1100D cold finger sensor cooling with TEC and water cooling

[Gina]

I've bought this silica gel:

http://cgi.ebay.co.uk/ws/eBayISAPI.dll?ViewItem&item=270968731230&ssPageName=ADME:X:AAQ:GB:1123

It is normally orange and changes to green when it's too wet.

Thanks for that Chris I like the idea of it changing colour - worth paying extra for.

[Gina]

I've started working out how the camera will fit in the box I'm thinking of getting. Here's one drawing showing the box in green and the water block in red, camera and FR parts are in shades of grey. I'll make another drawing showing the rest of the cooling parts later.

[Gina]

This drawing shows the cooling parts. As usual the cold finger is shown in blue, the middle plate in dark red and the hottest, the water block is shown in bright red. The plastic box is in green. A sponge rubber gasket both seals the water block to the box and applies pressure to the Peltier TECs, plate and cold finger to hold it all together.

[nightvision]

That looks really neat. The only thing I noticed was that the warm side of the TEC and some of the warm cooling block (don't know how warm it gets) are inside the box. Might just be a scale/drawing thing and not an issue in the build; if not maybe some insulation round the sides of them would prevent heating of the box air.

[Reggie]

nightvision, that's why i suggested aerogel, I know the slabs of the stuff are insanely expensive but perhaps a small amount of the aerogel pellets or some of the cloth cut so that all of the heat generated by the unit is encouraged away from the sensor, towards 1 face that is left un-insulated.

[Gina]

That looks really neat. The only thing I noticed was that the warm side of the TEC and some of the warm cooling block (don't know how warm it gets) are inside the box. Might just be a scale/drawing thing and not an issue in the build; if not maybe some insulation round the sides of them would prevent heating of the box air.

The water cooling keeps the water block close to ambient temperature - it is very efficient. So only a degree or two above case temperature and the exposed area shown above is small.

Actually though, I may put a copper plate onto the water block to which to attach the MOSFETs for PWM control of the TECs and the LM317T camera power supply regulator. This would stop them heating the air in the box. The MOSFETs will dissipate about a watt at full current (5A) and the LM317T less than that but that would be sufficient to provide some heating in a confined space. I think that with water cooling available I might as well make use of it.

[yesyes]

I don't know what MOSFET you are using, Gina, but mine doesn't even get slightly warm, even when switching both a 60W and a 90W TEC together. So that doesn't need cooling at all. The LM317 is a different story. I decided to use a switch mode regulator instead to prevent heat.

[Gina]

I don't know what MOSFET you are using, Gina, but mine doesn't even get slightly warm, even when switching both a 60W and a 90W TEC together. So that doesn't need cooling at all. The LM317 is a different story. I decided to use a switch mode regulator instead to prevent heat.

The power MOSFETs are IRF540A. The spec says the max Ron is 0.04 ohms so at 5A that gives a max dissipation of 1W by my calculations. Agreed that's worst case.

My 1100D takes a maximum of 200mA (when on long exposure or live view). LM317T is dropping 4.4v, so dissipation is 880mW.

[Gina]

Here is my circuit design for the board that's going in the camera box. Except that the TECs, temperature sensors, switch and pot are not actually on the board, of course

[yesyes]

Ah, I'm using the logic level variant of the same MOSFET, namely the IRL540N. I need the logic level variant because I'm connecting it directly to a 5V Arduino output pin. Are you going to run your 555 at 12V?

Also, interesting that the camera draws only 200mA. I never measured current draw of my 1000D but I expected it to be more.

[yesyes]

Ah, you just answered my question by posting your schematics..

[Gina]

Ah, I'm using the logic level variant of the same MOSFET, namely the IRL540N. I need the logic level variant because I'm connecting it directly to a 5V Arduino output pin. Are you going to run your 555 at 12V?

The power MOSFETs I'm using are the only ones Maplin do. I got them there as there were a few other things I wanted and I got them post free. They seemed quite suitable.

Also, interesting that the camera draws only 200mA. I never measured current draw of my 1000D but I expected it to be more.

Yes, I did too. I measured it because the max quoted current was 2A I think and wanted to see if it was less because the LM317T is rated at 1.5A when on a heat sink. I was surprised at how little current it took. When just standing by with no image processing the current drops to 40mA. My batteries last 3 or 4 hours on a charge so battery operation usually lasts for a whole imaging session but if I'm boxing it in, it needs a permanent supply. It has been a bit tedious changing batteries anyway

One thing I might add to the camera power circuit is an over-voltage crowbar and fuse (not that I've ever had a 317 fail of the several I'm using).

