I have been asked several times to detail the Mechanical and Thermal Design of the CP30T-EOS DSLR cooler. Note that although it was designed for a Canon 500D, most of these design notes may be used and adapted to any Canon DSLR camera.
Preliminary notes about CP30T-EOS:
- The design that I present here relies on an Aluminum frame that attaches to my custom made T-ring (received a second batch on 9/09) on one end and on the heat sink / TEC assembly on the other. This frame needs to be accurately designed and machined so that once everything is assembled, the heat sink does not apply any force or lever on the sensor.
- CP30T-EOS allows relies on a simple Direct CMOS cooling system that uses a single finger (Cold Finger) and no dew heater. The reason behind this is the use of PWM Temperature Controller. It is very responsive and use Pulse Width Modulation to adjust the power output of the TEC to adjust the temperature. With its accuracy, I can set the temperature above dew point and not be concerned about condensation anymore. It does mean this cooler will NOT bring the CMOS temperature as low but based on my Noise Measurements as temperature drops the reduction in noise gets lower and lower, to a point where (to me) making a more complicated (and risky) design wasn’t worth it. The sensor used to measure the Cold Finger temperature is located exactly where the Cold Finger meets (or enters) the camera body.
The first step of the design was to precisely locate the CMOS sensor in X, Y and Z. To achieve this I had to open the camera twice but it could be done just once if done properly. I used rulers and right angles that I used as references in all directions. Remember that the goal here is to locate the sensor in reference to the T-ring which is used to link the cooling system to the body. This may sound useless but it is an important requirement as the sensor assembly is fragile and is fastened to the body using a set of springs to maintain perfect parallelism with the telescope. If excessive force or lever is applied to the sensor assembly it could potentially lose its parallelism or even get damaged. At the end of this important first step, I want to know precisely the distances in X, Y and Z between the back side of the sensor (the surface that is in contact with the Cold Finger) and the center of the front side of the T-ring.
Additional Note: On the Canon 500D (and like most Canon DSLR), the CMOS assembly is composed of two layers. The first (front) one is the sensor itself while the second one is the controller. To access the gap that is in between, I had to remove the metal ground shield that covers the whole assembly. On the 500D the gap is about 1.40 mm, this is where the Cold Finger will slide in.
Next, I built a small CAD model to design the Aluminum Frame. I modeled all the parts, including TEC, Cold Plate, heat sink, Cold Finger. Working in “assembly mode” makes things easier as you can directly measure the critical distances we’ve defined in the previous step. Then it’s only a matter of execution. The shape of the Frame is somewhat irrelevant and will depend on one’s capabilities to reproduce these shapes (bents, cutouts, etc). When you get to this point, you realize that there is, in fact, only two important dimensions: Z or the distance between the front plane of the T-ring and the back plane of the CMOS sensor; and Y or the vertical alignment between the optical center and the center of the CMOS. These two critical values will dictate the two bends in the Aluminum Frame and the location of the Heat Sink.
Designing a device like this requires a lot of preparation and the main prerequisites are defining Heat Sink, Peltier device, cold plates and cold finger thicknesses. While there is, in my opinion, a lot of flexibility in the choices of these components, defining them early is required to be successful with the whole project because their precise and actual dimensions are required for the design. I am going to give some details on the choices I’ve made with CP30T-EOS:
- The Peltier device I went for is 30 x 30 x 3.80 mm, it has a maximum capacity of only 30 W, a maximum delta of 70 C and a maximum voltage of 15 V. Because of the very local cooling we are applying there will be minor heat loss which we do not need to compensate for, a CMOS sensor has a maximum heat output of only a few W, there is really no reason to use a powerful TEC because it will only make hot plate cooling more difficult. However, maximizing the temperature delta across the TEC will prove beneficial considering the low heat output, I went with the maximum delta I could find.
- I’ve chosen an Aluminum 60 x 60 mm heat sink because they are cheap and light weight and because we use a low power TEC module there is not need for an over sized cooler. The model I chose is a CPU AMD cooler that costs about $7. The image on the right shows a simplified model of the Heat Sink used to cool the Peltier module. Note the two sets of holes to retain both the Aluminum Frame and the TEC/Cold Plate assembly. Also note the pod that I added to recess the frame below the mating surface of the heat sink and TEC.
