1100D cold finger sensor cooling with TEC and water cooling

[Gina]

With the two threads of cold finger modification and water cooling both getting long and containing several false starts I've decided to combine the two in one new thread starting off from where I am now.

This thread will cover making a cold finger out of copper sheet to fit in between the image sensor and the PCB it is soldered onto and the camera modifications involved. It will also describe the fitting of a Peltier TEC (Thermo-Electric Cooler) to the cold finger and cooling the hot side of the TEC with a water cooling system. I shall also post the results of noise levels produced by the uncooled camera compared with the full sensor cooling.

Descriptions and photos to follow shortly.

[Gina]

Firstly, here is a rough diagram of the image sensor and it's surroundings, the cold finger, Peltier TEC and water block. The latter is a North Bridge water block which just fits the Peltier TEC.

Key :-

• Blue - Cold finger
  
• Red - Hot side
  
• Green - PCBs
  
• Black - Camera case and frames + USB connector and cable
  

[Gina]

Items bought for camera cooling :-

1. Cold finger - 22swg copper sheet - small piece bought from eBay for a couple of quid.
   
2. Peltier TEC - TEC1-12706 Thermoelectric Cooler Peltier 12V 51.4W :- eBay - The UK's Online Marketplace
   
3. Water Block - ThermalTake CL-W0083 W3 Waterblock for North Bridge | eBay
   

[Gina]

Here are some photos of the camera showing where the cold finger goes. Here's a list of the following photos - there is more info above each photo.

1. Back removed from camera.
   
2. Pointing out the shield over the sensor board that needs removing and where the cold finger goes.
   
3. Side view of the camera showing where the cold finger goes. The cold finger has holes to miss the brackets holding the main circuit board.
   
4. Main board removed and screws removed from sensor assembly ready to fit cold finger.
   
5. Angled view showing brackets that are in the way of the cold finger.
   
6. Side view showing the access for the cold finger. The imaging PCB is green and the cold finger goes just under this and above the sensor, as seen in this photo.
   

Back removed from camera showing main circuit board and shield over sensor assembly.

Pointing out where cold finger goes and the shield that needs removing to allow access for the cold finger.

Side view - cold finger goes in just below the connectors.

Main board removed also screws for sensor assembly.

Angled view showing brackets that need clearance holes or notches in the cold finger.

View of the gap where the cold finger goes with the main board removed. The edge of the green imaging board can be seen and below the gap, the edge of the sensor (dark grey).

[Gina]

Here are various photos of the cold finger and how it fits in the camera. To cater for going through the side of the case a small piece was taken out of the side of the back cover and a section taken off the cover for the connectors. The cold finger itself had holes and notches cut to clear screws and brackets etc.

Firstly, the two sides of the cold finger with holes and notches to miss brackets and screws etc. The insulating film layer can be seen on the back of the cold finger. This ensures that the cold finger does not short out the pins of the processing chip which protrude through the PCB.

Cold finger fitted into sensor unit between sensor and processor board. Thermal paste was applied to the cold finger on the sensor side.

Sensor assembly with cold finger fitted into camera. Plus main board screwed in and connected up together with the shield for the connectors (remote shutter release, USB port, HDMI port).

This photo shows the back cover side and connector cover cut out to miss the cold finger.

The components on the image processing board covered with insulating thermal paste to prevent any condensation causing short circuits.

Finally, the cold finger protruding through the camera side with the 40mm square pad for attaching the Peltier TEC.

[RAC]

The 1100d looks better suited to a cold finger design than the 1000d. The 1000d has little copper clips holding the sensor down that totaly get in the way of that small gap.

[Gina]

The 1100d looks better suited to a cold finger design than the 1000d. The 1000d has little copper clips holding the sensor down that totaly get in the way of that small gap.

I wonder if you could use the cold finger to hold the sensor in place. Maybe a couple of little screws with springs underneath.

[yesyes]

these seem to be ground clips only. I'll just remove them on my 1000D, maybe add an extra ground wire.

I think the 1000D is better suited because the connector board on the left is separate (video out and remote shutter). I'll just not put that board back as I don't need these connectors. That makes it easier to place the cold finger. I even removed th ebrackets holding that small board.

On the 1100D these connectors are now on the main board.

