Home Information Equipment Gallery Weather Links

UV/IR Filter modification and Peltier Cooling a Canon 350D

In December 2010 I decided to modify my trusty Canon 350D by removing the stock UV/IR cut filter and replacing it by a Baader ACF filter that alows 97% light at the important H-alpha and Sulphur-II wavelengths to pass through the filter, compared to only 25% and 15% with the stock filter in place, as is show by the figure below. The camera was dismanteled and the filter replaced using the excellent guides of Ashley Roeckelein (which has unfortunately become unavailable) and David Campbell, and has you end up with the camera looking like pictured below (right). Some other nice impressions on how to replace the 350D filter can be seen on Emiel Kempen's website.

Source: www.baader-planetarium.de

Originally, I only planned to replace the filter and then build myself a whole camera cooling box based on the splendid design of Gary Honis.  But with the camera dismanteled and me doing research on the web into by whole camera cooling and direct cooling of the CMOS imaging device, I started to move more and more to the direct cooling approach. So with the camera already opened up, i decided to go for it. I used the guides by Mark Shelley, Jason C (East Coast Astro Forum), and Sergi Verdugo as a starting point. As a start, the sensor assembly (with the new Baader ACF) had to be remounted back into the camera body, and as I decided to persue the direct cooling approach, the left copper clip that fixes the CMOS imaging sensor to the mounting frame needs to be removed. This way, the cold finger can be moved in from the left side to the back of the CMOS device, using the remaining copper clip as part of the mounting system, underneath which the cold finger will slide to be secured. With this done, I sat the camera aside and started working on constructing the cold finger and heat sink assembly.

The cold finger was cut from a 1 mm thick sheet of copper, with a rectangular shape measuring 60 x 50 mm, attached to another rectangular shape (the part that enters the camera) measuring 28 x 73 mm. Two sets of holes were drilled, one set to accomodate mounting of the heat sink and peltier element, and another to fix the cold finger onto the CMOS mounting frame in combination with the right side copper clip, underneath which the cold finger slides. To ensure no thermal conductance from the heat sink to cold finger, nylon bolts were used for mounting the heat sink to the cold finger. To be able to regulate the cooling, the temperature of the cold finger at the CMOS imaging device needs to measured. I ordered a Mini-Digital_Temperature-Controler from Ebay to be used alongside a PWM design by Sergi Verdugo. As this system is intended for use with home aquaria, the temperature sensor had to be removed from it's waterproof container, which was done by carefull remodeling with a hobby knife, leaving the NTC exposed for mounting to the cold finger. UPDATE: In October 2011 I started having temperature control and as it turned out this was due a malfunctioning temperature sensor. In fact, it had most likely suffered a burst glass cover. The modification to solve this issue is described further down the page. In addition, I also incorperated another different  Mini-Digital-Temperature-Controler, which solves another issue that related to the temperature delay build in the previous one. This resulted in the peltier device being swithed off occassionally in stead of changing duty cycle. In addition, this new one did not require modifying for usen with 12V DC.

The peltier element was salvaged from an old beverage cooler my parents once used on camping. Its an TEC1-12705 unit rated at 50 W, 15 V DC and 5 A. It was mounted in between the cold finger and heat sink using artic silver conductive paste to ensure optimal heat transfer from the hot to the cold side. One thing to be carefull of, is to not overthighten the mounting as this will damage the peltier element, however if it's too sloppy, cooling wil not work as it does need some tension for it to work at all. I overcame this issue by using a set of nylon washers to exactly fill up the space the peltier element takes up between the heat sink and coldfinger, ensuring it can not be overthightened.

As the cold finger will enter the camera at the left (back view) side of the camera where originally a small PCB housing the remote control socket, and A/V output was situated, there was a problem with keeping these connectors available, or at least the remote control jack as this is needed in long exposure mode for astrophotography. There are several options to accomodate this problem. For instance, one could relocate the entire PCB, but this needs elongation of the connector wires and is a not a nice job to do. Secondly, the cold finger could be shaped to allow the connectors to be used, but this needs good metal shaping skills and equipment, both of which I unfortunatly lack. So in the end, I chose to just move the remote jack connector from the PCB, and house it somewhere more convenient, for which I only needed the skillfull soldering of one of my nextdoor neighbours.

To accomodate for a new housing for the remote control connector, I opted to instal the peltier/heat sink assembly in a projectbox connected the the side of the camera, much like Jason C's design. But first, the ensure maximum cooling effect, the cold finger was isolated at the peltier linkage using some left over polystyrene. The remaining opening was cover with some electrical isolation tape, and some channels were carved out to allow both the temperature sensor wire, and a dew prevention system to pass through the polystyrene to reach their respective connectors. Also the temperature sensor (NTC) was installed on the cold finger at a position just next to the CMOS imaging device when mounted back in the camera. The NTC was fixed by some artic silver thermal epoxy adhesive for maximal thermal conductivity combined with epoxy type fixation. Finally the remote control connector was mounted in the project box with some kit, and the slots for relocating the USB 2.0 connection to the camera and the heatsink were prepared.  

