Peltier cooled webcam

[wella]

I was thinking of doing a few experiments with Peltier cooling of a webcam, in fact Ive already started to buy a few bits and pieces that i need.

I'm new to this (not that it will stop me!) so please PLEASE correct my ignorance and mistakes and sweeping statements.... OK off we go...

Now the way I see it the effects of cooling on a webcam are going to be much less than with a dSLR due to the fact that the exposures are likely to be shorter. But it should still be beneficial. Also I can do these experiments and it isn't going to cost me very much. I'm aiming to spend less than £40 on the whole project.

I have been reading all the threads on here about cooling SLRs (Ginas has been most informative), so I think I've got an idea of some of the pitfalls to avoid. Moisture seems the biggest one and the most likely to cause failure.

So far I have bought a Peltier a fan and a couple of heatsinks A & B (ill use one or the other) and of course a webcam of the xbox live type.

I am also going to put a little temperature sensor in the box but I was wondering if anyone thought it would be a good idea to put hygrometer in the box too so i could monitor moisture . Ebay has a joint temperature and hygrometer. photo below. link HERE

Anyone tried doing this before? Anyone got any tip? Comments on my purchases??

Thanks for any advance.

Paul

[SnakeyJ]

Paul - been doing this myself @http://stargazerslounge.com/topic/177027-firefly-mv-mono-is-it-worth-cooling/

Results looking very good with 75% reduction in dark noise so far which would be very noticeable on dimmer targets such as Saturn.

Which ru doing first xbox or asda cam?

Only thing not so sure on is how much driver/software control you have on these cams - just got two xbox cams, but not unboxed yet.

Intrested to see how you get on so keep on posting.

Jake

[wella]

Jake. I had totally missed your thread. Off to get a coffee and have a read. Thanks

[NickK]

From my experience with overclocking PCs with refrigeration you'll need to protect against the condensation. In addition you'll need to form a good heat transmission bond between the cooling block and the camera sensor (probably difficult as the sensor is mounted on the board). Lastly you'll need to keep the sensor clear of ice. You can get electrically insulating aerosol sprays to coat the board to protect against frost.

There's a thermal epoxy that you could use (or you'd need to clamp the cooling block to the target.

[simonmc]

Im about to start the exact same project, but in a custom housing made of 6068T6. So far I have the electronics, but waiting for the alloy which should arrive Tuesday. I am going to use to start with an led dimmer to control the peltier, but may go to a arduino nano for temperature control later if i find its worth while. The only problems i can see at the moment is that the xbox cam is not really designed for cooling and will probably need the whole board cooling rather than the just the CCD. However as the main chip gets warm this cant be a bad thing. Im still not sure if i will sort of encapsulate the back of the board with some sort of epoxy to provide a flat surface to mount to the cold finger, or use some of those pads used on processors to mount it to the cold finger..

It will be interesting to see how much noise is reduced by cooling, and it may seem im going over the top with a custom housing, but i have the kit, and like a few of my projects its an experiment...

Simon

[wella]

Simon - Im thinking about a little box to put things into too. So far im thinking about THIS one but that might change. I am away from home at work at the minute but once i get home and ive got most of the parts ill mock things up for size and make sure things fit before i go spending money on overly expensive boxes.

I do want it to look nice even if it is to be cheap.

Jake - I think ill use the xbox cam but might try and swap it for the asda cam as long as its not too difficult once ive got things sorted.

[NickK]

Duplicate - sorry!

[SnakeyJ]

No probs - hope its useful and look forward to how you crack on with the Xbox cam - it is a really valuable learning experience, though can take up a fair bit of time but is great for cloudy nights!

I thought about making my own set point controller for the peltier, but to be honest at £ 30 including shipping from China I decided this was a simpler/neater solution and my own soldering skills are a little lacking! I think you will find a 40mm fan/block a little small, you need to loose quite a lot of heat from the warm side of the peltier - I started with CPU coolers and will look to come down but think a good 60mm block/fan with heat pipes will be needed or possibly a slighlty bigger low profile block with the fan embedded.

Have fun - I've still got loads of testing/refinements to make and still very much learning!

[simonmc]

Paul,

I used alot of hammod boxes in my time and there quite good and easy to cut/modify. I will be making my own using a few 50mm squares of 20mm billet that i ordered for something else machined to fit in a sort of sandwich affair. I also will be attaching a VGA heatsink and fan i have. That way it will be quite small and with a front that i can attach a tube from an old eyepice on. OTT i know but machining is one of my other hobbies. I will probably also make up a case for an old trust 350 pro i have as well at some point.

Simon.

