Author Topic: Improoving Cooling on the 2.1 Sensor /Experimental  (Read 8009 times)

Nikon1

  • Hero Member
  • *****
  • Posts: 604
    • View Profile
Improoving Cooling on the 2.1 Sensor /Experimental
« on: November 02, 2020, 09:01:12 AM »
Has anyone yet messed around with the Cooling solution on the Sensor? 1.4 or 2.1?
 I am currently thinking about Replacing the Stock Aluminum Sensor Heatsink with Copper on my 2.1, which should allow for noticeable better Cooling performance on the Sensor itself.
 Would be interresting to see if that would help with noise.
 Allready took all the needed Measurements for the Heatsink block today, and also managed to get some 30x10 Copper Bar Stock.
 So, anyone got any ideas about this whole thing? Things to test to figure out performance differences or details to keep in mind for a project like this?
 If anyone from the Krontech team, who has access to the Blueprints and actuall Measurements could confirm if my measurements are about correct, would be also very nice.

CS223

  • Newbie
  • *
  • Posts: 37
  • 1.4 Color s/n 01266
    • View Profile
Re: Improoving Cooling on the 2.1 Sensor /Experimental
« Reply #1 on: November 05, 2020, 01:34:22 PM »
Peltier module coupled to the heat sink  8) ?

Nikon1

  • Hero Member
  • *****
  • Posts: 604
    • View Profile
Re: Improoving Cooling on the 2.1 Sensor /Experimental
« Reply #2 on: November 05, 2020, 01:41:56 PM »
Peltier module coupled to the heat sink  8) ?
Also had that idea, but there isnt really space for that, sadly.Backside of the Heatsink will cool the Main processor, front side of the Heatsink block will cool the Sensor (the 9x12,5mm Surface and the Pad), so there is 4mm of space in between there, and Heat coming from both sides, so peltier wouldnt really work. Best thing you could do is propably Heatpipes and attatch some more beefy cooling solution elsewhere, somewhere where more space is inside there. But Heatpipes are quite a bit of effort, and i never messed around with custom heatpipe Cooling untill now. Copper i can do...
 Also allready thought about Making it better and adding More surface area to have contact with the Moving air or even Water cooling, but that will add a ton of work...



Nikon1

  • Hero Member
  • *****
  • Posts: 604
    • View Profile
Re: Improoving Cooling on the 2.1 Sensor /Experimental
« Reply #3 on: November 05, 2020, 01:47:55 PM »
on this flange focal distance drawing you can see a cross-Section of the Heatsink, its just in between the Circut boards there. There isnt really that much space, sadly...
 https://www.krontech.ca/wp-content/uploads/2020/02/Flange-Focal-Distance-Chronos-2.1-HD.pdf

Nikon1

  • Hero Member
  • *****
  • Posts: 604
    • View Profile
Re: Improoving Cooling on the 2.1 Sensor /Experimental
« Reply #4 on: November 09, 2020, 09:08:37 AM »
Peltier module coupled to the heat sink  8) ?
Had to think about this a bunch more, and think, Watercooling would propably be the easiest and most effective way to go beyond the Current, basically Passive Sensor Cooling.
 Tricky bit is to get some very tiny tubes in there. Also pretty scary to have water runing around inside the Camera, but i guess no risk, no fun. Idea is to basically just drill a 2mm hole across the short side of the 4mm "base" of the Cooler and Solder on some 4x2mm Copper or Brass tube, and run water through there. with an 2mm Inside Diameter, there wont be that much volume of water moving trough anyways, unless you put a lot of pressure, so i would go with nice Cool water straight from the Tap, should give about the best cooling performance i can think of right now without a ton of effort or actually Hardcore Modding The Chronos, which i am not willing to do. This kind of cooler can still be swapped back to the Stock one.

muringuets

  • Newbie
  • *
  • Posts: 11
    • View Profile
Re: Improoving Cooling on the 2.1 Sensor /Experimental
« Reply #5 on: November 13, 2020, 11:11:43 AM »
Well I wouldn't go for water cooling, although switching the aluminum passive cooler for a copper one, with proper sanding and polishing on the contact surface and a proper thermal paste (it's an adhesive right now, need to study better options for that part) should be the best option for a passive cooler.

