A DSLR/Mirrorless Peltier TEC Project

[8472]

As these warm summer nights wreak havoc on uncooled imaging, I thought about designing a non-invasive (no cold finger mods!) and cost-effective way of cooling my cameras and hence reducing the noise a little bit. I'm not quite ready to jump into cooled mono + filter imaging, so I did a little research into what some DSLR imagers were experimenting with.

 

I don't need sub-zero sensor temperatures, as that brings it's own set of issues with moisture and condensation. Just something a little better than present.

 

I thought about peltier (TEC) cooling via an enclosed and insulated aluminium box, and even went as far as ordering the raw materials, but after much thought, I decided against it, as I thought that while feasible for a GEM, it seemed a little bulky for a smaller tracker-type mount.

 

Enter Plan B

 

A small group of imagers have gone down the route of cooling the back of their cameras with TEC cooling devices. This is fine, If the camera in question has a flip out screen, but not much use for other varieties.

 

Another solution, chronicled in an epic CN thread, is to TEC cool the camera via the base and tripod mount screw. Although not as efficient, the beauty with this, is that it can be fitted to a variety of different cameras, as they are all mounted through the common 1/4 UNC tripod mount. Here's the obligatory beer mat sketch.

 

 

My 12v power source feeds a 2KW, 40A PWM Controller ...

 

 

which (where necessary) smoothly throttles the 70w 5.8A TEC's supply voltage...

 

 

and is then cooled with a 90mm HSF.

 

 

The TEC and is then mounted to a 1.2mm copper plate (which I bent to 90 degrees) and drilled a hole to fit the tripod screw. Ideally, I would like to have made the horizontal length shorter than 70mm depicted, to improve the TEC's efficiency, but was constrained by the need to read the display on the back of one camera (T3) and flip the screen out 90 degrees on the other (E-PL5).

 

(0.7mm test plate pictured - not the final product).

 

 

Assembled device with camera fitted.

 

 

Along with an internal modification a performed to my cameras to improve the thermal conductivity from the body to the sensor area, I also lagged most of the exposed copper with closed-cell foam (This is the cold side, and you really have to minimise the loss of cooling as much as you can, as this is where the device is least efficient).

 

As per the initial sketch, I've also fitted a simple temperature probe which sits sandwiched between the copper and the insulating foam (not pictured). This obviously doesn't give you true sensor temperatures, but what you do get is a rudimentary way of matching Lights and Darks temps to a rough roundabout level. Throttling the PWM can fine tune this, to a certain degree -  Pun intended.

 

Coming Soon - Dark Frame Noise Testing...

[+Derbyshire Dave]

That’s astonishing Kev, well done for thinking through all the options and bringing it to a practical conclusion.

[+Bino-viewer]

Hats off Kev, amazing bit of work.

I hope you get some good results from it ?

[8472]

Thanks for the comments, guys!

 

The aim was to keep expenditure and complexity to a minimum and fortunately these TEC kits don't break the bank at all. A good job too - but I'll get to that later.

 

Once you get your head around the concept of how they work, it's not too difficult implement, either.

 

Cheers,

 

Kev

[philjay]

Top job Kev. Cooling certainly helps sensitivity.

[8472]

Cheers, Phil!

 

Before I put the 1.2mm copper version together, I assembled an initial TEC test bed, with the 0.7mm plate, to test functionality of the device.

 

In simple terms, the hot side of the TEC mates to the HSF and the cold side to the copper, which then conducts to the camera.

 

I used some decent quality thermal paste (left over from several PC builds over the years) on the TEC's surface to improve thermal conductivity and fired it up for a test run at full power.

 

In very little time, I could feel the copper plate getting cold to the touch and eventually condensastion and drops of moisture stated to form on the copper plate. My thermal probe had yet to arrive through the post, so a visuals were all i had to go by. Excuse the poor picture.

 

 

Seemed a sucessful test, so on to the full build. I re-assembled the device onto the thicker 1.2mm plate and added some insulation to most of the exposed copper. My DSLR was attached and a series of 300s dark frames were taken, in a dark room.