[wella]

Found a handy 12V supply in the garage which I recycled out of an old DIY audio DAC i built a few years ago.

and ive also had the cam arrive. I really dont know how they can sell them for so little money

[wella]

£3 to get the camera through the door from HERE amazing. Plus they are buy one get 50% off a second

[wella]

Have made some progress on this. A few steps forward, a few back.

I made up a test rig to test the cooling

I didnt really manage to get it very cool however, only managed 19.2 C when i only started at 23.3.

Under the fan that can be seen is a 40*40 mm cooler. Under that is the peltier and under that is some 2 mm copper plate which will become my finger eventually. The temp sensor is already mounted in the box that everything else will eventually be mounted in and around.

I have decided that the problem was the peltier not really being suitable so i have ordered a different one. I will try again when Hong Kong mail has done its thing.

Paul

[SnakeyJ]

Paul,

Try a bigger heat sink/cooler on the warm side of the peltier. My first attempts using a simple 60mm h/s and fan like this yielded a very low delta T and the warm side temp (taken within the heat sink) was running 20-25C above ambient. with a decent CPU cooler I managed to get the warm side temp down to 3.5C above ambient and the cool side down to -5C with the 60W TEC1-12706 starting from 19.5C ambient (total delta T -28C). Also the copper is working as a reverse heatsink - you will need to cut this to size/shape and insulate the areas where you do not require heat transfer.

How do you plan to shape the cold finger around the Xbox Live CCD to get max contact/heat transfer without obstructing the light path? I haven't looked closely at mine yet, but I don't think its possible to get anything in underneath the sensor. There seems to be very little surface area to work with here so this is probably going to be the hardest part:

Picture from Astro-Beano Xbox Mod page - http://astrobeano.bl...-ir-filter.html

[wella]

Jake thanks for your thoughts!

The peltier I initially bought was a 15.7v 25W one. But the power supply im using is a 12v one so already its not operating fully. So ive bought a 12V 60W peltier today from Hong Kong. Hopefully this will give me much more delta T. I am planning to initially try this with the fan and heatsink ive already got but im totally willing to up these too if i need to. From your experiments its looking like i will need to! Today my heat sink wasnt even getting warm.

My cold finger thoughts are still very much up in the air. I totally agree that the size of the bit of copper i had attached today wasnt helping the cooling at all! Its far too big and i will cut it down to size.

As of yet i havent opened the webcam up. Im leaving that for a while to protect from dust. But looking at the photos of a naked webcam it doesnt look like there is any space under the chip. I was thinking of maybe some thermal pads stuck to the surface and a bit of shaping to get maximum contact but i will have to see once i get to that stage.

I actually get the feeling that the cam is going to be the weak link in the chain and i might/will try some different cameras once the cooling is sorted. Maybe even have a play with one of the point grey ones you have been looking into. Im beginning to think that there is no point in cooling if you do not have the software control to take advantage of it!

Paul

[SnakeyJ]

Paul - reckon u should get 80% less noise at high gain if u can get -5C at the ccd, though this really cuts in at dimmer targets where you need max gain or at longer exposures. Not sure if theres an lx mod for this cam, but definitely worth a google.

All good experience and learning though so well worth the time in my book.

Good luck - jake

[Gina]

I think I can add a bit here having looked at cooling webcams in the past and with my experience with cooling DSLR sensors.

I would not reccommend high power Peltier TECs due to the problem of getting rid of the heat they produce themselves and also they are much less efficient that lower power ones at cooling the sensor.

Many webcams have their sensor chips touching the circuit board with no gap in which to put a cold finger. One thought I've had is that if there are no conductors underneath the sensor on either side of the board it might be possible to cut the board away with a Dremmel (other makes are available) - very carefully, of course. But if there are conductors beneath the sensor (or it's a multilayer board with the possibility) then you're scuppered.

There is one webcam I've found that actually has a cold finger already which conducts heat to its outer metal case by curved lugs on the cold finger. This is the Microsoft HD Studio. There is also a flat area of the cold finger about a centimetre square where you could put a tiny TEC. Unfortunately the amount of control you have over this webcam is very limited and it's only capable of short exposures. OK for planetary but no good for DSO imaging. And when I changed over from planetary to DSO imaging I stopped using a webcam for astro imaging.

[wella]

I tried with the higher power peltier.

The start temp was 22.2 C so the result was rather pants.

Gina, thanks for your input. I have followed your "butcherings" of the dSLR's with great interest and will be taking it all to heart. I think that is the avenue i will eventually follow but i wanted to get to know the cooling side of things a bit before pulling apart an SLR.

