic gate from pin 12 DD5, yet board shows this to be pin 13? Mine is put to pin 12.
DD5 pins 7 & 12 broken out to pads PB6 and PB0. Which I haven't yet done as this PB0 conflicts with VT1 gate pin from DD5. I need to look a little more closely at this.
Apart from these...just need to clean up orphans and layer restrictions. As you said, this also allows us some space for connectors etc.
edit:
Oh DD13 +5V supply was rotated wrong way..or something strange...on their board. Fixed on mine.
I can generate parts lists from Eagle PCB CAD....straight into places like digikey. As long as the parts were fully recognised.
Last edited by wasyoungonce; 25-07-2016 at 07:26 PM.
Still confused about DR6 3.3V feed to R10. This is not an issue really, its just they powered R10 after DR6 inductor, the schematic has it powered straight from 3.3V, not thru DR6 (I have it straight from 3.3V);
They corrected PB6/PB0 via. But these not on schematic nor mine but I will place them as per his board.
Still confused on DD13 +5V LDO. They have it rotated in silkscreen wrt to it's pins. No idea why. They have pins connected ok, as is mine.
They tended to use thinner traces...I tend to oversize most of mine, especially if I had room. The rest looks correct. I'm just laying extra 0V vias between layers and doing some layer restrictions to clean up orphan planes.
I won't be able to do this RCB for a few days from now. But I'll be thinking about it.
Last edited by wasyoungonce; 26-07-2016 at 01:34 PM.
Thanks Luka...I will export mine from eagle and compare. Sorry I assume its a Mega328!
Actually I'll email you my BOM exported from Eagle. Its in CSV format. Haven't checked it but it should be right. Looks terrible in CXv..sent as txt as well. Just open with excel. FWIW...it's not useable without a lot of work.
Last edited by wasyoungonce; 26-07-2016 at 04:26 PM.
By the way, your BOM shows mega328P (note the P at the end).
Grim's BOM shows only mega328 (no P).
The chips have the same footprint but the firmware may be different.
edit: Someone on the Ukrainian forum just posted a link to the sensor sockets from China.
of the possible solution.