[Gina]

Ah, I'm using the logic level variant of the same MOSFET, namely the IRL540N. I need the logic level variant because I'm connecting it directly to a 5V Arduino output pin.

Where did you buy that version Chris?

[yesyes]

here:

http://www.ebay.co.uk/itm/ws/eBayISAPI.dll?ViewItem&item=180895939940

[Gina]

here:

http://www.ebay.co.uk/itm/ws/eBayISAPI.dll?ViewItem&item=180895939940

Thank you

[Gina]

Here is an amended circuit diagram including camera supply over-voltage crowbar with fuse plus smoothing capacitors to reduce RFI from the TEC switching.

[yesyes]

Gina, quick question. You said you can reach -22C CCD temperature. How long does it take roughly to get to that temperature?

[Gina]

Gina, quick question. You said you can reach -22C CCD temperature. How long does it take roughly to get to that temperature?

I think it was about 10 minutes but I can check the files' time stamps and EXIF T in the name. It only took a few minutes to get below freezing point as I recall.

[yesyes]

Thanks Gina, no need to check exactly. Just wanted to get a rough idea. I was kind of hoping I hadn't waited long enough.. ;-)

[yesyes]

This is a temperature graph of my testing last night. The red line is the temperature measured very close to the cold side of the lower TEC (this is originally the hot side sensor, so it is labelled "Hot" and "Heatsink Temperature") and the blue line is the far side of the cold finger ("Sensor Temperature").

One unit of time on the X axis is 10 seconds, so between 2 separation lines it's 100 seconds, the whole graph is 1000 seconds (almost 17 minutes).

[Gina]

Fed time/temperature data into OO Calc spreadsheet and produced graph. Need to figure out how to convert times in hh-mm-ss format to seconds Also, I can't seem to find any way of making the x-axis spacing according to the time - it just spaces it with equal divisions for each data point. Maybe I need to find another spreadsheet software.

I had posted this rubbish graph but I've now found a way of giving the x-axis roughly equal time intervals and produced a better one - see below.

[Gina]

Had another go at it - typed in the file EXIF T values into a column and made a graph from that so the x-axis is reasonably proportional to the time. Scale is roughly 15 seconds per unit. Note that the y-axis is only to the nearest degree as it's the EXIF T reading rather than from a temperature sensor with higher resolution.

The graph starts after quite a long period of live view where I was setting up the scope to show some distant trees and get them focused. Then I set APT to schedule a run of short exposures (the subject was too bright to use long exposures and I wanted to see when the sensor dewed up and then froze). This is where the graph starts. The first part of the graph shows natural sensor cooling from getting hot running live view to the temperature resulting from the lighter duty of a short exposure every 15 seconds approx. Once the temperature seemed to have stabilised, I turned the cooling on. As can be seen, this resulted in a rapid cooling until the temperature approached equilibrium again.

Total time span is just over 13 minutes.

[nightvision]

I''l probably do a camera cooler thread but I thought you finger builders might be interested in the results of testing and surgery on the latest Aldi cooler-box. It cools down rapidly, from 21c to 1.5c in 25 minutes, that's a lot of air to cool so quickly

It has 3 heatsinks, a switched-mode supply and a NTC monitored cold side, so analogue feedback to a dual op amp and a couple of large MOSFETS and (at first glance) what looks like a discrete oscillator.

The heatsink arrangement may be of interest for cold finger design:

A large hot sink, the TEC, and then a solid (fin-less) 45x45mm (TEC size) block of aly about 25mm thick (surrounded with foam) and then a finned (cold-side) sink about 60% the size of the hot side.

[Gina]

I''l probably do a camera cooler thread but I thought you finger builders might be interested in the results of testing and surgery on the latest Aldi cooler-box. It cools down rapidly, from 21c to 1.5c in 25 minutes, that's a lot air to cool so quickly

It has 3 heatsinks, a switched-mode supply and an NTC monitored cold side, so analogue feedback to a dual op amp and a couple of large MOSFETS and (at first glance) what looks like a discrete oscillator.

The heatsink arrangement may be of interest for cold finger design:

A large hot sink, the TEC, and then a solid (fin-less) 45x45mm (TEC size) block of aly about 25mm thick (surrounded with foam) and then a finned (cold-side) sink about 60% the size of the hot side.

That's interesting, thanks Do you think you could take and post some pics please?

[nightvision]

Yes I will tomorrow.