A common procedure for cooling a Canon DSLR is to insert a copper cold finger between the back of the sensor chip and the circuit board directly connected to it.
Here is the rmoved 550D sensor assembly:
(http://www.markshelley.co.uk/webdisk/550d_sensor.jpg)
Looking from the side, the gap can be seen:
(http://www.markshelley.co.uk/webdisk/550d_gap.jpg)
The gap is tiny - I can't find my feeler gauges but I estimate the gap to be around 0.3mm This might not be sufficient to conduct away the heat. I'll need dig out my thermal conductivity equations and do some maths.
This might explain why CentralDS do a "Frankenstein" conversion of the similar 600D whereby the cold finger goes around the whole sensor and circuit board assembly:
http://www.centralds.net/cam/?product=cds-600dx-cooled-eos-600d-2
So I might have to go back to the drawing board :-(
Mark
0.3mm doesn't sound like much to work with at all, but getting some copper shim that size will be quite easy.
But to make the most of it the surface finish of the copper will need to be flat lapped to give optimum contact and conductivity.
Quote from: JonH
0.3mm doesn't sound like much to work with at all, but getting some copper shim that size will be quite easy.
But to make the most of it the surface finish of the copper will need to be flat lapped to give optimum contact and conductivity.
When I did the 350D cooling mod I was able to use a 2mm thick copper finger, so 0.3mm will only result in one seventh of the heat transfer.
I seem to remember an alloy that has a greater heat transfer than copper - Can't remember what though :(
Quote from: mickw on Mar 10, 2013, 18:22:54
I seem to remember an alloy that has a greater heat transfer than copper - Can't remember what though :(
Only Silver has a higher thermal conductivity than Copper.
Quote from: Mike
Only Silver has a higher thermal conductivity than Copper.
Cool! How much is a sheet of 0.3mm thick silver?
Not as much as you might think:
http://www.cooksongold.com/Sheet/Sterling-Silver-Sheet-0.30mm-Fully-Annealed-Soft-prcode-CSA-030
According to
http://en.wikipedia.org/wiki/List_of_thermal_conductivities
we have the following thermal conductivities:
Copper 385
Silver 427
Carbon Nanotubes 3180
Graphene 4840
Diamond is potential much higher.
Where can I buy 0.3mm sheets of diamond?
Mark
Quote from: MarkS on Mar 10, 2013, 22:05:19
According to
http://en.wikipedia.org/wiki/List_of_thermal_conductivities
we have the following thermal conductivities:
Copper 385
Silver 427
Carbon Nanotubes 3180
Graphene 4840
Diamond is potential much higher.
Where can I buy 0.3mm sheets of diamond?
Mark
Actually that would also be much cheeper than you might imagine...
Still several hundred at least though.
Diamond machine tools use very thin wafers of diamond bonded to a substrate. There is a limitation on the size though, it is about a 1" diameter i think.
I haven't a clue how they create the wafers though, i did ask a supplier few months ago but they didn't have a clue and just buy them in. Apparently it is a VERY closely guarded secret, or perhaps he was BS'ing...
Anyway, a company like nixon diamonds would prob be able to supply you a wafer.
Quote from: JonH
Actually that would also be much cheeper than you might imagine...
Still several hundred at least though.
Diamond machine tools use very thin wafers of diamond bonded to a substrate. There is a limitation on the size though, it is about a 1" diameter i think.
I haven't a clue how they create the wafers though, i did ask a supplier few months ago but they didn't have a clue and just buy them in. Apparently it is a VERY closely guarded secret, or perhaps he was BS'ing...
Anyway, a company like nixon diamonds would prob be able to supply you a wafer.
Yes, I saw it on the Royal Institution Christmas Lectures just a few weeks ago (The Modern Alchemist - the final episode I think). The lecturer produced a wafer of diamond - it looked just like glass - produced by chemical vapour deposition. Just holding it in their hand they easily cut through a block of ice because it was so incredibly thermally conductive.
So I did know such a thing existed ;-)
He also burnt some diamond in pure oxygen for a laugh.
Mark
Mark,
We have some copper sheet in work that is 0.5mm thick & has had silver deposited on it for improved conduction. I could probably get you a small piece if any good...?
John
Hi John,
Can you get me some pieces............about 7853 12" square please. ;)
Tony G
How about something like this? A thin active heat pipe (http://www.amecthermasol.co.uk/AmecThermasolFlatCoolPipes.html).
(http://www.amecthermasol.co.uk/AmecThermasolImages/AmecThermasolFlatCoolPipesProductShot.jpg)
That is cool (pun intended)!
