Adam J's DSLR Water Cooling Thread

[Adam J]

I have decided to create this thread to take anyone who is interested through the design and construction of my DSLR Cool Box over a series of posts.

I made my start in astro-photography / DSO imaging about 18 months back using a Canon 1000D on a 130DPS / HEQ5pro.

I have a very limited budget and so almost every piece of equipment I own is second hand or self built. Living in a yellow zone area I started out using a Astronomic CLS filter and this worked well for me for a short period. However, once I started guiding it became clear that I was limited by light pollution in longer exposures. To that end I decided that I wanted to try narrow band imaging, but I knew that I could not afford a mono CCD or CMOS camera to go with the narrow band filters and that with my DSLR I would suffer from low signal to noise ratio if I attempted narrow band imaging through the Bayer Matrix.

I initially looked into debayering a DSLR in an attempt to get more signal and I did in fact manage to successfully remove the bayer matrix and create a good quality mono sensor.  However, extensive testing convinced me that this was not the way ahead and that I had in fact lost performance overall due to the loss of the micro lenses along with the bayer filters. As a result I decided to focus on the other side of the S/N equation and have a go at reducing the noise through cooling.

All of this of course has been well covered by others in the past, however I would hope that my approach has proven to be a good one with some original design elements and so it sill worth sharing.

I began by researching ‘do it yourself’ DSLR cooling, as I said I am certainly not the first to have attempt this and a wealth of information exists on the internet not least this forum. It was immediately apparent that no two approaches are the same, but it was possible to group DSLR cooling into two main methods both of which make use of Thermal Electric Cooling (TEC) modules, a TEC being a solid state heat pump that uses the Peltier effect to draw heat from one side of the module to the other. The module itself consisting of two ceramic plates sandwiching a semi-conductor matrix. When a voltage is applied across the TEC one side of becomes hot and the other cold. 

Cooling Method 1: The first method is very similar to that used in commercially available CCD cameras and uses a copper plate or ‘cold finger’ in direct contact with the rear surface of the DSLR CMOS sensor to remove heat through conduction. This consists of a copper plate cooled by a TEC which is in turn cooled by a heat sink and fan. While this method is extremely effective in cooling the CMOS sensor it requires significant modification to the camera which carries a significant level of technical risk, problems can also occur with condensation inside the camera body due to the low temperature of the cold finger.

Cooling Method 2: The second method leaves the camera body intact and places it within a ‘cool box’ enclosure (essentially a miniature fridge). This effectively lowers the ambient temperature of the air around the camera which in turn leads to the temperature of the CMOS sensor being lowered. The effects of this type of cooling on noise can be simulated by placing a camera into a fridge on a hot day and taking a long exposure dark frame before and after cooling. While this method is lower risk than the direct cooling method it does come at the expense of bulkier less efficient and less effective cooling. However, I selected this method for my cooling project as the primary goal is to improve performance with minimum expenditure, accidentally destroying a perfectly good DSLR camera would not aid me in this goal. The ability to seal a DSLR within an air tight box would be essential in preventing dew from forming.

 

To be continued:-

[Adam J]

Construction of the Cooling Cell

 

The first stage was to design and construct the cooling cell itself. This would later allow the shell of the cool box to be designed to take into account the dimensions of the cooling cell. In its most basic form the cooling cell consists of a TEC module sandwiched between two heat sink / fan units. When a voltage is applied to the TEC it draws heat from the ‘cold’ heat sink and deposits it into the warm heat sink via the Peltier effect. The cold heat sink and fan sit inside the cool box enclosure and the warm heat sink and fan sit outside the enclosure dissipating waste heat into the environment.

It is essential to understand that for the cooling cell to operate effectively heat must be removed from the hot side of the TEC module. As the temperature of the hot side increases the rate of heat transfer reduces. As a result the higher the power rating of the TEC the larger the heat sink required to cool the hot side. This is further compounded by low efficiency of thermal electric coolers which results in a high proportion of the energy consumed by the TEC being converted into additional waste heat which must also be dissipated alongside heat drawn from within the cool box. In design terms the effect of this is to either limit the power rating of the TEC used or to make the cooling cell extremely bulky.