[Gina]

these seem to be ground clips only. I'll just remove them on my 1000D, maybe add an extra ground wire.

I think the 1000D is better suited because the connector board on the left is separate (video out and remote shutter). I'll just not put that board back as I don't need these connectors. That makes it easier to place the cold finger. I even removed th ebrackets holding that small board.

The 450D is similar and yes, I thought it was easier to fit a cold finger - no holes and notches needed for those pesky brackets. In fact it had me stumped for a while - until I worked out that I could fit the cold finger into the sensor assembly then fit the whole caboodle in one assembly, carefully fiddling the cold finger over the brackets and fiddling the ground strap screws behind the cold finger. It would certainly have been easier without those brackets. But a bit of "mind over matter" solved it

On the 1100D these connectors are now on the main board.

Indeed they are.

[RAC]

My peltier(s) is going right on the back of that shield, on the 1000d it doesn't have a hole in it. Every gap between the shield/circuit board and sensor will be filled with heatsink paste.

[Gina]

The next stage was to add a Peltier TEC to the 40mm square pad on the cold finger and arrange for cooling of the hot side. At this point I tried various air cooling options using CPU coolers salvaged from old desktop PCs and where I replaced the standard CPU cooler with something better. A 60mm square heatsink and fan proved insufficient - the heatsink got too hot to touch and little cooling of the camera's sensor was obtained. An 80mm size with a superior design of heatsink proved just about adequate and I was able to cool the sensor down to near freezing. But this cooler weighed 300g - a lot extra to hang on the focuser draw tube. Particularly since I'm planning to add a filter wheel as well.

So I wanted a way of cooling the hot side of the Peltier TEC that was lighter and preferably more efficient. This is where the water cooling idea came to the fore. So then came the search for a CPU cooling system that would cover the 40mm square hot area and not weigh too much. The complete cooling systems appeared to have large, bulky and probably heavy water blocks to attach to the CPU chip so I went for separate components.

A water cooling system consists of a number of parts.

1. A water block on the hot component (usually CPU).
   
2. A radiator with fan to cool the water and get rid of the heat.
   
3. A pump to circulate the water
   
4. A reservoir to hold the spare water and permit topping up.
   
5. Tubes or hoses to connect everything together.
   

At first I selected what seemed a reasonably small and fairly light CPU water block by Zalman - a well known and respected maker of computer cooling parts. Unfortunately, when I received it I found the cooling area was just a disc of 40mm diameter - insufficient to cool a 40mm square. An intervening sheet of copper was suggested but I wasn't happy to add yet more heat conductor and thermal paste into the equation.

In the end I found a North Bridge water block made by Thermaltake that was just slightly smaller than the Peltier unit and only 95g in weight - made entirely of copper. Rather than return the Zalman water block I decided to try this in combination with the 80mm finned heatsink in lieu of a radiator. ie. attaching the water block to the heatsink and using the 80mm fan to cool it. This has proved perfectly adequate.

Further description and photos to follow...

[Gina]

My peltier(s) is going right on the back of that shield, on the 1000d it doesn't have a hole in it. Every gap between the shield/circuit board and sensor will be filled with heatsink paste.

That's a good idea If there's room for it that is. My first thoughts were to have the Peltier behind the sensor with the LCD screen removed but I found that after having everything disconnected for some time the date/time needed setting before the camera would operate. This needs the display and buttons on the camera back. Also with the 1100D you need to have an SD card installed and the software saving to both HD and SD card in the camera otherwise it won't work. After a while the card fills up and needs reformatting before it's possible to take any more exposures. Previously I had thought the screen and control buttons on the back were redundant for AP - unfortunately not It would have been much nicer and neater to have the cooling through the back.

[Gina]

Here's a photo of the North Bridge water block and Peltier TEC on the cold finger. I rather overdid the thermal paste! . The screws, metal plates and knurled knobs were supplied with the water block. Behind the cold finger is a piece of plastic which provides thermal insulation between hot and cold sides.

BTW - The camera is shown with an old Pentax lens from my film days with an adapter ring to use it with the Canon. I use these lenses for widefield and for testing. I have no Canon lenses as the camera is for AP exclusively.

[Gina]

Current water cooling system consists of the following parts.