I'm not very familiar with the inner parts of the camera yet, but passive coolers get hot really fast, so, unless there's a good air flow inside the camera, switching for copper won't be so beneficial. Assuming a good airflow it shoud drop the sensor temperature on a few degrees celsius.

I would love to see the technical drawings for the cooler, I'd get it done.

Sorry for my English  :)

Nikon1

  • Hero Member
  • *****
  • Posts: 604
    • View Profile
Re: Improoving Cooling on the 2.1 Sensor /Experimental
« Reply #6 on: November 13, 2020, 12:15:10 PM »
Well I wouldn't go for water cooling, although switching the aluminum passive cooler for a copper one, with proper sanding and polishing on the contact surface and a proper thermal paste (it's an adhesive right now, need to study better options for that part) should be the best option for a passive cooler.

I'm not very familiar with the inner parts of the camera yet, but passive coolers get hot really fast, so, unless there's a good air flow inside the camera, switching for copper won't be so beneficial. Assuming a good airflow it shoud drop the sensor temperature on a few degrees celsius.

I would love to see the technical drawings for the cooler, I'd get it done.

Sorry for my English  :)
I posted my measurements above, those should be reasonably accurate.
 It also isnt really Thermal Adhesive, its a pretty thick thermal pad.
 .
 As i posted in an other Thread yesterday, i just replaced the Stock fan with an Noctua one, and am not yet done with Temperature testing, but from what i can tell from now, the 11% less airflow on paper for the Noctua will result in about 4 to 5C increase in Temperature with partially blocked intakes and a little less difference with fully open intake holes. So the Static pressure of the Stock fan makes a bigger difference if something is in the Way of the Airflow. I am also planing on putting an 40x40x28 (Will need to be mounted outside the Camera Body) High RPM, High Static Pressure Server fan on there to test if MORE Airflow does actually help a lot with cooling. The one i found laying around has DOUBLE the Airflow compared to the Stock Fan of the Chronos, which should show some noticeable difference, if airflow had an huge influence on Sensor Temperature. The Noctua Fan Mod i did there is mainly because of the Noise. Now with the Noctua its so silent, i actually forgot that i had it just sitting there on my desk running while temp testing, which absolutely wont happen with the Stock fan...
 .
 The Thing about the Watercooling is, i can easily force the sensor to below room Temperature, at a constant temperature. I will not go down below dew point, because for obvious reasons i dont want to deal with water condensating on the Outside of the Cooler, but just above that should be fine. Aiming for something like 15C Heatsink Temp, which should work out to 20C or well Below for the Sensor itself. Right now, sensor Temperature is sitting somewhere between 45 to 52C for all the Testing i did at 21,5 to 25,5C Room Temperature with the different Fans and sometimes partially blocket ventilation holes. Purpose of reaching such a low Sensor Temperature is to find out if Temperature will have a significant impact on Image Quality, more specifically Noise Levels and Dynamic range. Keep in mind, i could also run hot water through there, Turning it into a Water-Heating-Block. So i can easily test a huge Temperature Range with very stable temperatures.
 .
 Actually using some kind of water-Cooling Setup in the Camera for Daily use isnt very practical and i also wouldnt reccomend that.
 But the Copper Block is propably what i will end up using. I assume that just replacing the existing Aluminum Sensor Heatsink with Coper Material in the Same dimensions will proppably make back the few C i lost with the Noctua, which would allready be huge, so same Temps as a Stock Chronos but Way more silent, without any crazy solution. Replacing a Fan and using a different Material for the Sensor heatsink isnt that crazy to do, and i can even realistically See Krontech maybe even doing this at some point in the Future for their cameras, if it turns out to be working well, isnt that mutch more Work/ effort to do.
 .
 As of now the Cooling of the Sensor itself is technically actually partialy passive cooling (The Camera body itself takes away a lot of the Heat from the Sensor itself, which is of course somewhat cooled by the Fan. I think the Fan is mainly there to also make sure nothing else in there overheats.). This little Heatsink thing shown there is more or less just a "bridge" for the Heat to then go into the Metall body of the Camera. Of course A lot of airflow will lead to better cooling, since there is still a ton of surface on the Camera body itself and also a little bit on the Heatsink itself, and also on the Sensor and Circut boards. But there isnt really fins or something to really increasy the Surface Area to get the Heat away there more efficciently. Speaking about cameras overall, this is pretty much allready Way better Sensor cooling Than About any "usual" Video Camera or DSLR/DSLM has (Metall body with an Actual heatsink connected to the Sensor AND an fan!), but that Sensor in the Chronos is also doing way more serious work than most of those Sensors, and because of that also puts out a lot more Heat.
 .
 I might even go as far as to Modify the current original design by Krontech for the Sensor Heatsink a bit and add fins (There Seems to be quite a bit of space for Fins or someting) or even Heatpipes (actually found some Heatpipes laying around in my pile of old Server cooling and got to play with them a bit, seems doable, but certainly tricky to do in such a small space, but would help a lot to get away the Heat FAST there) for my own Camera, but for sure no Watercooling of any sort. Not even I am that crazy to go with a Watercooled Chronos for Daily use....
 .
 About Lapping the Heatsink, sure i could get that done or do it myself, but i think i would start at replacing the Fat Thermal Pad between there first, since with that in between there, this seems like a waste of time. Or at least replace it with a thinner one, if you want the best possible cooling solution. But that would in turn mean that you would need to Modify the Dimensions, which i dont want to do for the Contacting surfaces. An other Problem with getting good contact there is, that you dont have any kind of Spring Loaded Mounting there like you usualy have on an CPU/GPU-Style Cooler. This Little Heatsink Thing is Straight up bolted to the Camera body itself. So if you get anything just a little bit off or missaligned, bad things will happen. So the Thermal Pad Kind of needs to be there to allow for Some Imperfections and Tollerances and also Mounting. If you wanted to go with an direct Metal-to-Sensor Contact Style Cooling, you will need to come up with an completely different cooling concept, which i dont really need or want to do, unless it turns out, Temperature Matters a lot for noise, then this idea is maybe worth to think about more.
 .
 .
 #EDIT:Grammar
« Last Edit: November 13, 2020, 12:19:35 PM by Nikon1 »

