DSLR Active cooling MOD process - Part 2

[MarsG76]

On 08/11/2018 at 23:58, Gina said:

Good thermal insulation of the cold parts makes a lot of difference.

Absolutely.. my next stage is the insulate the heat syncs from the copper plate by inserting a piece of aluminium between the peltier and the copper plate on both side to increase the gap and fill out the gap with first Styrofoam than perhaps expanding foam as a final product is the styro insulates the heat away from the copper and stops it from heating it up....

[MarsG76]

On 09/11/2018 at 00:22, Thalestris24 said:

FWIW my 550d showed temps of 12 to 15 deg (filename) in a batch of 14 x 600s darks. The cooler was preset to +4 deg. So a difference in temp of 8 to 11 deg from the internal value. Unfortunately I don't know for sure what the ambient temp was at the time. They were taken indoors in the afternoon in March 2016 so ambient was probably around 15-20 deg. So clearly, with the cooling on, the actual sensor temperature was quite a bit different from the reported internal camera temperature.

Louise

Apparently the EXIF sensor is a bit of distance from the CMOS so there is slight a difference in reported temperature vs actual temperature so....

On 09/11/2018 at 00:26, Gina said:

I used a digital thermometer device directly on my cold finger close to the image sensor.

.. I placed a NTC sensor directly on the CMOS so this might show a bit more accurate temperature...

 

[Spaceade]

Thanks for documenting this Mars, I've been thinking of doing something similar myself.

[MarsG76]

4 hours ago, Spaceade said:

Thanks for documenting this Mars, I've been thinking of doing something similar myself.

Hopefully it ends up being a success and helps others thinking about this take the plunge into cooling their DSLRs.

[MarsG76]

Hello All...

Thanks everyone for your help and info during my cooling mod of my DSLR.

I do believe that I had a bit of success.

The cooling worked initially but as the heatsink heated up, it in turn heat up the copper plate and as a result the peltier was in a loop trying to cool itself... ultimately the temperature dropped around the 10 degree mark but soon rose 5 degrees and that's where it stabilized.

Even though the temperature drop was not great, I did do a dark frame test...

This was the result of uncooled and cooled ISO1600 darks... 24 vs 19 degrees. I did allow it to cool for 20 minutes before I exposed the cooled sub. The exposure time was 30 minutes on relatively cool night... so there is a bit of a difference in the amount of noise in the raw dark subs.

Eventhough there is a drop in the noise level, its not enough.....

Note that the actual temperature drop of the sensor might be lower than reported since the NTC proble does not sit on the sensor or the cold finger.

 

 

Edited December 8, 2018 by MarsG76

[Stub Mandrel]

If the peltier is getting hot your heatsink is not effective enough.

Use the most powerful fan-cooled heatsink you can get. Typically you need about 0.2C/W or better.

A 0.5C/W heatsink witha 40W peltier will give a temperature rise of 20C above ambient air temperature.

A 0.2C/W will give a rise of 8C.

A 0.1C/W one will give a rise of 4C.

For a rough and ready calculation, that rise has to be subtracted from the cooling effect of the peltier. So if your peltier cools by 25C but you only have a 0.5C/W heatsink you will only get 5 degrees of cooling after dumping the heat from the peltier itself.

This is a 0.5C/W passive heatsink, it roughly fills a cube 120mm per side:

That should convince you that you need lots of fins and lots of fan to effectively achieve adequate heatsinking.

I think you probably need (1) a bigger fan and (2) only to use a single-stage peltier.

 

[MarsG76]

Having to solve the heat feeding back problem, I decided to put another peltier/cheat sink on the other side of the copper plate and styrofoam in between the heatsinks and the copper.

I built a simple circuit board to power to the fans and the temperature controller as well as a relay circuit which supplies power to each of the pletiers through a relay, separating the current flow per veroboard track as initially having the total of 11.8 amps for both peltiers on one track did fry it.

As the heatsink fans are blowing air into the heatsinks, I also added two fans on the top and bottom of the heatsinks to speed up the extraction of the heat within.

 

Now the temperature drop was 10 degrees and stable, as reported by the NTC, but after a while I saw moisture forming on the exposed cold finger so again I'm guessing that the temperature was a lot lower than around the 16.5 degrees that the temperature sensor was showing. This test was performed inside during a warm evening, so for dew to form on the copper I imagine that the temperature was closer to the 5 degree mark.

