Working 1100D cold finger sensor cooling with TECs and water cooling

[ih8usrnames]

First, Gina you are an inspiration!

I recently purchased a t3/1100d for the purpost of hacking but had not gotten around to it when first found your thread; I am now in process of hacking it.

The question that occurs to me at this time is why not attempt to also cool the CMOS mounting bracket?

The chip has metal tabs that connect it to the bracket via screws.

I spent the day creating a template of the CMOS mounting bracket complete with points to indicate screw holes and pins. If I had taps of the appropriate size I might consider making an entirely new bracket from copper and tie it to the coldfinger too.

Initially I thought using a water system was a bit insane but after further consideration it is brilliant. If the housing can be made air tight then it would be possible to purge the oxygen and replace it with Argon or Kypton. Both are poor conductors of heat and dry - no moisture then no problem with dew or frost in the chamber.

The front lens that seals the camera inside could be heated with a few resistors.

Just a thought.

[Gina]

First, Gina you are an inspiration!

Very nice of you to say so - thank you

I recently purchased a t3/1100d for the purpost of hacking but had not gotten around to it when first found your thread; I am now in process of hacking it.

The question that occurs to me at this time is why not attempt to also cool the CMOS mounting bracket?

The chip has metal tabs that connect it to the bracket via screws.

I spent the day creating a template of the CMOS mounting bracket complete with points to indicate screw holes and pins. If I had taps of the appropriate size I might consider making an entirely new bracket from copper and tie it to the coldfinger too.

Interesting point. I have seen a mod where the frame was replaced with a fibreglass board to reduce cooling in the other parts but not copper.

Initially I thought using a water system was a bit insane but after further consideration it is brilliant. If the housing can be made air tight then it would be possible to purge the oxygen and replace it with Argon or Kypton. Both are poor conductors of heat and dry - no moisture then no problem with dew or frost in the chamber.

I don't think it can be made perfectly airtight without a lot of difficulty. eg. I want to be able to change clip filters and that will open the air seal. However, silica gell desiccant seems to keep the air dry once it's had time to work after sealing the box. I've decided not to set the temperature as low as previously thought - something like between -5C and -10C rather than -20C. That low a temperature would only be required for exposures of over an hour at ISO 3200.

The front lens that seals the camera inside could be heated with a few resistors.

Just a thought.

Yes, it could.

[ih8usrnames]

Very nice of you to say so - thank you

Interesting point. I have seen a mod where the frame was replaced with a fibreglass board to reduce cooling in the other parts but not copper.

In retrospect, after messing with it for a couple hours last night I realized making a new frame would be more a pain than helpful. Others have already done excellent work and it appears cooling the sensor with a cold finger is the most effective solution without extreme discomfort.

The dual stage peltier thing is something I was not planning on persuing, I may find I need to. Again, you did excellent work so I should capitalize on your efforts.

I don't think it can be made perfectly airtight without a lot of difficulty. eg. I want to be able to change clip filters and that will open the air seal. However, silica gell desiccant seems to keep the air dry once it's had time to work after sealing the box. I've decided not to set the temperature as low as previously thought - something like between -5C and -10C rather than -20C. That low a temperature would only be required for exposures of over an hour at ISO 3200.

I am not sure why I would ever want to use clip in filters. It seems to me, and I could be wrong, thread on filters are the superior, at least more flexible, solution as they are not camera specific; a 2" filter is pretty much universal in astronomy. In 10 years when we are all buying 600 megapixel 128bit Adaptive Optic Foveon imagers (with built in guiding) we can still use the 2" filters from the antiquated DSLR days.

If it is built right, and if one is able to control remotely all the settings required for astro via USB, then hopefully the seal does not need to be broken more than once every 1-2 years. This also means no cleaning the imager as it is sealed, just need to clean the anti-reflective filter sealing the camera in place. I am sure I am foolishly optimistic.

Tonight I'm reassembling my camera so I can use it this weekend and start sourcing parts for the mod.

I need to investigate using an arduino to monitor temps and power to the peltier.

If anything good comes of my work will post it here.

[Gina]

I must get round to automating the cooling or at least arranging interlocks so that it isn't possible to run the TECs without the water circulating! Trouble is the pump is mains powered and separate from the low voltage circuitry. The computer controlled automation would detect an excessively high temperature at the water block and cut power to the TECs. I think it was this that caused failure of the 12v TEC. I got a couple of 12v 3A 37W 30x30mm Peltier TECs from a UK source rather than Hong Kong and I have replaced the 12v 50W 40x40mm TEC with one of those and have the cooling system working properly again. I can get -10C with the control at just one third of full power, so there's still plenty of cooling.

