QHY5L-II Mono Options for Passive / Active Cooling

[SnakeyJ]

Okay I've had my QHY5L-II mono for just over a month and have been itching to take a look under the hood to consider options for cooling this super little camera.

The camera uses the Aptina MT9M034 1.2Mpix (1280x960 active pixel) 1/3" sensor with Peak QE at 74% (datasheet available at http://www.aptina.com/products/image_sensors/mt9m034/). This is the same sensor used in the ZWO ASI120MM. The current drivers/firmware support shutter speeds from 0.020s to 600s, 14bit ADC giving 8bit or 12 bit colour depth and with USB2 interface and ROI support can produces some very respectable frame rates even on Saturn ( 400x400 ROI @ 8bit giving 75FPS for L @ 85% gain, 65FPS for R @95% gain - dropping to around 45-50FPS for B @ 99% gain.

Noise levels seem very good, even at max gain for shorter exposures, way better than my previous experience with the Pt Grey Firefly Mono - but there are a few (3-4) hotpixels showing up on high gain short exposures and the noise levels build up with exposure length. The camera body is noticeably warm to the touch during extended video capture and this will only worsen as ambient temps increase towards summer.

Interestingly, QHY list a 'Super Radiator' accessory for this on the product data sheet. Though I have been in contact with QHY, who say this is not available currently and were unable to advise on manufacture/availability - though did clarify that it would be a passive sleeve/fin design to slip over the 1.25" body of the camera.

My curiosity suitably piqued and finding no information on the internals or cooling mods online - I decided to risk voiding the warranty for a little (cautious) investigation:

The good news is that the camera is very nicely made and is already designed to draw the heat of the cmos and main processor via an internal aluminium bulkhead to the outer case. So it should be very simple to make an external passive heatsink and fan. For active (TEC) cooling it would be very simple to add an external 20-30W peltier fitted to a copper collar and it should also be possible to fit a smaller device 5W? directly to the back of the bulkhead. The case itself would lend itself to fitting a small dessicant tube and the modular construction and long ribbon cable would also make it very suitable for rehousing in a custom case if desired.

Given the experience with cooling the firefly, I think this offers considerably more options for a neater and more efficient cooling solution and I would hope to easily achieve 25-30 deg below ambient with 80-85% reduction in noise at max gain. Given the current price of £ 229 (roughly 8x what I paid for the firefly), there is certainly more financial risk - but the mods look far simpler and will avoid any direct contact with the CMOS, so I feel well up for the attempt!

[tonedeaf]

thanks for the nice pictures...

...had mine also disassambled except the front sensor pcb (the two screws were glued)...

one option as you said would be a peltier on the back of the cmos, the heat goes to the casing where at least cooling ribs would be fine

but the peltier would be a small one...so i don´t know if it would be that noticeable.

peltiers on the casing...hmm i don´t think that would be the best idea..or it would be hart to make cause its round

and then there would be dew inside...

i think the peltier on cmos would be best...

a small hole through the casing (for the peltier power line)

for the cooling of the casing maybe a cpu computer aluminium cooler drilled with 1,25" hole and stuck on the camera body...

hmmm

[SnakeyJ]

thanks for the nice pictures...

...had mine also disassambled except the front sensor pcb (the two screws were glued)...

one option as you said would be a peltier on the back of the cmos, the heat goes to the casing where at least cooling ribs would be fine

but the peltier would be a small one...so i don´t know if it would be that noticeable.

peltiers on the casing...hmm i don´t think that would be the best idea..or it would be hart to make cause its round

and then there would be dew inside...

i think the peltier on cmos would be best...

a small hole through the casing (for the peltier power line)

for the cooling of the casing maybe a cpu computer aluminium cooler drilled with 1,25" hole and stuck on the camera body...

hmmm

My front PCB screws were also glued in, but with clear silicon or similar soft clear glue - a sharp cross point was sufficient to unscrew and then I spent a little time removing the excess glue from the screw heads and boards (don't want any stray bits getting stuck to the cmos or window).