Thermal conductivity up in the diamond zone. But it is way too fat :-(
I got a feeler gauge from Halfords tonight and the gap is just slightly more than 0.3mm
Quote from: JohnP
We have some copper sheet in work that is 0.5mm thick & has had silver deposited on it for improved conduction. I could probably get you a small piece if any good...?
Thanks for the offer but 0.5mm is just too thick. I'm probably better off getting 0.3mm sheet rather than grinding 0.5mm down to 0.3mm.
Quote from: MarkS on Mar 11, 2013, 23:42:43
I got a feeler gauge from Halfords tonight and the gap is just slightly more than 0.3mm
Have you thought about extending the flexible copper circuit cables (http://www.contour.uk.com/flexi-circuits.htm) and shimming the back board away from the CCD? Would mean rehousing the camera - but you are never going use it off a scope anyway...
Interesting idea.
However I don't think the sensor can be separated from the control board that sits immediately behind it - they form more or less a single unit. In any case it's not something I would want to attempt because of the complexity and the risk of rendering it inoperable.
I'm sure I used 2mm copper sheet, with heat transfer past between the copper and camera chip.
I also sealed the chip PCB with lacquer and went over it with a hot glue gun. I didn't lacquer connectors or push on chips, as the lacquer I used was very runny.
As of yet I have had no condensation problems with the camera.
Quote from: julian
I'm sure I used 2mm copper sheet
The 450D you modified has a much bigger gap behind the sensor.
The gap has shrunk to almost nothing on these latest Canons.
I've done some simple thermal conductivity calculations which indicate that to remove 1W of heat from the area behind the sensor (assuming it's unformly distributed) will require a 5C temperature gradient in the 0.3mm copper strip, which will then obviously result in a 5C temperature gradient across the sensor itself.
But I can't easily determine how much temperature reduction (below ambient) would be achieved by removing that quantity of heat, because there are too many unknowns - especially how much heat is generated by the sensor. and nerby circuitry.
I'll just have to experiment and see.
If you can operate the camera with the circuitry/sensor exposed, I've a laser thermometer you're welcome to borrow.
It might not be sensitive/accurate enough but should show any deviation.
It should be easier than that, just read the EXIF data for the internal thermometer. I think APT records that info.
The temperature recorded in the EXIF is the temperature recorded at some place within the camera body (not entirely sure where). It is not necessarily the same as the temperature of the sensor itself, which directly affects dark current. I did once extract the EXIF temperatures as the sensor warmed up and they were not a good predictor of dark current (except for the first frame or two before the sensor began to heat up).
The best way to measure the thermal gradient across the sensor is to measure the noise at various positions in difference frame between 2 darks and remove the read noise component (using a formula very similar to Pythagoras's Theorem). This has the advantage of directly measuring the quantity that will affect the signal to noise ratio in your final images.
Here's the prototype cold finger created from 2mm and 0.3mm copper strips soldered together with plumbers solder:
(http://www.markshelley.co.uk/webdisk/550peltier1.jpg)
Put into position in camera:
(http://www.markshelley.co.uk/webdisk/550peltier2.jpg)
Screwed in place:
(http://www.markshelley.co.uk/webdisk/550peltier3.jpg)
Camera reassembled. The microphone socket had to be permanently removed to make room:
(http://www.markshelley.co.uk/webdisk/550peltier4.jpg)
Peltier and heatsink added:
(http://www.markshelley.co.uk/webdisk/550peltier5.jpg)
At room temperature of 20C, 1.5A at 12V was required to reduce the sensor temperature to 15C below ambient i.e. 5C
At this temperature I measured the thermal noise in a 5 minute dark frame to be 2.2 electrons i.e. lower than the read noise of 3 electrons. The darks revealed no obvious left/right thermal gradient.
Boosting the Peltier current to 2A reduced reduced the dark current by half again, giving a thermal noise of only 1.5e in a 5min exposure.
Mark have you seen this post may help you get more cooling from the TEC .
Are you going to put the camera in a box as I had to due to condensation?
If you are, I would see if you can do it in a way that you can use a camera lens as well as a telescope as I didn't and wished I had, as I used a 2 " adapter due to the flash unit on top of the camera being in the way.
http://stargazerslounge.com/topic/146756-1100d-cold-finger-sensor-cooling-with-tec-and-water-cooling/
Julian,
That's a very serious mod in the link you posted! Not sure quite how practical it is though.
No, I have no intention of putting the camera in a box - I'd like to use the camera occasionally on the Hyperstar and for that I need the TEC and heatsink to be removable - that functionality has yet to be designed. But I'm intending to replace the heatsink with a low profile copper one.
If condensation (on the imaging window) proves to be a problem I'll add a heating wire loop on the window.
Mark