I've been investigating methods of cooling a DIY refurbished, Astrodon Inside modded 5DMKII (aside from a centralds mod), which led here and there, with mixed results.
This camera makes a brilliant wide field instrument. I love it.
I read somewhere that Spencers use a passive method of cooling the sensor. Basically, bleeding heat to the camera frame - which gave me an idea. Good or bad, is yet to be determined.
As you can see from the images below, I have made a cold plate/finger to fit to the back of the sensor - it will be insulated from the electronics with a moisture barrier.
This particular mod will use a very low power linear driven TEC device installed under the tongue of the cold plate, consuming no more power than my dew heater.
Heat from the other side of the TEC will be conducted to an external active heat sink by a copper strip. The copper strip will be insulated from the camera chassis, held in place by heavy duty double sided tape.
It won't get that hot. Not like the one used on a previous temperature regulated 1000D cold finger, which had a massive heat sink. All it needs to do is take the edge off the dark current without frosting the thing to bits.
The mod could dispense with the TEC by extending the cold plate tongue and fixing it to the camera chassis or extending it outside to an active heat sink, which I think would work very well.
I'm waiting on a 30mm x 15mm x 3.6mm TEC module that will tuck in nicely.
Sounds like a sweet idea using the frame as a heatsink It should coup easily with a low wattage tec. I think to many people go nuts using massive TEC's and trying to dissipate the heat, all thats needed as a small tec to coup with the load thats causing the problem. That dsi pro mod works just perfect a 3watt tec does the job just beautify and using the existing passive setup to cool it. This is it here http://www.iceinspace.com.au/forum/s...d.php?t=101266 in case you missed it rowland, lol ill be going the same route by using a very small TEC if the 1100d mod causes to many issues.
Extremely nicely done, hope to see some results soon and progress come along. I think this would own any cooler box out there and maybe the cheapest and best solution to date to achieve less noise without breaking the bank and going to extream's. As alot of testing has shown sub zero temps really arn't needed in newer sensors and if this takes the edge off enough it could be the best solution ever Lol this could become a diy kit if you got the heatsink laser cut or pressed out XD
I didn't see the DSI - how did I miss that? Yes... my first mod was over engineered.
A kit is a good idea
Agree. Sub zero is not necessary. Just a consistent low temp that doesn't fog up the sensor. Dark scaling used by some processing programs should cope with +/-5C over a nights imaging.
The black spots associated with mismatched darks can be masked by dithering by 12 - 15 pixels while imaging. Seems like the easiest solution.
Moisture barrier for sensor electronics (BluTak pressed around components). I've used this technique before. It is quick, convenient and effective. The cold plate will hold it in place as well.
Fitted cold plate
TEC module - Mistake with my measurements, here. Too wide and blocking the CF card holder. Narrower TEC required. There's a metal retaining tag folded over the plastic sensor filter frame that sticks out 2mm...
At this stage the TEC could be left out. Extending and widening and bending the tongue to contact the chassis, with plenty of contact area, might be adequate. A slim heatsink might do the trick too (40mm x 13mm x 5mm) Perhaps extend outside the body under the CF card to an active heat sink.
Another method might be to machine the whole cold plate from aluminium block with appropriately shaped cooling fins.
A trimmed 30 x 15 TEC. Removed one line of couples and resoldered the +ve wire and plugged with BluTak for testing - finished with silicon. Another option is to use two 12mm square modules (not so readily available) switching both or one for two levels of linear temperature control. I've used a $2 seconds module.
Main board fits nicely in place. TEC wires are a bit of a juggle.
Copper strip to external active heat sink.
Tested at 1.6 volts ~2amps - heat sink only - no fan. Hot side is warm to touch. Cold side bites when touched. This will improve when I get the power supplies sorted out and have a fan running.
12V SMPS will supply 12volt fan and 5A DC-DC adjustable convertor will run the TEC, at 1.5 - 2 volts and ~2 - 3 amps.
