This is just a "what if" idea, I am coming back to a concept I did some research on last year, thinking about trying to build a liquidcooling system for one of my 450Ds. The idea is to get the heatsink/fan weight off the camera and have the radiator heatsink/fan mounted on the pier. I would put a 40x40mm (TEC sized) waterblock on the camera cold finger hot side and just run two flexible plastic tubes away to the pier mounted pump and radiator/fan unit. I have been looking for efficient 40x40 water blocks, and most seem bigger than that but the standard AMD ones could work, and there are many simple (and cheap) Chinese ones made out of copper and aluminium. I would like to keep the weight down so am leaning towards one of the ones with a copper bottom (TEC side) and aluminium top section (copper is better at absorbing heat from a source), obviously sealing and gaskets become an issue. Not sure how much pressure these things run.
Anyone have any PC experience with these cooling systems? I would be concerned about hose flexibility as they have to move with the mount, and how long they can be for any given pump used, etc.
I know from my research that QSI offer a recirclating pump and a water block that attaches to some of their cameras, suppose to offer an additional -10C off the delta T of the normal camera TEC cooling. I am not looking for deeper cooling just to move the load off the camera, so if QSI has made it work on a CCD camera I am thinking it is do-able.
I've given this some thought with using a copper plate from the one of the CPU heatsinks with embedded tubing. Aquarium Pump, automotive windshield wash tubing, ice water bucket ..... Not sure how efficient it would be, parasitic heating, achievable delta but water is a better heat transfer medium than air and you could make quite a stable system. One advantage might be size of install at the camera not having a big heat dissipater but lots of other bits hanging off would be a pain.
One advantage could be in the field, a 12volt aquarium pump uses a LOT less power than a TEC.
I hadn't got to the point of a TEC at the output end. Needs more thought as to what you would gain with a more complex setup.
Brent I have spent some hours trying to find a waterblock that would fit the existing cold finger setup. The idea is the water block would cool the hot side of the TEC, replacing the current heatsink/fan combo, and the pump and radiator/fan would be off the scope. Rowland did some heat calculations for me. If I go through with it I would be using CPU water cooling components and not aquarium stuff. There are plenty of water blocks on the market but finding one that is not overkill on a 40x40mm TEC is not easy, and it has to be light. I also need to find one with side hose barb entry so the hoses are not pointing outward. The water block maybe the hardest thing to sort out. It would be a closed system, using soft hoses that won't restrict mount movement, guiding, etc. should be no more an issue to manage than dew heater cables - hopefuly.
Glen, did you look at GPU coolers? I can't see the dimensions but something like this: https://www.pccasegear.com/products/...pu-water-block
If it is too bit, it probably does not matter? If it is too small you could use a copper sheet as a heat spreader?
Also if you are mixing copper and aluminium, make sure you use corrosion inhibitor add-in fluid.
Just thinking of a possible issue is water running through the pipe and then dispersed through restriction in the cooling element may cause unnecessary vibration. It is not a requirement to worry too much for CPU's but for obvious reason is critical if the camera vibrates with the water flow.
I myself though of going through this process but will not consider direct cooling instead enclosed cooling method.
Trying to remember more from the days when I had a water-cooled computer...
You need large hose diameter for better water flow (or a bigger pump). The hoses that are generally used for computer cooling are quite stiff and would not work with a moving telescope on the mount. They are stiffer than a typical cheap garden hose.
Movement may also lead to leaks on the joints... Hm, what are they using for car brake lines?
Well I have gone ahead and ordered some bits for this project. I went with one from PCcasegear, the GPU one mentioned below. I found a spec sheet for it and the base is 50x50 so that is fine for the TEC, a little big but not an issue. I will need to make a small carbon fibre bracket to hold it fast to the cold finger, sandwiching the TEC. It has the tube barbs on the end so that is good. I ordered some flexible tubing. Still looking at pump and reseviour options, the pump will need adequate lift pressure to get to the camera. I also need to work out how I will mount the pump and the radiator in the observatory. Of course it's going to be hard to move the camera into the house now as it will be tethered to the pump etc. by the tubes.