muringuets

  • Newbie
  • *
  • Posts: 11
    • View Profile
Re: Improoving Cooling on the 2.1 Sensor /Experimental
« Reply #7 on: November 13, 2020, 01:04:08 PM »
You're pretty much on top of everything, I've read your post about the Noctua fan, nice to see the results.

Lapping the cooler won't do any help because of the thermal pad, correct. I probably didn't make myself clear in my previous comment, but I meant that it would be preferable to pick a thermal pad with a better conductivity (Gelid has some fancy ones last time I looked)

I like the idea of carving fins.

Also, from what I read, a lot of the problems in image quality derive from heating, so you need to keep doing black calibration every now and then...

Nikon1

  • Hero Member
  • *****
  • Posts: 604
    • View Profile
Re: Improoving Cooling on the 2.1 Sensor /Experimental
« Reply #8 on: November 13, 2020, 01:45:18 PM »
You're pretty much on top of everything, I've read your post about the Noctua fan, nice to see the results.

Lapping the cooler won't do any help because of the thermal pad, correct. I probably didn't make myself clear in my previous comment, but I meant that it would be preferable to pick a thermal pad with a better conductivity (Gelid has some fancy ones last time I looked)

I like the idea of carving fins.

Also, from what I read, a lot of the problems in image quality derive from heating, so you need to keep doing black calibration every now and then...
Yes, a better Thermal pad would also be very good improovement. Need to do some Research though, cause i am not really that much into Thermal pads tbh. I know a bit about thermal paste, but no expert in that field, but Thermal pads i know very little about. Never actually bought any so far, but worked with them, if they where allready in whatever Devices. Also i dont know what is even in there right now. I expect some rather generic thermal pad with an reasonable price-Performance value, which still performs pretty good, but for sure nothing overly fancy. I am sure, there are better ones, but would have to find out about what that would be then.
 .
 Carving finns would need to increase the Size of the Heatsink by quite a bit to even have anything to Carve into. There is room, so i think i will end up doing it. However, i will start to do an 1:1 reproduction of the Original one to get an propper comparison Value. Going Straight with a Heatsink with fins would be somewhat unfair. From the 1:1 Copper one, i will see how much room there is to extend Fins out of the Block itself.
 .
 There are a bunch of problems with Image Quality. First of all, there is everything that isnt really Temperature, that would be Signal Processing and optimisation of Readout. Those Lines that can be very visible are mostly that i assume. Temperature wont affect those things not that mutch, if at all. They are caused by different things, and as i heard, should be improoved by an next little bit with the upcoming Firmware Update.
 .
 Then There is Random Sensor noise. That one is actually Highly affected by Temperature.