The next step was to mount the cold finger/peltier/heatsink assembly in both the camera and project box. This proved to be relatively easy as everything had been carefully measured out in preparation (OK acctually more trial and error fitting and adapting), but in the end it fits near perfect. Thermal conductivity was ensured by applying arctic silver thermal paste between the cold finger and CMOS imaging chip, being carefull not to get the stuf on the connectors of the CMOS chip. OK, despite the electrical tape present to prevent this from happening it did anyhow. I removed this spillage with some 70% alcohol solution until my swabs no longer removed any thermal paste and hoped for the best, it would short out on reactivation of the camera, which would have spoiled a hard days work in a few seconds.  The cold finger was secured to the frame holding the CMOS imaging chip by 2 spare screws originally used to mount some of the plastice cover parts on the entry side of the coldfinger into the camera. To prevent condesation on the PCB directly under the cold finger (where the remote and A/V sockets were originally mounted) some foam was placed between the two. On other parts where the coldfinger is close to PCBs 2-3 layers of duct tape were applied on the cold finger to act as some form of insulation to prevent condesation, as the foam proved to put too much stress on the main PCB for my comfort. 

Finally, it's starting to look a Canon camera again, and it is now time to continue the reassembly of the camera. Firstly, the main PCB is refitted, with its 9 flatband connectors. This is where an extra pair of hands from one of my next door neighbours came in very handy (as well as in some of the soldering that needs to be done to remove a protective cover shield). Next, the back panel is reconnected, but before closing up the camera, the camera is signed with the date of the modification. The camera is then sealed again and the battery and CF-card are refitted, as the tension rises to power up the camera. Will it still function or will we have check all the ribbon cables to see if one is faulty or not. In my case, everything had gone perfect from the start, and so the camera functioned at the first power up. Pfew, no thermal paste left on the CMOS connectors, cleaning them thoroughly seems to have payed off. The next thing to be done was sealing off all openings to the camera/projectbox junction, as well as on the projectbox, to prevent moisture from entering the camera from the outside which may condensate inside te camera, or freeze to the CMOS device.

The next task was to build the temperature controler. First thing to do was to remove the AC/DC converter from the aquarium thermostat and reconnect the power line.  Next up was the soldering PWM board components. Although this was thought to be straight forward, at first me and my neighbour fried both 555 timer ICs (one for the peltier PWM and another for a simple dew controler) as they were wired up in mirror image, D'oh! But when this mistake was corrected by wiring up 2 new 555 timers the designs works very satisfactory, keeping the set temperature usually within 2C from setpoint. As you can see it does however become quite the wire mess, so we put everything away in a nice project box, so it looks all nice and tidy on the outside ;-).

Source: Sergi Verdugo

At last, it was time for some bech testing, although later it turned out I was running the TEC flat out as the 555 timer ICs were already blown on first power up. However, this does provide an excellent view of the maximum capacity of the peltier system. For this bench test, I let de camera take 10 min dark frames at ISO 1600 and monitored the temperature on the cold finger inside the camera, right next to the CMOS imaging device. This experiment was conducted at room temperature of 21.5C, and when the camera started taking 10 min dark frames the cold finger temperature rose to 26.0C after about 3 dark frames, which is something that is known to happen when the CMOS sensor heats up during imaging. There some astrophotographers use interval shooting to let the CMOS sensor cool down between frames, in order to limit the build up of thermal noise. When I kicked in the peltier device, the temparature on the cold finger dropped quickly to -3.2C in about 3-4 minutes and settled at -1.8C after 3 dark frames. This gives the peltier device a maximum delta T of nearly 28C, which is similar to others using the same approach.

The show the effect of cooling on the thermal noise, I have put up a sample of the dark frames taking during bench testing the peltier device. On the left, an uncooled 10 min dark frame at ISO 1600 is shown at ambient temperature of 21.5C (cold finger at 26.0C) is shown, and on the right a cooled 10 min dark frame at ISO 1600 at ambient temperature of 21.5C (cold finger at -1.8C) is shown. The difference is very clear, and except for the ampglow present, the cooled dark frame is very clean and almost looks like a bias frame. Beneath both dark frames, 100% center crops of these darks are shown, to enhance the noise present in both frames.. Conclusion: it has been well worth the work done and me and my neighbours who helped had a drink on the results.

Modifications to resolve apparent issues with temperature regulation.

Due to a broken temperature sensor, my cooling had been out off commission since October 2011. After much trial and error, I finally suspected the temperature sensor. So I devised a plan to incorporate a new one with a metal shell instead of the glass one in the earlier model. To make this change as little invasive as possible I decided to place it at a reachable spot on the cold finger. To be precise, this spot was on the border of the project box that houses the peltier device and the entrance into the camera body. It was fixed in place with thermal conducting epoxy (artic silver), and subsequently covered in silicone. In addition, I also swapped the temperature controller for one that did not suffer from a not by passable timer delay, and that was ready for 12V DV straight out of the box. Only misjudgment was that this one had a blue display, which needed several layers of red foil to restore everybody's night vision during star parties. Despite these improvements, peltier cooling is still giving me some problems during cold winter nights. It seems the PWM disturbs the autoguiding of the NEQ6 mount, however at warmer temperatures this is much less apparent. So during the freezing winter conditions, I just shoot without cooling and use the peltier device to shoot dark frames during the day time. In the warm summer months I can then benefit to the max, with cooling during the night and shooting dark frames during the day. 

All content on astrovirus.nl 2010-2013 by T. Schuurman