OK, it was time for another dark frame test, again gain was ISO1600 but this time I'll expose the dark frames for 1800 seconds, which is the time I generally spend per sub for my narrowband imaging. I did another dark sub test with the top fan blowing into the heat sinks for comparison.

The result was substantial....

What this shows is the darkframe comparison uncooled to cooled with, first, the top added fan extracting... than with the added fan blowing into the heat sink. 

This is a comparison to subs on the left which are RAW unprocessed with all of the noise intact. Photoshop CameraRAW does reduce noise as it converts to TIFF, and does a pretty good job of it, as shown in the center, As a final step I pre-process the subs with a "Dust & Scratches" filter to remove the individual pixels which were left by CameraRAW... 2 pixels @ threshold 30 did a good job of only removing the noise/overactive pixels resulting in subs with the noise level shown down the right side. These are the subs I would normally stack.

All of these pictures are center cropped 640x480 from the original files.

The orientation of the top fan made little difference but by having both fans extracting the warm air, it seems to have had the edge and so that how I'll keep the fans for the final version.

 

I started doing some wide angle imaging, mostly as a test. The wide angle subs are 120 second subs at ISO 1600 and so far the results are awesome... my subs are noise free.

Next clear night I'll image some long exposure narrowband subs through one of the scopes.

Since the current state of the setup has cables, the circuit board and temperature controller hanging down and held in place with alligator clips, I'll have to clean up the whole cooled DSLR setup into a nice package with only needing to plug in the power supply. I might also try to move the temperature probe into a place where it'll report a more accurate actual sensor temperature.

 

So far I'm very happy with the results.

 

Clear Skies,

MG

 

[MarsG76]

58 minutes ago, Stub Mandrel said:

If the peltier is getting hot your heatsink is not effective enough.

Use the most powerful fan-cooled heatsink you can get. Typically you need about 0.2C/W or better.

A 0.5C/W heatsink witha 40W peltier will give a temperature rise of 20C above ambient air temperature.

A 0.2C/W will give a rise of 8C.

A 0.1C/W one will give a rise of 4C.

For a rough and ready calculation, that rise has to be subtracted from the cooling effect of the peltier. So if your peltier cools by 25C but you only have a 0.5C/W heatsink you will only get 5 degrees of cooling after dumping the heat from the peltier itself.

This is a 0.5C/W passive heatsink, it roughly fills a cube 120mm per side:

That should convince you that you need lots of fins and lots of fan to effectively achieve adequate heatsinking.

I think you probably need (1) a bigger fan and (2) only to use a single-stage peltier.

 

Yeah, I figured that the heat sink might be too small, but so far the result is OK with the fan overkill shown above.

Another mod I might do later is place a thick piece of copper between the peltier and the heatsink to increase the gap between the heat sinks and the copper plate, and fill the gap with expanding foam, which is (allegedly) the best temperature insulator outside of a vacuum that is easily available.

 

[MarsG76]

On 08/12/2018 at 09:15, Stub Mandrel said:

only to use a single-stage peltier

As I was exposing last night, I noticed that one of the peltiers isn't working, so one peltier was all that was used with a heat sink and fan on the other side... Temperature wise it didn't seem to make so much difference, if any... but I might have had a faulty/open circuit peltier from the beginning...

Do peltiers fail easily?

 

[Stub Mandrel]

9 hours ago, MarsG76 said:

Do peltiers fail easily?

ood question!

I found: " Thermoelectric cooling modules are considered to be highly reliable components due to their solid-state construction"

[MarsG76]

On 09/12/2018 at 21:02, Stub Mandrel said:

ood question!

I found: " Thermoelectric cooling modules are considered to be highly reliable components due to their solid-state construction"

I don't think it is a odd question... regardless to how reliable it's claimed to be I noticed that one side of the cooling system is not warming up, hence the question. Tonight I tested both of the peltiers individually and one was not working, so I guess it's not as reliable as it's cracked up to be.

 

[Stub Mandrel]

2 hours ago, MarsG76 said:

I don't think it is a odd question... regardless to how reliable it's claimed to be I noticed that one side of the cooling system is not warming up, hence the question. Tonight I tested both of the peltiers individually and one was not working, so I guess it's not as reliable as it's cracked up to be.