I shall be trying a 12v 20W and a 12v 37W TEC and see if that's OK. Also, I shall try single TECs of the lower powers and see how well that works. The game isn't finished yet!

[yesyes]

Looking forward to your test results, Gina, to see what's possible with which combination of TECs. Then I can finally continue troubleshooting my cooling system. I must find out why I can't get anywhere near the low temperatures you are able to achieve. Lots of possible reasons in my mind. Main suspect is currently the fact that my test setup is all open-air.

[Gina]

Looking forward to your test results, Gina, to see what's possible with which combination of TECs. Then I can finally continue troubleshooting my cooling system. I must find out why I can't get anywhere near the low temperatures you are able to achieve. Lots of possible reasons in my mind. Main suspect is currently the fact that my test setup is all open-air.

I expect to be continuing the testing shortly. I bought a second hand Zalman Reservator v2 very cheap from eBay and have that set up so that I can do water cooled testing indoors. It's very convenient as it has automatic valves on the ends of the plastic tubes that connect to the main unit and valves in the unit too that shut off and prevent leakage of coolant when disconnected. On the other end I have screw couplings to my water block in the camera box (scope camera). I also have screw connectors on the plastic pipes on the mount so that I can swap over. The weather forecast isn't good for the week so I may well have a go this week. Though I'm still waiting for some very low power 5v TECs, I have 10W and 20W 5v ones and also 12v TECs of different powers - 20W and 37W which I plan to try.

Do you still have your TECs on a longish cold finger outside the body of the camera, Chris? TBH I think the folded CF with minimal length works well. Also, being folded the cold is not lost so much as the back of the CF faces itself and the whole thing is very compact.

[Auntystatic]

One snippet of information I did come across was that the cameras cmos temperature sensor is on the PCB directly behind the cmos sensor its self, these two are now separated by the cold finger and will not give an accurate temperature reading of the cmos sensor, a figure of up to 5deg C difference was mentioned. No biggy I know but interesting information I thought.

[yesyes]

Do you still have your TECs on a longish cold finger outside the body of the camera, Chris? TBH I think the folded CF with minimal length works well. Also, being folded the cold is not lost so much as the back of the CF faces itself and the whole thing is very compact.

Yes. But the temperature sensor on the copper sheet right next to the cold side of the lower TEC has never shown less than -9 °C. Once I got that to an acceptably low temperature (-20 would be good) I will see how I can make the cold finger shorter.

[Gina]

One snippet of information I did come across was that the cameras cmos temperature sensor is on the PCB directly behind the cmos sensor its self, these two are now separated by the cold finger and will not give an accurate temperature reading of the cmos sensor, a figure of up to 5deg C difference was mentioned. No biggy I know but interesting information I thought.

Yes, I agree - interesting I know the temperature sensor isn't in the image sensor - I was under the impression it's in the image processing chip on the imaging PCB. My cold finger wraps right round that so it's probably close to the cold finger temperature and the image sensor will probably be slightly warmer. But EXIF T is the nearest I can get to sensor temperature. The EXIF T does differ from the CF temp by several degrees at times, which is rather strange as the EXIF T reads lower than the thermometer in contact with the CF.

I think it's the relative temperature between one run and another that's important though rather than the absolute temperature. eg. you want the darks taken at the same temperature as the lights. I shall be controlling the CF temperature with the set point temperature control rather than reading the EXIF T as I've designed it. The EXIF T is not available often enough with long exposures to give proper set point control.

[Gina]

Yes. But the temperature sensor on the copper sheet right next to the cold side of the lower TEC has never shown less than -9 °C. Once I got that to an acceptably low temperature (-20 would be good) I will see how I can make the cold finger shorter.

I see. Strange. But as I said before it might be that the TECs you're using are too powerful.

[yesyes]

possibly. I was going to test again with a single TEC. Just too busy with other projects at the moment. There just isn't enough tinkering time to do all the projects I have in my head..

[Gina]

Oh I know what you mean - not enough hours in the day! I have the same problem.

I've brought a camera box indoors to do some more experimenting. One snag - I've mislaid the TECs I've been clearing up - fatal!!

[Gina]

Testing with 5v 20W 25x25mm and 12v 37W 30x30mm TECs and long exposures (currently 2m) and getting

• -10C at half power (equal mark/space on the PWM).
  
• -13C at full power.
  

And that's it until I can find my other TECs When it gets dark and I don't get light leaking into the camera I'll take some darks.