 

I returned an hour later to the TEC's HSF stone cold and of course the DSLR red hot! Yep, in my haste I had accidentally flipped the TEC the wrong way around and cooked my camera. ?‍♂️

 

I carefully checked the camera over, and touch would, other than being warm to the touch it seemed to be unscathed. Dodged a bullet, I hope.

 

With TEC righted, I repeated the test and checked the dark frames, along with their corresponding EXIF temperatures, to paint a picture of the temperature drop during the course of the test.

Sure enough, there appeared to be a tangible decrease in dark frame noise after each and every frame, along with a steady drop in temps, as the TEC got up to speed.

 

Frame 1 -  Camera Temperature : 44 C

 

 

Frame 2 -  Camera Temperature : 41 C

 

 

Frame 3 -  Camera Temperature : 39 C

 

 

Frame 4 -  Camera Temperature : 37 C

 

 

Frame 5 -  Camera Temperature : 36 C

 

 

Frame 6 -  Camera Temperature : 36 C

 

 

Frame 7 -  Camera Temperature : 35 C

 

 

Frame 8 -  Camera Temperature : 34 C

 

 

Frame 9 -  Camera Temperature : 34 C

 

 

Frame 10 -  Camera Temperature : 34 C

 

 

Frame 11-  Camera Temperature : 34 C

 

 

Frame 12 -  Camera Temperature : 33 C

 

 

An initial 10C drop in temps seem a good return, and I shall be improving things with some insulation and conductivity mods.

 

Predictably, the TEC proceeded to go open circuit not long thereafter, entirely due to my botched previous reversed installation, no doubt.?

 

I have since fitted a replacement (correctly this time) and it ran a full imaging session (600s narrowband subs + full set of calibration frames) without hitch, just as it should. Further tests went smoothly, too. ?

 

Next, I will probably be repeating the test, with a different camera.

 

Clear Skies!

 

Kev.

 

[8472]

Full Spectrum E-PL5 Dark Noise test results

 

10 x 300s Darks taken from TEC switch on (again, these are EXIF sensor temps)

 

Dark Frame 1 - 25.5C

 

 

Dark Frame 2 - 22.6C

 

 

Dark Frame 3 - 20.2C

 

 

Dark Frame 4 - 18.3C

 

 

Dark Frame 5 - 17.0C

 

 

Dark Frame 6 - 15.9C

 

 

Dark Frame 7 - 15.1C

 

 

Dark Frame 8 - 14.4C

 

 

Dark Frame 9 - 14.0C

 

 

Dark Frame 10 - 13.6C

 

 

I didn't perform any more than 10 darks as the battery on my secondary power tank went flat (these TECs can consume a huge amount of juice) but my leisure battery has plenty of reserves though, so no worries there when it comes to all-night imaging.

 

The gradual shrinking of the RAW files in size is also a good indicator of the reduction of noise in the darks, so all in all, I'm satified with the resultant 12C reduction in sensor temps over the course of an hour and the corresponding noise drop.

 

Kev

[BAZ]

That's an impressive bit of shedology there Kev and it certainly seems to make a noticeable difference.

[8472]

Thanks, Martyn.

 

Seems my full spectrum MILC has better read noise levels, and far better sensitivity than my T3, so i'll maybe sticking with that for the time being, for astro.

 

Less weight hanging off the focuser, too.

 

 

[8472]

Another incremental update to the build via an extra bit of copper.

 

 

Doesn't look much, but it appears to lower sensor temperature a good few degrees lower than just the coldplate before. It sits sandwiched between the rear of the camera and the tilt display mechanism.

 

With the lower temps, It seems condensation could be a limiting factor now, as TEC is currently set to running no higher than roughly 75% full power.

 

The round Delrin boss i made screws into the camera tripod base, and the other end is 1/4" female threaded at the base to enable me to fit it to my Star Adventurer L-Bracket, should I so wish.

 

 

 

Camera in position.

 

The black cable at the bottom of the above picture is the temperature probe...

 

And there is just enough room in the PWM Controller chassis to fit the Temperature probe display.

 

 

From the initial sample size, the temperature delta between the EXIF temps and the probe appears to be only roughly 3-4 degrees or so after the latest mods. So good enough for my purposes. ?

Edited September 23, 2019 by 8472