It's a block that bolts onto the back of the heatsink on the camera. Takes heat away from the heatsink. You don't remove the heatsink/TEC from the camera.
I use 1/4inch tubing and a low flow rate. I'm using a fairly flexible tube and run it off the scope via the counterweight shaft. No vibration issues. You don't need a powerful pump - it syphons around.
I use a 20litre drum of water - it will slowly warm up during the evening. Don't try cooling the water with ice. If you lower the temp of the heatsink below the dew point, you'll have a lot of water condensing on the camera and potentially fry the electronics.
I don't use mine often. Bit of a pain to setup and I'm happy with -20 during winter. May use it this month seeing it's still a bit warm.
Thanks for that info David. Where did you buy the QSI pump? I will probably try out your method of running the cooling lines on the counterweight bar.
I have ordered the rest of the equipment I believe I need to get the cooling system operational (Pump, connectors, radiator, mounting brackets, etc).
I am thinking of putting a couple of quick disconnect fittings on the lines so that I can remove the camera from the rest of the system. I am looking at wall or floor box mounting in the observatory for the pump/radiator.
It is pretty clear to me that this is a more expensive heat shedding option than the heatsink/fan combo that is traditionally used. I have already spent three times the amount on getting these components together. The out of the box integrated water coolers, like the Corsairs, might be great for the inside of a PC case but they are unsuitable for cooling a camera.
Longer tube runs require more pump head pressure, and a special water block is required so keep the weight down.
Testing the differences in performance will be interesting.
I just bought a cheap 12V pump off ebay - less than $20. I bought an adjustable voltage controller so that I can slow it down. There are quick release connectors in the block that came from QSI so I leave the block attached to the back of the camera and only attach the tubing when required.
You don't need a powerful pump. The water siphons back down, so i think you're only overcoming the flow resistance in the tubing.
All of the parts for the liquidcooling system arrived today. Laid out on the counter it looks pretty complicated but at least nothing is missing. The coolant concentrate bottle is in the back right of the photo, it contains the ethyl-gylcol stuff and corrosion inhibitors. The hose (3m) is very flexible but does not collapse in on itself when bent. I already have a fan to stick on the radiator, and enough distilled water to fill the system. The cooling block that goes on the camera cold finger is nice and compact and will remove weight from the camera. I will try and get it all together this weekend.
It took awhile to assemble everything, but I managed to get a bench test done today. Started by flushing the radiator with vinegar, because I found some advice online that these small radiators for liquid coolers sometime had manufacturing burrs and rosin inside, and the vinegar is suppose to desolve the rosin, well it must have worked because when I tipped it out it was blue grey. I brought the camera temperature controller in from the observatory and setup all the power connections for the pump (wires everywhere), stuck a small blower fan on aimed at the pump heatsink, and the installed the radiator fan (just reused the old heatsink fan for now). Filled the system with a concentrated anti-scale, biocide, etc fluid mixed with demineralised water.
Test #1. After filling the system and checking for leaks (there were none). I checked the hose vibration at the camera and there was nothing. I started the first test. The ambient room temperature was 25.4C when I started. Turning on the camera controller the temperature started at 25.4C and rapidly (within one minute) dived to single digits. The Set Point was at 1C so I reset it to -10C to see how low it would go. Within a few minutes it dropped below 0C and after about five minutes it got down to -8.0C, which means a Delta T of -33.4C! This is the best result I have achieved with this Jaycar TEC (52 W, 6amp). The previous low temp benchmark on this mono camera was a Delta T of -27C, with a insultated cold finger and it took much longer to achieve that.
I should add that there was no insulation on the cold finger, nor argon bag for this test, and you can see the frost layer in the photos below. So I believe I maybe able to get it to a Delta T of -35C with the cold finger insulated as I normally have it for imaging. The system I have is designed to shed more than 200W of heat, and going by Rowland's 2X rule it should be capable of cooling more powerful TECs, or two stage TEC, if I wanted to go there.