 And then there is also The Combination of the Two, when the Camera tries to remove all the Weird stuff the Sensor does for different Reasons. Main Problem Here is, that Temperature usually changes a lot, especially if the Camera is just Turned on. As i Said, quite a bit of the Cooling is Passive also, and Since the Body is a huge Billet of Aluminum, this thermal Mass first needs to get up to temperature bevore it stabilizes into an Equilibrium. From all the Testing i did, after like 1 to 1,5 Hours the Camera reaches a Very stable temperature, which only deviates by 1,5C or less, if Room Temperature is constant. Up until a Stable State is Reached, Temperature allways Changes noticeable, even in the Short time from Doing a Black Calibration to actually starting a Recording. If you do an Black Calibration once its on a Stable Temperature, The Calibration usually doesnt run away from the actual sensor Noise that much or will even Stay where it should be. Still would propably not hurt to do a Black Calibration every now and Then when Shooting for Hours, even when warmed up.
 .
While now the Lines and Weird Stuff is propably mostly Software, and the Exposure/ Sensitivity and Random Noise is mostly Temperature CHANGE, there is yet an other problem, Absolute Noise Levels. When the Camera needs to adjust for More Noise or just basic different exposure for different pixels/ Rows (which gets worse the hotter the Sensor runs), it needs to bend the Values from the Sensor More, then if it had not to do that by that much. As a result of the Heavy adjustments the Camera needs to do for an Hotter image, even if it gets rid of most of the weird looking stuff, it will still mean a lower Dynamic Range as a result (At least in theory, this is kinda the Point of this whole project to find out, How much difference it really makes in the Real World, cause in Theory Colder (A very Constant "cool" Temperature, big problem with "good" cooling is, that the Camera will get very "quick" and responsive to enviromental Temperature Changes, as Heat is exchanged a lot faster, and as we learned earlier, even a few C can Cause a big difference in Noise and Exposure, resulting in very visible Noise or other Image Artifacts. Because of this, the Huge Thermal Mass of the Chronos Camera Body is actually pretty good for Image Quality, once it is heated up = Has Reached A Stable Temp.) Overall Temperature Level should have Less Noise, and because of that also Higher Dynamic Range.

Nikon1

  • Hero Member
  • *****
  • Posts: 604
    • View Profile
Re: Improoving Cooling on the 2.1 Sensor /Experimental
« Reply #9 on: November 15, 2020, 08:26:45 AM »
Today i finally got around to start Milling out an Copper Heatsink, but sadly i messed up with one Measurement on the Contact pad for the Sensor when rough-Milling, and now its a full Millimeter to small, so i have to start over again another Day. Surface finish i am Getting on the Finished Surfaces themself, is so good that i would not even Bother to do any grinding or Polishing on them For now. But first i need to make one in the first place, without reading the dimensions from the wrong pad...

Rainer

  • Newbie
  • *
  • Posts: 8
    • View Profile
Re: Improoving Cooling on the 2.1 Sensor /Experimental
« Reply #10 on: November 15, 2020, 12:13:31 PM »
I am really impressed by this project and very curious how the image noise will be affected. Great work!

muringuets

  • Newbie
  • *
  • Posts: 11
    • View Profile
Re: Improoving Cooling on the 2.1 Sensor /Experimental
« Reply #11 on: November 16, 2020, 05:18:49 AM »
Today i finally got around to start Milling out an Copper Heatsink, but sadly i messed up with one Measurement on the Contact pad for the Sensor when rough-Milling, and now its a full Millimeter to small, so i have to start over again another Day. Surface finish i am Getting on the Finished Surfaces themself, is so good that i would not even Bother to do any grinding or Polishing on them For now. But first i need to make one in the first place, without reading the dimensions from the wrong pad...

Nice job! Keep us posted, curious for the temperature tests!