I think this is about as far as i can push this little set up so i will have to have a re-think. I was hoping not to have to increase the size of the fan and heat sink but it looks like i might have to. ill have a think.

Paul

[SnakeyJ]

Paul,

Out of interest what temps were you seeing on the heatsink?

Also wondering if the mounting bolts are circulating heat back into the cold plate.

Jake

[wella]

I never actually checked the heat sink temp but it wasnt very warm, certainly not hot.

I wondered if the contact between the peltier and the metal was pour on both sides. The heat sink certainly isn't flat, it has a slight bow in it which i have tried to even out a bit with some sand paper but it is still about 1/2 mm off flat.

I read your points (and caths) about heat transfer back to the cool side through the bolts. Personally i dont think it would be a huge issue ( i may be wrong!) The contact between the bolts and the heat sink is minimal because it is only passing through it.

I do have an old cpu cooler which is LARGE which i might butcher down (slightly, it will still be much bigger than this setup) to fit this rig. not going to have much tinkering time for a bit so ill have to do thinking time instead.

Paul

[wella]

I should of said that to counter the metal contact issues I have bathed everything in thermal contact CPU cooler stuff.

[emadmoussa]

Paul - been doing this myself @http://stargazerslounge.com/topic/177027-firefly-mv-mono-is-it-worth-cooling/

Results looking very good with 75% reduction in dark noise so far which would be very noticeable on dimmer targets such as Saturn.

Which ru doing first xbox or asda cam?

Only thing not so sure on is how much driver/software control you have on these cams - just got two xbox cams, but not unboxed yet.

Intrested to see how you get on so keep on posting.

Jake

Sorry, Jake. the link doesn't work. Could you post it again please? Thanks!

[wella]

Should work

[SnakeyJ]

I never actually checked the heat sink temp but it wasnt very warm, certainly not hot.

I wondered if the contact between the peltier and the metal was pour on both sides. The heat sink certainly isn't flat, it has a slight bow in it which i have tried to even out a bit with some sand paper but it is still about 1/2 mm off flat.

I read your points (and caths) about heat transfer back to the cool side through the bolts. Personally i dont think it would be a huge issue ( i may be wrong!) The contact between the bolts and the heat sink is minimal because it is only passing through it.

I do have an old cpu cooler which is LARGE which i might butcher down (slightly, it will still be much bigger than this setup) to fit this rig. not going to have much tinkering time for a bit so ill have to do thinking time instead.

Paul

It's well worth measuring the temperature on the heat sink as this is the reference/start point for your DeltaT, rather than the ambient room temperature. My least successful attempt with a heatsink saw this temprature rise to 23C above ambient and the coolside a very disspointing 1-2C above ambient - DeltaT was still 21-22C, but badly compromised by inadequate heatsink. The thermal interface is also very important to dissipate heat as rapidly as possible - I found arctic silver to be very good, but greasy and messy - which isn't too useful around optics. I tried standard thermal pads (Akasa), but found the performance of these was very poor, so went back to the arctic silver but then sealed everything back up in insulting tape and foam. I also found that a few layers of tape was very effective at holding everything together without screws of clips (dependent on the design of the heatsink, fan and airflow).

I did purchase some thermal Artic Silver epoxy - though not used this or tested as its supposed to be a one time permanent joint, so not ideal for testing purposes!

I took my cooled camera out last night and did some lunar imaging - revealing lots of dirt on the cmos and some issues. The bright new moon isn't a terribly good subject as I was getting a very full histogram with no gain and fastest shutter speed - though it did prove that the setpoint cooling was working nicely down to hold the camera steady at -10C (ambient was just above zero). Due to infocus problems I wasn't able to test at high magnification, but there were no discernable issues caused by vibration of the fan at 1000mm focal length.

Hope to get another go over the weekend at up to 5000mm focal length if seeing permits - and I'll use IR-PASS filters so that I can push the gain and exposure settings up as high as possible! If conditions are good I might even get a go at Saturn, which should prove a far more serious test at around 24-30dB gain.

[NickK]

Careful with arctic silver - it's electrically conductive. However the same company does a ceramic non-conductive paste. As with CPUs, it only requires a very thin smear - not a glob. We used it use a razor blade on top of the CPU core silicon to make a consistent thickness layer.

I would try with non-epoxy first then once you have it working nicely - use the epoxy version (or as I've seen - a non-epoxy+epoxy mix).

Lastly - you can get some spray varnish - cover the electrics with this. It will prevent condensation from shorting things

Ahh them be the days.. nothing like running your CPU faster than it should run and using more power todo it than the cost of buying a faster chip!