I have attached some photos taken during the test. Fairly pleased with the result. I need to install the larger 120mm fan for the radiator (that's the size its designed for), and install the pump and radiator in the stand I am making for it in the observatory. There is plenty of hose length for my rear camera mount RC but I will need to check on the hose route for the newt. Adding more hose will not affect performance as the head pressure of the pump is over 5m. You will notice in the photos of the camera that it has a much smaller out-hanging cooling arm, and the weight seems a little less than with the heatpipe sink and fan.
Must admit it makes for a very small and tidy setup at the camera and the performance is quite astonishing. I have gotten a delta of 20*C which is very useful and I think I can get more now with a sealed front to the camera.
I just query the amount of extra hardware and complexity and whether it is worth the gain. Are you going to measure the noise value to see definable results ?
Glen, I can't help but think you'd be better off with a better behaved sensor to start with, otherwise it just makes it look like you like tinkering
Yup, tinkering ....
All part of the DIY fun
Looking for my next 'project' now. Lathe and workshop is idle.
Pier's done, 450D Full spectrum is done, got alignment and guiding all sussed.
ASCOM direct control to the mount is next, might run that and guidng from a second PC.
Then an ASI174mc cooled probably with it's attendant learn the new processing requirements.
Then the 8F8 Astrograph back on the mount and .....
Glen, I can't help but think you'd be better off with a better behaved sensor to start with, otherwise it just makes it look like you like tinkering
Tinkering! You maybe right, but there is nothing wrong with my sensor's behaviour. Sure I guess I really didn't need to do this, the air cooled version easily reached 0C.
I did manage to do some brief testing last night before the high cloud rolled back in. I can confirm it does work well, if anything too well, at least as far as the cold is concerned. I think I might have fogged the front UV/IR Cut filter by cold transfer to the front of the T Adaptor where it was mounted, driving it down to -5C on a humid night, but importantly the argon bag sealing around the water block worked really well and there was no condensation inside the bag. I shot five darks at -5C and compared them with 0C versions and there is not really any detectable difference (as we already knew).
Ergonomics and operations are an issue, as the hoses create a managment challenge during mount movement. I had been worried about the weight of the water inside the hoses creating drag on the camera during guiding, and despite focusing carefully before I did any test shots they turned out to be slightly fuzzy. There is not really any pump 'pulsing' finding it's way to the camera (the radiator is in between to smooth this out). On the other hand I may not have locked the focuser after I focused.
As a proof of concept I think I have shown that a DSLR can be liquid cooled, but there are issues that need solutions.
I am probably going to park this project for now, and may scrap it altogether and go back to the air based cooling simply to reduce complexity. I can always use the components to build a 12V fridge for the obs.
After reviewing the imaging test it turns out that I failed to lock the Moonlight focuser during the test, thus it shifted out of focus under the weight of the rig. After reassessing, I have decided to go to a lighter water block ( the Raystorm GPU one was heavy copper plate), and purchased a cheap aluminium one from ebay. http://www.dx.com/p/aluminum-water-b...0#.VwOYI1Jrm-c
I will stick a 40x40mm thin copper plate between the TEC and the aluminium block to improve the heat transfer off the TEC.
The aluminium block takes a smaller ID hose as well. After discussing fluid dynamics with my son (the engineer), I have decided to keep the present pump/radiator system and use a 'fluid bridge' and reducer to feed the new block. I will go to thin wall aquarium silicon tubing for the feeds to and from the water block, with the larger hoses retained to the bridge, radiator and pump. Pressure will increase in the line feeding the water block but volume is decreased. Just sourcing the reducer fittings and bridge now. The bridge will be mounted on the pier and the new thin lines will run from there to the scope. More testing to come, stay tuned.
Hahahahahah, can't leave it alone can you ?
Aluminium is the next best conductor to copper so I'd be for not including the copper ( less surfaces ) and just using a good thermal paste.
don't take my comment as criticism, I think it's great what you guys have been doing...inspirational!