EDIT: The $5 ebay Buck convertor, adjustable voltage, will be calibrated to corresponding temperature differentials with reference to an ambient temperature of 20C (room temperature) and corrections applied for ambient temperature variations, applying ROT a table or graph. Setting temperature 2C above the camera internal dewpoint should be relatively easy - it's SMPS. Grounding the camera chassis is a prerequisite as well as good power supply filtering for reducing EMI effects. I don't think this will be too troublesome.
Note: The BluTak moisture barrier will need 10mm wide strips of speed tape on the left and right side for added protection to the tops of SMD components that come close to the backside of the copper plate. They don't touch and as you can see, I have drilled holes to provide clearance to those that sit higher than the sensor backplate.
Finished the TEC installation and added a temperature sensor - it failed and had to be replaced. The replacement was a neater installation than shown here.
Insulated cold plate with speed tape given the proximity of the main board metal covers.
It needs a suitable heatsink. The one I have is too big. Not sure what impact that will have on performance. I suspect the one I am using at 200 watts exceeds requirements by a mile with no added advantage.
At 1.3 volts the TEC pulls just under 2 amps. A further increase in voltage, as suspected, heats the system. So the sweet spot is ~1.25 - 1.35 volts on my multimeter.
A DC-DC convertor which is yet to arrive will provide a little control over temperature control, but not as much as I had imagined. The system is limited by heating of the conducting strip, but this is preferable to using a large external TEC which may risk condensation internally or damaging cold - it's a trade off.
I suspect better performance can be had with an additional and bigger TEC installed along with the heat sink... perhaps just a waste of energy. Alternatively, and possibly a better solution is a lower powered TEC. More energy does not necessarily equate to more cooling with this technology.
I think the TEC is too big at 5A and 4V for the job - 2 amps would have been adequate. The advantage of the conducting strip is that it keeps the air in the camera warm and may assist with defogging of the sensor
Furthermore, I don't know how reliable the location of the temp sensor is in providing an fair estimate of performance. So I will go with the improvement in dark signal as a guide.
Screenshots before and after cooling with dark noise reduction. Left column is cold plate temperature. Right hand is room temp. I haven't evaluated the actual reduction and will get to that later. As far as appearances go, it looks a lot better.
That's it for now.
EDIT: Nearly forgot to mention that the CF door obviously had to be modified. I'm using tape to hold it in place at the moment and have one or two ideas about how to make the card more accessible - WIP.
I couldn't resist trying a second external TEC - 15V 8A at 12 volts, reducing the cold to hot side temperature differential of the internal TEC, which is a measure of performance.
Running the internal TEC at 1.2 volts and the external at 12 volts, reaching 20C differential it was turned off allowing the sensor temperature to drift up to the internal TEC max differential of 8 - 9C. With the external TEC at 10.1 volts (need a different power supply to get a lower voltage) max diff was 15C.
A few more degrees differential would be nice, requiring a fresh look at the dimensions of the copper strip and the rating/performance of the internal TEC.
Great work. I made a heatsink for a Nikon D70 years ago. Unfortunately I damaged something in the process and it was a failed project.
That copper looks quite thick. Is it about 2mm thick? You could even use thinner copper like .7mm and cut it with snips. Or 1.2mm. You can get that from Austral Bronze or hobby stores often have small sections of copper sheet.
I presume the vast bulk of heatsinks around are aluminium because of cost or is it also very efficient? I would have thought a copper one would leave aluminium for dead.
The Buck convertor arrived, with two more on order rated at 6 amps. A cost/time effective way to cater for the various voltage/distribution requirements of my setup. The footprint is very small. Given the steady temp sensor readings these convertors are quiet.
Rated at 5A, input is quoted as minimum 10V - but works down to 8V. Output 0.9 volts minimum, which permits setting the internal TEC at 1.2v (see previous post). An external TEC can be used to control temperature through a second convertor. If small differentials are required in the colder months both convertors can be wound back as required.
In pursuing a linear method of temperature control, buck convertors seem to be the way to go. Essentially, they provided a means of setting a differential, but not a set temperature. As previously explained temperature variations can be handled within reason by dark scaling and any mismatch artifacts dithered out during acquisition of light frames. Aggressive dithering of light frames is a prerequisite for high SNR DSLR images.