Small peltier directly attached would be my favoured approach - but to be efficient this needs to be closely matched to the heat output from the cmos. I also think that the thermal junction TIm between the cmos and stud could be considerably improved to allow the heat to dissipate more quickly away from the sensor. Agreed that an external one would not be as efficient/elegant, but would be an improvement on a purely passive approach or heatsink/fan combination and gives an option to anyone not happy with opening their cameras/voiding the warranty.

Good idea re the CPU heatsink for passive cooling - I know that some of these have a central cylindrical copper plug with aluminium fins/body around it. Unlikely that the copper plug would be a neat 1.25" diameter, though I think its probably 25mm+ from memory - provided the aluminum body had sufficient material it may be possible to bore this out to size and get a neat fit. My original thought was to source an inch of 1.25" ID aluminum tube and then wrap and solder some fins around this, which is somewhat less elegant.

[tonedeaf]

ok just made an attempt:

i´ve drilled a hole with my "home" tools, not perfect...but hey it should do it.

further i´ll need heat paste, and a half angle usb to install a fan on the back side....

now waiting for clear skies....maybe in 2-3 weeks

[tonedeaf]

btw...what do you think....

how big is the "hole" on the back side of the chip?

8x8mm? or 15x15mm?

have to make a shoping list for tomorrow

[SnakeyJ]

Nice job on the heatsink - where did you find a 32mm drill bit so quickly or did you use a hole cutter? I had a hunt around at work this morning, but had no hss drills over 18mm (though lots of hole cutters, diamond and masonary drills)....

Worth grabbing some dark frames for a before and after comparison to see if the cooling is having the desired effect.

I didn't measure whilst I had the oppurtunity last night, but the raised stud would be 8-10mm diameter - I also didn't take the thermal pad off of the underside of the CMOS - though it should be possible to get the contact dimensions from the product datasheets from Aptina. The die size is 11.43mm x 11.43mm, so I would think 8mm square would be the maximum dimensions for direct contact to avoid shorting out any pins. Would have to be a non conductive TIM as well, though nothing permanent like the Arctic Epoxy.

[tonedeaf]

hey thanks, so 8x8mm peltier onto the list...

regarding the drilling

i know...thats for wood, but wd40 as lubricant and not to heavy pushing, also some minutes of drilling time got the aluminium drilled

[Cath]

Theirs not much to the camera is there Jake, looks nice and simple.

I couldn't resist putting a peltier on the back of that sensor lol, perfect opportunity with the way they have it laid out.

[tonedeaf]

got it measuered 7x7mm

now looking for the right peltier element

[SnakeyJ]

That hole cutter looks a little fearsome, but certainly did the trick

Hmm 7mm x 7mm (nice image BTW) - and it appears to be recessed/flush with the front side of the pcb - might need a little square of 1.2mm copper as a spacer - but then that introduces another thermal joint and we certainly don't want a thick pad here as they not terribly efficient. On the other hand at least the back of the chip is accessible which is a big advantage over my firefly - I'm wondering now if the bulkhead could be removed/modified or replaced with a copper disk and slightly longer square section of copper which could be soldered together..... more to consider.... At least the hole does not expose any of the contacts/pins on the back/of the sensor.

I couldn't resist putting a peltier on the back of that sensor lol, perfect opportunity with the way they have it laid out.

Looks a whole lot simpler than the firefly - though its a bigger investment/risk should anything go wrong with the mod - but the QE, resolution and long exposure capability on this camera make it well worth the effort. I think I'll knock up a PWM Peltier controller with an Arduino Nano, Mosfets and three digital temp probes and may add a couple of heater/resistors to keep the window dew free. The sensors can monitor hot and cool sides plus ambient temp, to give some more accurate set point control.