 

'ood' was meant to be 'Good' not 'odd'!

[Gina]

Reminded me of "The Ood" from Doctor Who ?

[MarsG76]

On 21/12/2018 at 02:15, Stub Mandrel said:

'ood' was meant to be 'Good' not 'odd'!

Than that completely changes the meaning of your comment... The peltier is definitely dead.. I tried to directly power it but its dead..

As I'm currently exposing a wide angle constellation shots in between moonlit/cloudy nights, I do not want to pull the setup apart to investigate the state of the peltier. If it's clear tonight, I collect the final set of data and hopefully on the weekend I can pull apart my cooling system to investigate the reason for the failed peltier. That said, with only one peltier cooling, the ISO1600 subs are all noise free...

 

 

[MarsG76]

Hello astronomers... I have good news and bad news...

I cleaned up the cooling mod, replaced the dead peltier, inserted a NTC probe in a way which touches the copper plate, giving me the temperature of the cold finger rather than the air inside of the camera. I replaced the styrofoam with expanding foam in between the copper and heatsinks to insulate the heat transfer, added plugs to the system for east connection and tested and the cooled DSLR. 

On startup the system worked beautifully, the initial temperature was reported at 27 degrees (summer time here) and it started to fall FAST.. within 30 seconds the temperature dropped to 13 degrees and settled on 14... and stayed there. This is measuring the temperature of the copper and not on the actual sensor, but I though that the proof will be in the pudding" so I setup to capture some wide angle subs, than later DSO imaging (Rosette Nebula) in RGB and HAlpha.

Results are beautiful, there is near no thermal noise... as shown in the attached unprocessed subs, RGB is 120s exposures and the HAplha is 900s both at ISO1600. Center cropped and scaled full frame below them. 

The little noise that there is there is easily removed with single pixel noise reduction... These results were consistent in the 7 or 8 nights of imaging both the wide angle constellation exposures and the Rosette nebula during last two nights.... I was very happy... I say was because I didn't heed @Thalestris24 warning about condensation protection (thinking that dew doesn't conduct) and last the Dew Point was at 19 degrees while the camera was cooled to around 14 degrees and as a result last night my 40D died... Sometimes it powers on, takes a exposure, every so often goes into live view mode but shortly after power up it hangs. I know that the sensor is OK since for the shot while when it works it sees the light but the powered on state doesn't last longer than 20 seconds before hanging.

I'm still hopeful that when it dries out it'll come back to life but otherwise looks like I'm replacing the main PCB. Of course I'll be spraying it with conform acrylic and by the time the spare PCB arrives, I'll insert a thin temperature probe between the sensor and the cold finger which will give me the actual temperature of the sensor.

And so the cooled DSLR adventure continues.

 

Edited January 5, 2019 by MarsG76

[tooth_dr]

4 hours ago, MarsG76 said:

Hello astronomers... I have good news and bad news...

I cleaned up the cooling mod, replaced the dead peltier, inserted a NTC probe in a way which touches the copper plate, giving me the temperature of the cold finger rather than the air inside of the camera. I replaced the styrofoam with expanding foam in between the copper and heatsinks to insulate the heat transfer, added plugs to the system for east connection and tested and the cooled DSLR. 

On startup the system worked beautifully, the initial temperature was reported at 27 degrees (summer time here) and it started to fall FAST.. within 30 seconds the temperature dropped to 13 degrees and settled on 14... and stayed there. This is measuring the temperature of the copper and not on the actual sensor, but I though that the proof will be in the pudding" so I setup to capture some wide angle subs, than later DSO imaging (Rosette Nebula) in RGB and HAlpha.

Results are beautiful, there is near no thermal noise... as shown in the attached unprocessed subs, RGB is 120s exposures and the HAplha is 900s both at ISO1600. Center cropped and scaled full frame below them. 

The little noise that there is there is easily removed with single pixel noise reduction... These results were consistent in the 7 or 8 nights of imaging both the wide angle constellation exposures and the Rosette nebula during last two nights.... I was very happy... I say was because I didn't heed @Thalestris24 warning about condensation protection (thinking that dew doesn't conduct) and last the Dew Point was at 19 degrees while the camera was cooled to around 14 degrees and as a result last night my 40D died... Sometimes it powers on, takes a exposure, every so often goes into live view mode but shortly after power up it hangs. I know that the sensor is OK since for the shot while when it works it sees the light but the powered on state doesn't last longer than 20 seconds before hanging.

