Cooled (or not) DSLR - Real life comparison

[FrenchyArnaud]

Out of boredom (mount out of commission for a while), curiosity (I like to make my own opinion), necessity (I don't have uber money) I endeavoured to DIY a cooled box for modded canon DSLR (I sacrificied a 450D) .  With a few specific things that would decide if it was a success or a waste of time; namely, I wanted to be able to

• set a specific temperature for the sensor, so as to be able to calibrate my darks (that is one of the most often quoted argument AGAINST cooling boxes) 
• cool the sensor 10/12C under functional temperature (as the thermal noise is reduced by half for every 5C of cooling)
• keep under 1.5kg
• be able to SEE the difference on my lights without pixel peeping. 

I am not going to get into the build itself and all the tests I have conducted for the past 2 or 3 weeks, because it would be long and boring, but simply jump to the final result. 
This are two unprocessed, 100% crops. Canon 450D modded on SW130PDS/CG5GT, guided, 600s, ISO800, taken about 1 hour apart.  One is not cooled (ambient temp: 20C, sensor temp: 28C) one is cooled (ambient temp:18C, sensor temp: 8C) 
No points for figuring out which is which. 

It was not a total success because there are a few details that did not pan out as planned (and because I proved to be an imbecile who did not consider that the ambient temp would fall later during the night!) and therefore, more dev is left to undertake BUT I think it is fair to say that for an investment under £150 including the camera, the work in my opinion is totally worth it. 

One of the things I have come to realise during the design/build is that, while everyone seems to consider the temperature of the sensor vs ambient temperature, this is actually irrelevant; what counts is the temperature the sensor reaches when cooled vs when uncooled, regardless of ambient temperature, because uncooled the sensor will get much warmer than the air (typically for this specific dslr, 10C).  And this difference does not have to be massive to see a real improvement : in fact, other comparisons and tests I have made convinced me that cooling by just 6 to 10C make an absolutely massive difference in the "cleanliness" or level of detail of the light frames.  In this case, uncooled : a very noisy image purely due to thermal noise and the infamous ""canon banding" is very visible; cooled by 20C, the light frame is virtually noise free with a very smooth background. 

I hope this will inspire others  

[Rallemikken]

Big difference. Some attach the peltier element to the camera itself (with or without fan - AstroBiscuit did this on one of his hilarious videos), while others make an insulated box, with element, vents and fan (Gary Honis has a couple of really good descriptions out). What's your approach? Image mostly in cold weather, but being a tinkerer, this is tempting!

[FrenchyArnaud]

It's a cooled box "with a twist". Here it is : 

 

Red : current temperature of the heat shield, blue, target temperature, 22.2C = cold heat sink temperature and 23.6C = ambient air temperature. With these and looking at the current local dew point it is possible to set the ideal target temperature, as cold as possible while avoiding condensation/fogging/frost. And because it can be set, it is straight forward to make libraries of darks that are perfectly calibrated. The excellent freeware ExifLog also helps a lot to check temps variations over a series (and was critical during tests to put the shield probe EXACTLY at the right place so that the set temperature = sensor temperature) 

 

 

And here it is on  the telescope : 
 

The 4 cables coming out are : cold cell PSU, DSLR usb connexion, DSLR PSU, ambient air temperature probe

The twist is that the box itself is separated in 2 compartments, the cooled box itself and a separate zone where the electronics are (because along with the wiring they also give off quite some heat) The partition is made of wood because it is easy to work and is a very good thermal insulant.  Beside, it decreases the volume of the cooled zone as much as possible. 

  

You can also see in the middle of the back of the box (flipped over) the temp probe of the thermostat, placed so that it presses against the heat shield of the sensor. The bubble padding adds insulation and acts as a spring to ensure intimate contact. 

The cable that goes inside the cold heat sink is another temp probe that measures the temp of the cold side of the peltier cell (to judge how fast it is cooling / warming up and make sure it is not far below 0C) 

The back of the camera is gone permanently, the fan on the cold heatsink ensure the air circulates straignt within the electronics. Under the heat sink where the USB is connected, and under the cell PSU cable I pack silica gel (2x 20G). As the box is pretty much air tight it prevent condensation very well. 

The camera in its original state heats the sensor up to Ambient +10C. In this config the sensor can be Ambient - 9C. That's -19C and is actually too much as I learned the hard way yesterday (frost on the outer side of the field flattener when the air temp fell to 10C and the cold cell went down to -8.5C, sensor just above freezing) The thermostat allows to maintain a safety temperature without fogging or frost (sensor around 8C) as the field flattener is in effect wormed up by the metal focuser. So the trick will be to set the proper temperature which is 1) above freezing 2) just under the dew point (the sensore being 1 or 2 degrees above the set temperature) Yesterday I f**** up because I just forgot that the temp would drop at night... and I left the thermostat to -50C, forcing the cell full blast for 3 hours.

The current PSU is 12V3A but the peltier is rated for 6A so another psu is on its way. Also, I will ad 8mm neopren foam on the front side to limit heat exchange (the insulation is thiner there to ensure proper square contact of the ring against the front). It's not about getting cold faster but about preventing fast warming when the peltier cuts off at the target temperature: this will allow a more constant temp. It seems that with this specific sensor my ideal target temperature is around 6-8C which should be safe in most situations, except maybe in heatwaves where I'll need 10-15C; but even at 15C, the image is so much cleaner than uncooled that I can absolutely live with whatever little noise is left.  

[Rallemikken]

One mighty project! I'll try something similar when time permits, have the parts lying around, but my approach will most likely be a little simpler. Tempting with all those led's, though.... My first taught was to run it uncontrolled, and deal with the coma corrector with some sort of dew heater around the focuser tube. When I image small targets with my 200PDS I don't use the CC at all, so it's no reason not to go all out.

[jfrijhoff]

Seriously cool (pun intended)! For v1.1, will it be able to keep my beer cold as well 😬?

[iapa]

Hmmm, interesting results, definite improvement.

I built a Peltier cooling unit following Martin Pyott’s project.

I donated the Astro modified 600D before I f=for to use it properly.

Now have a pair of faulty 700D, unmodified. 

In the name of “science” you have prompted me to do a side by side with those, one cooled, same lenses, same target, concurrent exposures.

Don’t hold your breath.

[GTom]

Good stuff! Back in the day I put dry ice packs around my D5100 with great results. That one also catches and freezes all mosture around.

My trouble is with most cooling boxes is weight&size: looking for a solution for my travel rig, these are way too clumsy for a mount already working near limits.

[Dark Raven]

Perhaps a less destructive middle ground solution with smartphone coolers.