1. Water block on the TEC hot side - North Bridge water block :- ThermalTake CL-W0083 W3 Waterblock for North Bridge | eBay
   
2. Water cooler - Zalman CPU water block, 80mm heatsink and fan. i might replace this with a proper radiator and fan later.
   
3. Pump - Aquarium water pump low power :- LOW POWER MULTI FUNCTION WATER PUMP - AQUARIUM / FISH TANK | eBaya
   
4. Reservoir - Empty plastic bottle.
   
5. Connecting tubes - Very flexible silicone tubing obtained from farm equipment suppliers (feeding tube) and plastic tubing.
   

Here's a couple of photos of the reservoir, pump and water cooling unit.

[redondo]

HI gina and excellent mod but please be careful smearing thermal paste all over your cameras components as most contain nano particles of metal such as silver,copper or other exotic conductive materials and you can guess what might happen .Good luck and keep up the good work....

Carl.......

[Gina]

Here's a diagram of the cooling system. The reservoir containing the circulating pump is mounted on a piece of wood which is attached to the mount using the polar scope viewing hole. Also mounted on the wood is the water cooling unit consisting of the Zalman water block, heatsink and fan. Copper tube is used to shed some surplus heat to provide a dew heater. The cold finger at the side of the camera has Peltier TEC and North Bridge water block connected by very flexible tubing to the copper tube and the pump. Allowance will be made for moving the camera further away from the scope for a Barlow and also for the rotating joint between scope and mount. Flexible plastic tube provides these functions.

Using a fairly tall reservoir (table sweetener plastic jar) allows for the tilt of the mount and by mounting the whole assembly on the mount polar axis means that the tilt is never enough to spill the water - it just tilts in different directions while remaining just 39 degrees off vertical. This arrangement keeps piping short and free of the rotation of the mount on it's axis - only cables cross this junction.

[Gina]

HI gina and excellent mod but please be careful smearing thermal paste all over your cameras components as most contain nano particles of metal such as silver,copper or other exotic conductive materials and you can guess what might happen .Good luck and keep up the good work....

Carl.......

Thank you And for the warning However, I was very careful to choose thermal paste that was guaranteed totally electrically non-conducting.

[redondo]

Thats good or in future maybe use dielectric grease thats what is use by overclockers when using phase change cooling .Something else for you to look at sometime............

Carl.......

[russellhq]

Thats good or in future maybe use dielectric grease thats what is use by overclockers when using phase change cooling .Something else for you to look at sometime............

Carl.......

Ah yes, DC4 Silicone Compound. That's the stuff I used when trying to freeze processors. Was pretty good from what I remember:

2793695 - DOW CORNING - COMPOUND, SILICONE, DC4, 100G | Farnell United Kingdom

[Gina]

Yes, I was thinking of using silicone grease to seal the camera body.

[Gina]

This is the cooling setup I have ATM. I haven't yet added the dew heater copper tube though I do have it ready. In fact, with fairly low ambient temperatures the water returning from the hot side water block is only a couple of degrees above ambient and I doubt it would have any appreciable effect on dewing. I don't seem to have any dew problems with the ED80 as yet.

[Gina]

I've yet to tidy up all the pipes and wires/cables etc and when I've done that I'll take a photo and post it.

Meanwhile here's a comparison of dark noise cooled and not cooled. I did a test run at various exposures and ISO settings right up to and including a one hour dark sub at ISO 3200 with cooling, resulting in an EXIF T of 1C.

Here's a couple of JPEGs scaled to 1024 pixels wide for here, uploaded to Photobucket and imaged here.

1. 3600 secs = 60 mins = one whole hour at EXIF T = 1C with Peltier TEC & water cooling
   
2. 3600 secs = 60 mins = one whole hour at EXIF T = 28C without cooling
   

[SamAndrew]

Impressive stuff looks like another project to add to the todo list.

I would be interested in seeing some results of a stacked set of lights and darks with and without cooling to see the ultimate results of this mod; do you know of anyone who has done this test ?

Thanks

[russellhq]

Very nice work! Have you considered a small radiator instead of waterblock & heatskink combination? That would be a more efficient solution for removing the heat I would think... though I'm guessing you probably though that but are just using the parts you have.

[angusb1]

That is a very noticeable difference! I'm planning on doing the filter removal mod on mine and might do this one too while its in bits now you've shown us all how.

Thanks!