I have the experimental rig set up indoors with cooling provided by the Zalman Reservator v2. Having been running at full power for some time the unit feels tepid. The 120mm fan cooled radiator ran much cooler even with the 20W + 50W TEC combo at full power. Of course the Reservator is completely silent (the fan isn't) and would probably do a very good job of CPU cooling - its intended purpose.

It's nice and dry indoors and I'm not getting any condensation or frost. It's also well over 20C.

[Gina]

Now doing a dark subs run with various exposures up to 1 hour at ISO 3200 and -10C set temperature. Left it running overnight.

[Gina]

Completed the darks run overnight with the EXIF T holding nicely at -10C as set. Exposures of 30s, 1m, 2m, 4m, 5m, 10m, 30m, 60m at ISO 3200. Another time I'll do some at other temperatures. Here are selections of one the 60m darks in JPEG format. Firstly, the full frame resized to 1024px wide and then crops of 1024px wide from the corners, edges and centre.

Full frame resized.

Cropped.

top

centre

bottom

[Gina]

This shows that the cooling and hence the noise is constant over the whole area of the image sensor so I'm pleased with that. Next tests will be at higher temperatures. Think I'll have to sort out where light is leaking into the box and stop it so that I can do the darks in daylight - either on the scope or indoors with the camera body cap on. Of course, with an hour's exposure at ISO 3200 the sensitivity is very high and even the smallest light leak shows up.

[Gina]

I now have the camera box lightproofed Put it in a cardboard box with black polyether foam all round the camera box (recycling packaging that my replacement UPS battery came in). Pipes and cables enter in two corners, split down a bit for the pipes. The PSU and a nice heavy sealed lead acid battery for an UPS hold the top of the cardboard box closed. I've set the temperature to 0C. PWM is currently running at 50% duty cycle. I'll post the results when I have them

Not sure which other temperatures I want to test at. Maybe -5C and +5C but probably not worth doing any others. But I'm open to requests

[Gina]

We have a problem, Houston... After a number of subs the camera is playing up Ran it with EOS Utils instead of APT and it showed Err 70 Had that before as I recall but forget what it was. It isn't anything I've done this time as it was running unattended and it was only when I checked that I noticed it was stuck with APT saying "Busy". Someyhing to look into tomorrow but no long exposures at 0C for now. It was on a 10m sub when it stopped. Curses! One thing it wasn't was overheating - the cooling was working fine.

[Gina]

Found a solution by Googling - http://www.reikan.co.uk/photography/blog/?p=1550 - will try that today.

[Gina]

Opened the box and connected a camera back to the main board so that I could use the camera menu. Didn't need to do anything special - just went through the menu until I found the "Reset Settings" or whatever it's called. Then went back through the menu and set things as wanted. After that it's "all systems go" - working perfectly again

[yesyes]

soooo, you're saying there are a few things that have to be done with the back panel of the camera connected? This was not possible to do via the EOS tools?

[Gina]

soooo, you're saying there are a few things that have to be done with the back panel of the camera connected? This was not possible to do via the EOS tools?

Ah good point - I didn't check that. I'll take a look

[Gina]

Well... Completed a successful dark subs run with no cooling and around 25 to 27C then started a run at around freezing - actually 1C - and that ran for several hours before giving up. Symptom was that the exposure didn't stop at the end of it's time. This was the second half hour exposure. Tried switching off the power and on again and no USB recognition. Opened box and switched off camera at ON/OFF switch. USB responding again but now getting "Err 70" after first exposure. So...

Opportunity to try the EOS Utilities.

I was able to choose Clear Settings from the camera menu and put it in the working window form where I was able to send it to the camera. This did clear the settings but NOT the "Err 70". I am unable to clear the error from EOS Utilities.

[Gina]

Well... it seems the above is irrelevant as the camera has stopped working entirely - dead as a do-do I've tried using a fully charged battery with a back on and connected to netbook. No display, no red light, no computer response from the USB, no nothing! I think it must be a component failure

I can swap the power section for the one from the other dud camera (clutching at straws!) but I don't have a spare working main board - I took a component off the spare to repair this one (successfully). I have a spare imaging assembly and shutter unit but I doubt the problem lies there. Prime candidate must be the main board.

I can get a US version of this camera via eBay from a UK supplier for just over £200 - I see no reason why I should not use the US version.

[Gina]

The camera may have died but the results I managed to get are still relevant...

I didn't manage 60 minutes exposures but I did do 30m ones so here ares the dark frames for 39m at ISO 3200 at temperatures of 29C, 1C and -10C. Firstly, as captured and just resized in PS.

Secondly, having applied Curves to stretch the histogram 4 times and then resized.

Thirdly, after applying curves again with 2x stretch giving 8x altogether.