- The Cold Finger thickness is dictated by the gap between sensor and controller. I went with a 1.2 mm copper plate which is just a little thinner than the gap so that thin electronic insulation material can be used to prevent the cold finger from shorting out the controller.
- The Cold Plate is the part that will be in contact with both the TEC module and the Cold Finger. I went with a 1/8″ thick copper plate. It has to be thin enough to minimize weight and thick enough to remain perfectly flat when fastened onto the TEC and Cold Finger. I didn’t want to attach the Cold Finger directly to the TEC because TEC requires quite a bit of pressure and I was afraid that the thin Cold Finger would bend under such pressure.
With all of these components modeled in CAD, I was able to design a frame that locates the heat sink and TEC exactly where needed. The image below shows the location of the cutout for the Heat Sink in reference to the center of the T-ring front plane.
The great thing about CAD models is that if you are confident with initial measurements, you can and will make something that works. The image below shows the alignment verification. You can see the Cold Finger perfectly aligned with the optical center. Note the dZ measurement of 63.18 mm which is precisely the distance I’ve measured between the back of the sensor and the front of the T-ring.
The view on the right shows the dZ and how all the TEC assembly components are stacked. When all of the components have been decided and modeled, then only variable is the Aluminum Frame which is the most critical part. Also note the Custom T-ring which once more proves a great feature to mechanically attach any custom made cooling system to the camera body. Thanks to it, no force or lever is applied on the fragile CMOS sensor which remains perfectly parallel to the telescope no matter what.
Unfortunately I do not have many pictures to share of the modifications that were made inside the camera body but I will still do my best to detail this important step.
The first modification was the removal of the metal ground shield that covers the CMOS assembly, this was easily done by loosening two screws and prying off a corner of the shield.
Next on the modification list was the cutting of the two metal posts that were in the way of the Cold Finger. This image (click) shows the two metal posts. Notice the gap between the two layers of the CMOS assembly, this is where the Cold Finger slides in. To cut these posts I used a high leverage wire cutter, I didn’t expect this to work as they looked really hard but I guess I just had a very good cutter. If you try this and your cutter does’t seem to do the job, you could use a Dremel but make sure you protect your work first, you do not want any dust or debris in there. You can use a vaccum cleaner close to your cuts while you are working with the Dremel to deal with the dust and debris.
The last modification I had to do was cutting an opening in the soft plastic panel that covers the IO ports. It is very machinable and a very easy modification. There is no risk of debris as you can work on this part alone.
Now it’s time to assemble everything back together. What gave me the most trouble was the putting back one ribbon cable, this specific ribbon cable is partially covered by the body and it took me a bit of time, but with enough patience and calm… it eventually went back in. The mechanical re-assembly starts with the installation of the frame on the custom T-ring which is easily done with four M3 screws. I then installed the Aluminum Heat Sink on the frame, followed by the TEC and Cold Plate without forgetting to install some cut-to-size thermal insulation material around each of these components. I used some 1/8″ thick foam rubber, it works well. I used a torque screw driver (set to 3 in.lbs) to ensure even pressure on each screw holding the TEC / Cold Plate assembly.
This is the moment of truth: After spreading some white (low temperature) thermal compound in the proper locations of the Cold Finger, it is then slid in between CMOS and controller. Note: Do not forget to electronically insulate the top of the Cold Finger as it could short the controller if it was to touch pins and/or soldering points. It went in without any trouble. Judging by the amount of force I had to apply to slide it in, I could have certainly use a little more insulating material to make a better fit, but I decided I should test first and evaluate the performance of the cooling system first. I then gently moved the Cold Finger as I was aligning the holes on the other side. Once the Cold Finger was screwed in I just installed some insulating material around the surface of the Cold Finger that was exposed to air.
This concludes this brief documentation on the design of my CP30T-EOS cooler for Canon 500D.