My solution to this problem and my first design choice was to overcome this design limitation through the use of water cooling as opposed to the more often used air cooling.

Water cooling allows me to make efficient use of a high power TEC while also keeping the weight of the cooler down, but at the expense of portability. However, I was willing to sacrifice almost all portability as it has always been my intent to do all my imaging from home.

The final cooling cell design makes use of a 60mm square heat sink and fan on the cold side of the TEC and an inexpensive 40mm water block on the hot side of the TEC. A 40mm 12volt 10amp (120watt) TEC was chosen to match the footprint of the water block while keeping power draw within practical limits for my power supply (considering that the intent is to solely use the cooler in my back garden battery capacity is not a factor). In order to secure / sandwich the TEC between the water block and the heat sink via the use of a bracket (placed over the water block) two 3.5mm holes were drilled into the bottom surface of the heat sink and threads added using a M4 x 0.75 tap. A further four holes were drilled and threaded in each corner of the heat sink to allow the cooling cell to be secured against the inside wall of the cool box.

A standard water cooling setup makes use of a low voltage water pump in combination with a water reservoir and radiator within a closed loop system. However, as the camera is to be used primarily within a back garden observatory I chose to simplify the setup by providing a continual trickle of cold water through the water block via a custom garden hose attachment.

 

To be continued:-

[Adam J]

A bench test of the cooling cell showed it to be highly effective at reducing the temperature of the cold side heat sink causing a thick layer of frost to form.

The cooling cell was subsequently tested while sealed within a plastic lunch box in order to confirm its ability to reduce air temperature within a semi closed system. Over a 30 minute period a temperature reduction of 17°c below ambient air temperature was achieved, a promising result considering that the lunch box is not insulated.

Further testing demonstrated that only a modest water flow was required in order to effectively cool the hot side of the TEC. As a result I chose to reduce the diameter of the piping used from 8mm to 4mm greatly increasing the flexibility of the pipe work. A significant build-up of condensation within the test box confirmed that it would be essential to create an effective air tight seal around the camera in addition to making effective use of desiccant to reduce humidity. Failing to do this would, at best, result in ice / condensation on the sensor and at worst cause damage to the electronics of the camera.

To be continued:-

 

 

 

[Adam J]

Cool Box Enclosure Design and Fabrication

 

I set myself the aim of making the cooling box enclosure lightweight, airtight, well insulated and compact. Depending fabrication skill level it is entirely possible to make use of commercially available enclosures such as electronic project boxes, lunch boxes, biscuit tins or electrical junction boxes; however, it is unlikely that any of these solutions will deliver optimal cooling performance. With this in mind I chose to design and fabricate my own enclosure to tightly fit the dimensions of the DSLR and cooling cell. Accurate measurements were taken of both the DSLR and cooling cell, particular attention was paid to the relative location of the T-ring and tripod attachment thread on the bottom of the camera which was used to secure the camera within the cool box.

Using these measurements a technical drawing was produced of the panels making up the shell of the cool box using a PC drawing package. When this initial design was completed the drawings were cut out into thick construction card prototype using a CO2 laser cutter. However if you dont a laser cutter its easy to print a design onto paper and then use it as a template. This prototype enabled my design to be tweaked without the need to fabricate the design from more expensive materials.

To be continued:-

[tooth_dr]

Good write up. I've the same camera so am interested in this. 

[al-alami]

Considering that I am in the process of trying to work on a cooling system, I have to say this is a very interesting thread, looking forward to seeing how well it works. (was actually thinking yesterday that I have the parts from a computer water cooling system, maybe I can get them to work!) 

Keep up the good work.

[Adam J]

Once the design was finalized the cool box panels where cut out in 2.5mm ABS plastic using a CO2 laser cutting machine. However, as with the card if you do not have access to a laser cutter ABS plastic is relatively soft and easily worked and the required level of accuracy can be achieved by printing a paper template and gluing it to the ABS plastic sheet. The sheeting may then be roughly shaped using a jig saw and accurately filed / sanded to the outline indicated by the template.

Two holes were created in the front of the cool box, one to allow a nose piece to attach to the t-ring (which presses up against the inside of the cool box wall), the second hole allowing clearance for the front edge of the flash which protrudes past the front font of the camera t-ring. A single 40mm square hole was cut into the panel to the left of the camera to allow the TEC and water block to pass through the wall of the cool box.