Nikon1

  • Hero Member
  • *****
  • Posts: 604
    • View Profile
Re: Improoving Cooling on the 2.1 Sensor /Experimental
« Reply #12 on: November 21, 2020, 01:20:18 PM »
Quick Update:
 Finally got around to making another one, which is now to size for the Critical Dimensions. Kind of messed up again, the 2.0mm Drill i used to predrill the Holes broke on the Second Hole, so i had to mill it out with an Carbide endmill, and the Hole ended up about 0.4mm oversized in Diameter. But since this is just a Clearance Hole for an M3 Screw, its no problem.
 .
 I also got myself a bunch of pretty fancy Carbide endmills and was able to improove the Surface Finish on the Contact surfaces by quite a bit. its now about as Close to an Mirror Finish as i will be able to get on my old Manual Mill.

Nikon1

  • Hero Member
  • *****
  • Posts: 604
    • View Profile
Re: Improoving Cooling on the 2.1 Sensor /Experimental
« Reply #13 on: December 30, 2020, 06:13:02 AM »
Here is an other Quick update, made some tiny custom Connectors for the 2mm OD Brass Tube that is planned to be used for the Watercooling Tests, to fit inside the Little Space between the Circut Boards.
 .
 Still quite a bit of Way to go untill it is finished, Still need to mill some Channel into the Copper cooling block itself to connect it to the Tube and connect all of that Stuff together (also probably still need at least one more Of the Bigger Adapters made and find a way to attach a Temperature probe to the Incoming Cooling Water, and idealy also make some Clamp to Mount all the Stuff on the Outside of the Camera, so the tiny tubes wont break from the Weight of those big connectors and the 10mm ID Hose that will go onto it).
 Initially i tried to get away with just using straight up 2mm Brass tube without any soldering, bent from a single solid piece, inside the Camera and only have some Kind of Connectors on the Outside to be absolutely save from Any potential Leaks inside the Camera. But that sadly didnt work out, even after quite a bit of trial and error. Just couldnt produce clean Bends as tight as i needed to reliably, so i chose to go with (some admittedly pretty overengineered, but i really dont want to risk any Water leaks inside there...) 90 Connectors, so i will only have to use Straight bits of tube without any bending (hopefully, or at least avoiding the Extreme Bends, if i still need to do any bending at all). Those will be Soldered when everything is ready for Assembly.

Nikon1

  • Hero Member
  • *****
  • Posts: 604
    • View Profile
Re: Improoving Cooling on the 2.1 Sensor /Experimental
« Reply #14 on: December 31, 2020, 12:44:01 PM »
So, i made an Small Waterblock today to fit an TO220 Size Resistor. for once as a Proof of Concept for the Final Test Cooler for the Chronos to see how well this kind of Thermal design will handle the Expected Thermal Load, and also to see how difficult it will be to drill such a small and Deep Hole. Allready very happy, i did an Testpiece bevore doing this to the Actuall cooler itself, because i broke at least 6 or Something drills trying, which all got stuck way inside the Part and are impossible to get out now. Turns out, drilling a 30mm Deep and only 2mm Diameter Hole is seriously hard to do.
 .
 Resistor is one of those:
 https://asset.conrad.com/media10/add/160267/c1/-/en/001624828DS01/datenblatt-1624828-tru-components-tcp50u-a3r90ftb-hochlast-widerstand-39-radial-bedrahtet-to-220-50-w-1-1-st.pdf
 .
 So, an 50W rated Resistor In a TO220 Package, i ordered an 3,9 Ohm one, so it will output about 36,9W of Heat when i feed it with 12V Power (Given it is rated at 1W in Free Air without any cooler attached, i guess it will dissapate up to 1W on its own Surface Area, so the Heat it pushes into the Waterblock would propably be closer to 36W), Which is close enough to the Maximum Heat output i expect from the Sensor itself and the other Big Chip that Cooling Block is also Cooling on its Backside (according to The 2.1 Datasheet, Power input is given as 40W, and considering that also includes the Fan, the display and all the Other Electronic parts inside the camera (maybe also includes some amount of power to charge the Battery when empty), 36W is really worst-Case Scenario for Sensor+The other chip, i assume it should be quite a bit less than that actually, but i dont really have any other data, so just to be sure, test it with 36W input, if that kind of whaterblock design can handle that much heat, it will be able to cool that camera).
 .
 Will still need to wait a bunch of days for the Resistor to arrive, so i just took the next best thing i found laying around in an TO220 Case to get the size right, which happened to be some random used Voltage Regulator, will be replaced by the Resistor whenever it gets here.