The heatsink, fan ON or OFF, handles heat dissipation well, which tells me it can be smaller, providing it's active. Perhaps a relay to turn off cooling in response to fan failure, for protection.
BTW: Tellurex make a 25 x 12mm module only 3.4mm deep rated at 2.5 amps - ideal for the internal TEC application, but I cant find a reseller. Presently, I'm waiting on Tellurex to suggest an alternative shipping method other than the single $97 dollar option for delivery to Aus. Lets see if they come through? The dimensions of this module offer a few more options in terms of room for better insulation of warm/frigid components from the camera body.
Picked up an old stock (Phenom II) Cooler Master for this project. The copper base is 39mm wide which is ideal for a 40x40 TEC. Light enough and small enough not to present a significant balance problem, given its displacement from the optical axis.
Heat sink calculation is probably the most complex part of cooling. Sourcing the right heatsink was a challenge. I think this will do it.
Thermal resistance is probably around 0.3 - 0.5C/Watt, which is OK, - a weight/cost trade-off. It will run about 10C warmer than the Alpha Novatech used on the 1000D cold finger mod.
10C is quoted by Tellurex as an ideal ambient air temperature / heat sink airflow differential. That's fine but needs a 0.16/Watt heatsink for this cooling application. I don't need to cool to that extent. 20C would be great for summer time. Winter requires as little as 6 - 8C differential, because the camera is already cold.
EDIT: While on the subject of heat sinks, it is apparent through testing that the 40mm copper strip is more of a heat repository than heat sink. If using a second external TEC there is more thermal mass to overcome before reducing the hot/cold side differential of the internal TEC, with greater power demands.
A better approach is to reduce the width of the copper conducting strip to that of the internal TEC. With the arrival of the more efficient Tellurex ZMax 25 x 12 TEC (regular post) I will replace the existing copper with a 25mm strip ending in a 40mm square to fit to the heat sink.
The test set-up is an internal TEC constant at 1.2v and an external TEC rated at 12v 6A. Heat sink as in previous post
Rounding, I found that the temperature differential, starting at 8 volts increased by 1C/volt. A 3 minute exposure at each voltage increase, decreased the differential by 1C. ( an acceptable mismatch and manageable with dithering).
The first image compares performance without cooling and temperature rise measured at the bottom left corner of the sensor frame over 240 seconds.
A graph showing temp differential vs voltage and 3 minute exposure effect.
Noise comparison from PixInsight Noise analysis script and noise evaluation during debayering.
EDIT: It crossed my mind that a two TEC arrangement (not stacked, which is usual) has advantages over a single brute force external TEC (which can achieve moderately higher temperature differentials) because the metal in the vicinity of electronic and mechanical components is warmed by the hot side of the internal TEC.
In this installation, the hot side of the TEC is immediately adjacent to the face of the sensor and the rim of the chassis which surrounds the sensor face. There is also a small air gap between the chassis and the sensor.
The Astrodon filter is clear of condensation at -3C, measured at the bottom left corner of the sensor frame. Unfortunately Backyard EOS is not recognising the camera, which prevents reading sensor temperature.
An essential component of cold finger installations is a metal collector plate (practically speaking, the bit that clamps the TEC to the heatsink) on which ice and dew forms, keeping the internals free of the same. This was observed by another DIYr, and given 4 different conversion styles to date, it seems to be correct.
I was running a few more tests the other night after fitting an improved heat sink clamp and graphed the data yet again. Interesting. The original set up was obviously not clamped adequately as the diferential has increased to 19.5C with 1.21v and 11v to the internal and external TECs respectively.
Furthermore the increase in C/volt is closer to 0.6 - 0.5C. So the data in the previous post is rough at best. However, the relationship C/V remains linear 8 - 11volts (still waiting on a second DC-DC convertor to get lower voltage to the external TEC and graph the data further)
Above 11v there is no increase in differential because cooling of the copper strip to the hot side of the internal TEC has reached the optimal hot/cold side ratio, for the power ratings of the TECs and the thermal mass properties - I think?
As mentioned this set up seems to provide advantages over a brute force system, because a condition of cooling by the internal TEC is the production of heat near the camera chassis and the co-location of the TEC and sensor. I have not seen any sensor fogging as yet - localised warming near the sensor face.