The edges of the ABS panels were lightly rubbed down with sand paper to improve surface adhesion and glued together using super glue gel (with the exception of the rear access panel). Once complete the seams of the box were made air tight by running a bead of silicon along the inside joints, improving strength in the process.

A clip filter acts as a sensor window and helps to form part of the seal keeping dry air around the sensor. The T-ring itself presses up against the inside of the cool box and when the nose piece or coma corrector screws in it compresses the front surface up against the t-ring forming a seal. This process is further aided by a strip of foam glued around the circumstance of the t-ring which is also compressed to form a better air tight seal. The thickness of the ABS plastic (2.5mm) has been specifically selected to maintain the correct coma corrector back focus and replaces the spacer on the MPCC MK3 M48 thread. A 1/4 inch thumb screw secures the camera to the box via the camera tripod mounting point.

To be continued:-

[al-alami]

looking good

[Adam J]

In order to create a broader surface to the rear of the box to mount the rear access panel against four strips of 1cm thick acrylic plastic where glued around the outside perimeter to the rear of the box with the aim of providing a wider surface with which to create an air tight seal against the rear access panel.

In order to achieve greater rigidity the stand alone rear access panel was fabricated from a Carbon / Glass fiber composite similar to carbon fiber but less expensive in preference to ABS plastic. As the CO2 laser is not able to cut this material the rear panel was produced using the previously described paper template method.

A paper drill template was also printed and used to align the screw holes produced in both the back panel and the orange plastic ridge to the rear of the cool box. These holes were then taped with a M4 x 0.75 threads to allow the rear panel to be secured using 8 x nylon thumb screws. A rear interface panel (containing the temperature display, USB and power connectors) was added to the access panel and set proud of the surface by 10mm to accommodate the thickness of the insulation material to be added later.

To be continued:-

 

 

[Adam J]

I chose to insulate the exterior walls of the enclosure as opposed to the interior walls, this allowed the interior to remain uncluttered and made it easier to mount components to the internal surfaces. (I later added additional insulation in select locations on the inside).

This leaves the insulation open to the environment and it was necessary to select a durable material, a light weight insulation such as expanded polystyrene would likely become damaged during use. I settled on neoprene foam rubber to insulate the exterior of the enclosure due to it’s excellent thermal properties (its used in wet suits) and ability to absorb impacts. Neoprene can be purchased in sheets of varying thickness and in this case 10mm thick material was selected as the optimal balance between bulk and insulating efficiency. I should also be noted that 10mm was selected to match the thickness of other components and spacers were fabricated for the rear thumb screws to allow them to tighten flush to the surface of the insulation.

Neoprene can be easily cut using a sharp craft knife / steel ruler and can even be shaped using 120 grit sand paper in order to achieve an aesthetically pleasing finish (a face mask should be used when sanding). Cut outs were made in the neoprene insulation for the water block, rear interface panel and telescope focusing tube / coma corrector. The neoprene was fixed to the outer surface of the enclosure using super glue gel following preparation with sand paper.  

To be continued:

[al-alami]

Great write up so far.  One question (at least for now), from the placement of the fan and heatsink inside the box it seems like you no longer have access to the miniusb connection, so how are you connecting to the laptop? or is it wireless?

[Adam J]

In order to aid in creating a air tight seal when the rear access panel is tightened down onto the main body of the cool box I cut a 2mm gasket from neoprene sheet.

One factor limiting the performance of cool box designs is the insulation provided to the CMOS sensor by the camera’s own casing. The electronics contained within the camera produce waste heat that becomes trapped by the casing; this often raises the CMOS sensor temperature by as much as 10°c above ambient levels during long imaging sessions. As such an optional method to further improve the cooling of the CMOS sensor is to remove the rear panel of the camera (including the LCD display). In this case an additional cooling fan may also be placed on the inside of the cool box access panel with the resultant air flow directed over the rear of the CMOS sensor housing. My testing has shown this design adaptation to reduce the temperature differential between the CMOS sensor and the cool box internal air temperature from 10°c to 4°c during long imaging sessions in effect providing an additional 6°c drop in sensor temperature over that achieved with the camera casing intact. While removing the rear of the camera does not strictly involve modification, this procedure is not without some risks and will invalidate the camera’s warranty. While my Canon 1000D is able to function via the USB to PC interface without using the LCD and or having the back panel attached this may not be the case for all DSLR camera makes and models.