I'm still hopeful that when it dries out it'll come back to life but otherwise looks like I'm replacing the main PCB. Of course I'll be spraying it with conform acrylic and by the time the spare PCB arrives, I'll insert a thin temperature probe between the sensor and the cold finger which will give me the actual temperature of the sensor.

And so the cooled DSLR adventure continues.

 

Great result despite the dead camera. You will have this sorted and look forward to the results. 

[MarsG76]

56 minutes ago, tooth_dr said:

Great result despite the dead camera. You will have this sorted and look forward to the results. 

As of now, allowing the camera to dry with both batteries out for 24 hours did not resurrect the camera, same as before... on for a short while than hangs. I need to sus out a cheap main 40D PCB to replace...

[Thalestris24]

 Even distilled water rapidly picks up ions and will eventually conduct... Maybe try drying for longer with the back off? It's possible that it failed for reasons not to do with any condensation. My 550d mono failed after a longish imaging session and I've not been able to resurrect it. As it stands, it's an ex-cooled mono 550d... It powers up but operating the shutter results in it hanging with the red activity/busy led staying on. That indicates a problem acquiring data. Do you get an error number in the viewfinder? Is the camera tethered to a PC?

Hope you can sort it, especially after all your efforts!

Louise

[MarsG76]

9 hours ago, Thalestris24 said:

Even distilled water rapidly picks up ions and will eventually conduct... Maybe try drying for longer with the back off?

That's exactly what I'm doing, I'm still hopeful of it coming back because a couple of months ago my Obsy laptop was rained on and drenched to the point where it did not power up at all.... I took it inside and the next day it powered up and boot up but with a bunch of USB unrecognized errors, PCI-ex not recognized and the WiFi was not working at all, simply not found in the device list. A week later the laptop was back in full operation, USB fully functional, PCI-ex recodnised and WiFi operating.... so there is still hope.  

[MarsG76]

9 hours ago, Thalestris24 said:

It's possible that it failed for reasons not to do with any condensation. My 550d mono failed after a longish imaging session and I've not been able to resurrect it. As it stands, it's an ex-cooled mono 550d... It powers up but operating the shutter results in it hanging with the red activity/busy led staying on. That indicates a problem acquiring data. Do you get an error number in the viewfinder? Is the camera tethered to a PC?

Hope you can sort it, especially after all your efforts!

That red light hanging phenomenon is thing which is happening to my 40D... also when the Live view occasionally comes up, shortly later (if not just frozen camera with a black LCD display) it becomes seqment split lines that resemble a freeze of digital version of an old un tuned CRT TV noise pattern... To me it looks like the main PCB has been compromised and can't process the data without crashing,,, zapped ram due to the short circuit?? The processor was zapped?? either way its all on the main PCB which is the closest circuit board to the cold finger. 

The shutter looks like it still works, the LCD on top is OK, and the back LCD is good too... 

I did win a used 40D on ebay so I can attempt the mod again, but before I pull apart this "new" 40D which looks like its it very very good condition, I'll attempt to replace the main PCB which I found on ebay for $70.

Another item I found is a thin temperature probe, which I can insert between the sensor and the cold finger and get an actual temperature reading of the sensor and with the temperature controller even keep it a degree or two above the dew point if need be.... although with the acrylic conform spray I might not need to worry about this unless my sensor starts to dew over......

Let the mod games continue... ROUND 3, ding ding.... 

[Thalestris24]

11 minutes ago, MarsG76 said:

That red light hanging phenomenon is thing which is happening to my 40D... also when the Live view occasionally comes up, shortly later (if not just frozen camera with a black LCD display) it becomes seqment split lines that resemble a freeze of digital version of an old un tuned CRT TV noise pattern... To me it looks like the main PCB has been compromised and can't process the data without crashing,,, zapped ram due to the short circuit?? The processor was zapped?? either way its all on the main PCB which is the closest circuit board to the cold finger. 

The shutter looks like it still works, the LCD on top is OK, and the back LCD is good too... 

I did win a used 40D on ebay so I can attempt the mod again, but before I pull apart this "new" 40D which looks like its it very very good condition, I'll attempt to replace the main PCB which I found on ebay for $70.