I have noticed electroluminescence near the edge of the sensor in dark frames, but can't say whether this is related to the proximity of the cold plate - here
The dc dc converters arrived. Seen here housed in an old webcam box, mounted on the camera base, that swings open for voltage adjustment/temperature control. The third converter will drive a dew heater. All needs testing - hoping to get out tonight and do some wide field shots.
The first image also shows the Teensy temperature test set up. Internal and external temp sensors.
Dam nice work Rowland, has got me motivated to get stuck back into mine tommorow Grab some crops of cooled and un-cooled if you get a chance would be nice to see.
Thanks Jay. It's pointing at Rho Oph right now. At 5C ambient, cooling is off. It was running, but very light condensation became a problem. I need to plot a few more voltages for small differentials, to provide margin above dew point. That will mean winding the internal TEC voltage below 1volt.
I pleased with the low power requirements. At a diff of 12C Toal current is 1 amp or less.
It's a bit ugly in the two TEC configuration. Balance on the mount is a bit precarious.
With the internal TEC set at 1.2 volts, which seems to be its optimum and the external TEC set sequentially from 1 - 7 volts, cooling is less effective > 4 volts. Not published here, but from 8 - 11 volts cooling is ~0.5C/V, after which there is no significant benefit.
Without the external TEC fitted (that is, 0v) the internal TEC max delta is ~8C, which is consistent with the trend shown below. The only way to reduce the delta further (which would have been nice last night with cold temps and dew point near freezing), is to reduce the internal TEC voltage in 0.2 - 0.3v increments and see what gives.
The 8 - 12v data was acquired using a good quality SMPS. The reason for not including it here is that the difference between 7 and 11 volts is just 3C, using a 6amp 40 x 40 TEC. At 8 amps the delta is around 20C at 12volts - marginal at best.
EDIT: Not impressed with the latest batch of dc dc convertors. Not robust at all. 8 volts ~3 amps, rated to 6amps, burnt out. Suggest if buying these, limit input to 7 volts or provide lots of cooling.
This set up only achieves the goal of 10 - 12C differential with use of an external TEC - maximum differential is ~19 - 20C, which is suitable for warmer climates.
It may do better with a good quality internal TEC and may just trip the 10 - 12C goal, without an external TEC; however
5D MKII sensor heating, during long exposures, is considerable and almost defeats the capacity of the internal TEC - differential is around 4C at most.
Heating is required in the vicinity of the sensor face. A 5 x 10mm Nichrome anti dew strip could be attached to the plastic sensor frame where the dust reduction mechanism was attached - another project, when I pull it all apart to improve on some design issues, discussed in an earlier post.
This is the sensor dew heater that fits into the piezo dust reduction receptacle on the filter glass plastic frame. Made up of bits lying around with a touch of red liquid insulation and some thin wire capable of carrying the current.
6 x 100R parallel network ~20R. Voltage supply DC DC converter 0.8V upward. Hot to touch at 4V but plenty of temp control otherwise.
Further to the sensor dew heater in the previous post, the newly installed Tellurex 25x12x3.4mm TEC fits comfortably inside the space available with a very tight fit, using 1mm sheet for the copper finger, fixed to the camera chassis with 5lb external quality wall mounting double sided tape - the black and red roll found at hardware stores.
The Tellurex module though smaller, is speedy. Much faster than the standard ebay TECs. And marginally better differential at 9C @ 2.6V, without an external module to help it along - still not the magic 10 - 12.
Tellurex is ~16US, but international freight, which appears to be non-negotiable, is ridiculously high. I made other arrangements to keep the costs low.
Hope to complete assembly tonight. New moon is not far away, but forecast is rain.
A word on DC DC converters. I learned the hard way... first calculate power requirements, voltage x amps and double that to get the converter rating.
Cool modding a 5DMKII
It should coup easily with a low wattage tec. I think to many people go nuts using massive TEC's and trying to dissipate the heat, all thats needed as a small tec to coup with the load thats causing the problem. That dsi pro mod works just perfect a 3watt tec does the job just beautify and using the existing passive setup to cool it. This is it here 