In addition to the fan a slotted compartment is also attached to the rear access panel of the cool box (bottom right of image) and filled with silica gel beads. These beads act as a desiccant and remove moisture from the air contained within the cool box lowering humidity. This has the effect of preventing condensation as the temperature drops within the cool box and in turn prevents electrical problems and the formation of ice on the sensor surface itself. A USB B socket is mounted to the exterior of the access panel along side a four pin power connector and a temperature / humidity display (terrarium sensor). The USB B socked was sourced from a short USB B to mini USB converter cable. The cable was stripped down removing the outer cladding in order to increase the cables flexibility. In the case of a 1000D the USB socked is located to the bottom left hand side of the camera and as such the heat sink and fan have been raised slightly off the bottom of the cool box to enable clearance. For other cameras it can be useful to use a specialist low profile 90-degree mini USB cable, these are typically used to save space when attaching go-pro type cameras to quad copter type drones and can be sourced from many RC hobby shops. However, the cable is configured as a video cable and as such to make it into a normal USB cable the connection between pin 4 and ground must be severed, the mini USB is then soldered to the USB B socket in place of the original connection (I found this a necessary step when I upgraded to a 550D).

To Be Continued:

[al-alami]

Excellent addition info  

 

[tooth_dr]

Some job this. Looks heavy though?

[Adam J]

3 hours ago, tooth_dr said:

Some job this. Looks heavy though?

Its specifically engineered to be light weight, its plastic, carbon fibre and foam rubber. With the coma corrector it totals just over 1.1kg including the 0.5kg camera, that is easily within the handling capacity of my 130PDS and results in no tilt, probably would not try it on a cheap focuser though. Its when you get people building them with metal boxes and huge external heat sinks instead of the water cooling that you get problems, I have heard of examples that are more than 2kg. If you want to see a bulky heavy one look up the Orion DSLR Cooler.....and that is commercially sold.

Also its probably not so large as the pictures make it seems, there is virtually no clearance around the camera for the most compact fit possible without hacking the camera apart.

[mikeyj1]

Wow, very professional.  Attention to details like sealing seems critical - more than i had expected.  Really appreciate you taking the time to post this.  Will let you know if I give it a go!

thanks

Mike

[Gina]

Wow!  This thread takes me back to my experiments with DSLR sensor cooling many years ago.  At that time cooling below -5C didn't improve S/N ratio.  I used the cold finger approach and various arrangements of Peltier TEC cooling including water cooled.

[Adam J]

On 10/27/2017 at 13:30, Gina said:

Wow!  This thread takes me back to my experiments with DSLR sensor cooling many years ago.  At that time cooling below -5C didn't improve S/N ratio.  I used the cold finger approach and various arrangements of Peltier TEC cooling including water cooled.

Thanks, I was actually going to ask you something recently about the peltier units you used, I seem to recall you saying that the Farnell / RS ones are much better than those found on Amazon / E-bay. The one I have was from amazon and its started to lose performance (On recent runs am only getting 17 below ambient as opposed to 22-24 below), so I was going to replace it, but the Farnell ones are so much more expensive I was wondering how much better they actually are in your opinion? 

I agree that once you are below zero on a CMOS it does not improve much more, I can easily do 20 - 30min subs with that level of cooling without much noise. 

[Gina]

The dearer Peltier TECs are a lot more efficient and therefore give you a lot less heat to get rid of for a given amount of cooling.  I think it's worth it. 

There are modern CMOS sensors that are a lot more suited to cooling than the older Canon EOS 1100D DSLR that I tested.  The ZWO astro camera rang, for instance.

[geoland]

Very nice... a work of art...

You can pay a lot for TEC modules. I have a little Tellurex unit, which blows the socks off similar size ebay units and price to match. Materials and design. Having said that, I use a generic Chinese model on my cold finger mod.