Another item I found is a thin temperature probe, which I can insert between the sensor and the cold finger and get an actual temperature reading of the sensor and with the temperature controller even keep it a degree or two above the dew point if need be.... although with the acrylic conform spray I might not need to worry about this unless my sensor starts to dew over......

Let the mod games continue... ROUND 3, ding ding.... 

Hi

I wish I could tell you what was causing the failure on mine. I did start replacing boards but got distracted and busy with other things. It's been sitting on my table in bits for months...  I have a suspicion the problem is with the sensor but can't be sure. As you know, it's a pain to take a board out and replace it. I might get round to having another go one of these days but I acquired an Atik 383l+ mono back in September so the 550d is a bit redundant now. Still, it would be nice to get it going again. Maybe I'll have quick look tomorrow

Louise

[MarsG76]

12 minutes ago, Thalestris24 said:

Hi

I wish I could tell you what was causing the failure on mine. I did start replacing boards but got distracted and busy with other things. It's been sitting on my table in bits for months...  I have a suspicion the problem is with the sensor but can't be sure. As you know, it's a pain to take a board out and replace it. I might get round to having another go one of these days but I acquired an Atik 383l+ mono back in September so the 550d is a bit redundant now. Still, it would be nice to get it going again. Maybe I'll have quick look tomorrow

Louise

I like to doubt that the sensor would cause this problem since it would most likely just deliver a distorted image instead of freezing the whole camera. I read a bit about electronic, damage and dew online and apparently condensation can destroy a electronic circuit quite quickly, either way I'll get to the bottom of it.. if replacing the main PCB won't do the trick, than I'll mod my other 40D, properly protect with acrylic it this time.

I'm curious to see if you can get your 550D working and to what your camera fault is... yes I'm sure that the Atik made the Canon redundant but (to me) there is something special when creating astro images using a self modded camera, especially when such extreme mods are involved rather than purchasing a already cooled CCD camera. The way I see it that, sure, a dedicated astro imager will deliver better images than a astro modded cooled DSLR when all things are equal but either way none of us will ever match hubble images and a big part of the hobby (to me) is the engineering part of it, not just capturing images but capturing them with a camera I modded (maybe one day a camera I'll create from scratch), but also create electronic circuits such as, for example, a auto close circuit and system for my observatory for when it rains, this is why I'll keep tinkering.

 

[Thalestris24]

2 minutes ago, MarsG76 said:

I like to doubt that the sensor would cause this problem since it would most likely just deliver a distorted image instead of freezing the whole camera. I read a bit about electronic, damage and dew online and apparently condensation can destroy a electronic circuit quite quickly, either way I'll get to the bottom of it.. if replacing the main PCB won't do the trick, than I'll mod my other 40D, properly protect with acrylic it this time.

I'm curious to see if you can get your 550D working and to what your camera fault is... yes I'm sure that the Atik made the Canon redundant but (to me) there is something special when creating astro images using a self modded camera, especially when such extreme mods are involved rather than purchasing a already cooled CCD camera. The way I see it that, sure, a dedicated astro imager will deliver better images than a astro modded cooled DSLR when all things are equal but either way none of us will ever match hubble images and a big part of the hobby (to me) is the engineering part of it, not just capturing images but capturing them with a camera I modded (maybe one day a camera I'll create from scratch), but also create electronic circuits such as, for example, a auto close circuit and system for my observatory for when it rains, this is why I'll keep tinkering.

 

I know what you mean but I didn't mod the 550d myself - I paid someone to do it for me! I don't have the time, patience or eyesight to attempt it myself.

[MarsG76]

7 minutes ago, Thalestris24 said:

I know what you mean but I didn't mod the 550d myself - I paid someone to do it for me! I don't have the time, patience or eyesight to attempt it myself.

Still modded.. still "against the grain" ?

You mentioned that you 550D was mono.. that's one thing I couldn't achieve... I destroyed 5 sensors before giving up on monoing my 40D....

[Thalestris24]

6 minutes ago, MarsG76 said:

Still modded.. still "against the grain" ?

You mentioned that you 550D was mono.. that's one thing I couldn't achieve... I destroyed 5 sensors before giving up on monoing my 40D....

Yeah, monofying is high risk!