Being the last major league baseball franchise in Canada, the Toronto Blue Jays unify Canadians from coast to coast to coast like no other phenomenon. Even I was swept up in the fervour of last season’s electrifying drive to just shy of a berth in the World Series. And this is from a guy whose interest in baseball extended only to his son’s involvement in Pee Wee softball a decade ago.
The Jays are in an even better position this year to clinch the championship. So perhaps it’s not all that remarkable that one can build a cooled DSLR for less than some unobtanium baseball tickets. But I will show you how it can be done for a budget of under $300 and give you a DSLR capable of producing astronomical images that rival those taken by the best amateur astronomers.
I’ve always been a strong advocate of using a DSLR (digital single lens reflex) camera to image the night sky. Most people already own one for recording family moments so there is no further financial outlay. If you want to modify a DSLR for enhanced performance, used bargains can be easily had on Kijiji and Craigslist. All this activity has not gone unnoticed by the big camera manufacturers, both Nikon and Canon offer models that have modified IR blocking filters to allow improved hydrogen alpha response and models that dispense entirely with the low pass filters for improved resolution and sharpness. Pentax does one better by using a hotshoe mounted GPS and sensor stabilization motors to allow sidereal rate tracking and moderately long night sky exposures with only a standard fixed tripod.
One of the most important characteristics of digital imaging sensors is that they exhibit thermal noise also known as dark current. Heat generated by the densely packed electronic components of a DSLR cause spontaneous generation of electrons within the pixel wells of a digital sensor even if no photon signal has been recorded. This is also why it is referred to as dark current because even if the sensor is kept completely in the dark, noise can still be generated.
Dark current is very well behaved and the amount produced is linearly proportional to temperature and time (length of exposure). Since it’s so predictable it is possible to remove it from an image by performing a dark frame subtraction. But that presupposes that the internal temperature of the DSLR is constant, the same at the beginning of a series of astro exposures as at the end. In reality the tight confines of a DSLR body doesn’t allow quick heat egress and continuous operation of a DSLR, exposure after exposure, causes the DSLR internal temperature to continuously rise from ambient to over 40oC. This results in an inaccurate dark frame subtraction leaving or adding noise to your images. On the other hand, actively cooling the imaging sensor of a DSLR will allow us to stabilize the temperature at a set value which will allow the generation of an accurate dark frame as well as reduce dark noise in general making its elimination easier and more effective.
I started by searching Kijiji ads for a used Canon T2i DSLR, this is a 6 year old camera that I’m familiar with since I own one. I purchased one for $200 including kit lens which I did not need and later resold for $100 so the DLSR effectively costs only $100. Aside from appearance, the shutter actuation count is the best way to determine how used a DLSR really is. Canon makes it very difficult to find this information and you have to purchase third party software that can reveal this number when connected to the DSLR via USB cable. Entry level Canons like the T2i are probably good for 50,000 shutter actuations.
Reviewing the disassembly procedure on websites like Lifepixel.com can be very helpful. Take care with removing the flat flexible cable connectors and use the largest Phillips screwdriver to loosen the screws on the first try otherwise you run the risk of stripping the heads. The T2i exterior shell must be completely dismantled before the main circuit board can be removed and the imaging sensor board accessed.
Figure 1: The main internal components of a DSLR.
With the sensor freed, now is the time to remove the IR blocking, I usually keep the 2nd low pass filter that is integral to the ultrasonic dust shake cleaning system.
Figure 2: The glass components over the CMOS imaging sensor. IR blocking filter is bonded to the primary low pass filter
I then determined that the gap between the back of the surface mounted CMOS imaging sensor and its circuit board was 0.35 mm with feeler gauges that I use to check the valve clearances on my vintage motorcycle. Using a piece of cardboard as a template, I cut and folded a design that would allow a piece of metal to be inserted behind the CMOS chip and fold over the main camera motherboard before exiting the camera in the vicinity of the tripod screw hole. Aluminum is a poor choice of material since it conducts heat half as well as copper. Silver is the best material and 0.5 mm thick sheets can be obtained from local jewelry supply stores. 99% pure copper sheet can sometimes be purchased from roofing companies.
Figure 3: Gap measurement between CMOS imaging chip and circuit board for coldfinger dimensions.
Figure 4: Paper template of silver coldfinger that will be mounted to CMOS sensor circuit board using stock screws, folded over the main computer motherboard and exit the camera over the tripod screw hole.
Figure 5: Silver coldfinger in place. Temperature sensor position is highlighted.
I chose the tripod screw hole as a strong point to attach the heavy Peltier thermoelectric cooling semiconductor and its large heatsink and computer cooling fan which I removed from a portable 12VDC beverage cooler.
Figure 6: Complete cooling system in action and holding sensor temperature at 6oC. Cooling exacts a 4A draw and you would need to dedicate an entire battery to run it all night.
You need a thermostat to keep the imaging sensor at a stable temperature and to also prevent that temperature from slipping below the dew point and having frost form on the optical window of the sensor. A mechanical thermostat is not compatible with the Peltier because the sudden on/off cycles leads to rapid mechanical deterioration of the Peltier. Temperature control is accomplished by pulse width modulation of the Peltier and purchasing a commercially made unit from eBay for only $20 is the most easiest and economical route. The unit comes with a thermocouple style temperature sensor which I affixed as close as possible to the metal coldfinger underneath the CMOS imaging chip.
Figure 7: eBay listing for PWM thermostat for Peltier cooler.
Figure 8: DSLR was programmed to shoot five minute exposures continuously. With the cooling system on dark current attains a stable plateau level. Without cooling, dark current continues to climb.
Figure 9: The proof is in the image details. NGC281 shows no sign of dark noise, space between stars is smooth and clean.