BTW - I finally got the water block yesterday, for the revised firefly cooling experiments - so will be knocking up the new heatsinks this weekend if the kids allow me the brain space and the weather keeps me in

[SnakeyJ]

Ok as a point of reference here's a set of dark frames from the unmodified camera taken indoors (ambient 20C). I shot these in firecapture 2.2 using 1 sec exposure, full resolution with Gain at 250, 500, 750, 900 and 1000. It should be noted that this gain is not a linear control with a definite boost cutting in between 99% and 100%. I shall e-mail Torsten and see if he can quantify how this slider should work and if there is a direct dB equivalence. Each test was run successively with no time for the sensor to cool and 5 frames taken of which I am taking a 300x300 crop from the centre of the final frame for the samples posted below. Recorded settings are from FC logfiles with some addition (Exposure/Colour depth as these were incorrectly recorded/not present). Test was run once at 8 bit colour depth and a second run at 16 bit depth (according to FC).

8bit Darkframes:

16bit? Darkframes:

I don't think the precise settings are too terribly important, provided the tests are repeatable for testing efficacy of modifications. The camera had a black lens cap fitted and was shut inside a laptop case to eliminate any possible stray light....

Whilst the sample crops show significant hot pixels above 500 Gain, and vertical banding beyond 900 Gain - The Noise (avg deviation) is the figure to follow and I shall use this to plot a curve for subsequent comparisons once I have a better understanding of the true gain figures in dB.

Currently not sure why there is such a huge difference in noise levels for the 16 bit capture - a 20x increase was somewhat unexpected and an increase of this magnitude is not apparent in the sample crops - so this may be a firecapture calculation issue or issue with the drivers.

[tonedeaf]

i´m now going a different approach...

down with the cooler...(it hasn´t that much contact points)

also the peltier inside...hmm

so made with the help of an fellow collegue

we made an adapter (*caugh* at work -> long live lazy friday )

it´s ab bit big 40x40mm (so it will sit very deep on the camera)

40x40 peltier element <- which is on stock in my electro shop (powered by 12V 1,5A power supply -> ~18-20W power)

on top of that a cpu active cooler somewhere there an on/off switch...

as a guiding cam it will be a bit overweight on my 9x50 finder

photos of the whole assambly wil be made today at home

but now some peaks on the adapter:

first milling done

second round

ready made (with the old 1 minute technical drawing <- the real measurements are a bit off the drawing)

[tonedeaf]

oops cant edit

as i see right now, the side view is incorrect...should be "mirrored"

(as this fault happend, you can see technical education is going down the drain in austria!

five years learning were nearly worthless )

[SnakeyJ]

Nothing wrong with your technical drawing skills, mine were tested 30 years ago to CSE Grade 2 - brought back some fond memories of drawing boards, pencils, pens and throwing the rubbers around when the teacher wasn't looking!

If I'm reading this correctly this will give you a bolt on collar/block of aluminium, cooled externally by something like a TEC1-12706? This will work and certainly give some cooling - but I think there are a lot of inefficiencies in this route, besides the concerns with all up weight and vibration from a fast 60-80mm fan. I think the major issue will be the aluminium body of the QHY5-II - in that you will pump the heat from the collar very efficiently (with some insulation), but the body of the camera will be absorbing heat from the atmosphere and via direct conduction from the focuser, in addition to the heat being produced by the camera. Even at 60W on the TEC, I think you will struggle to sustain sufficient deltaT to the internal bulkhead studd/cold finger.

Sorry if I've misunderstood, but not sure that the above method would be optimal (putting it midly) - An external passive cooling option will certainly provide some improvement, perhaps reducing noise by a factor of 2-3 without altering the camera (provided the standard design dissipates the heat out to the body at a reasonable rate). But to cool below ambient you will need to mod the camera/body, either by cooling externally and bringing an insulated cold finger in through the camera body to the cmos or to introduce a small TEC directly to the back of the CMOS. Pretty sure the second option will give the best results.

The more I think about it, the more I'm considering building a project housing to allow me better access to test, without destroying the case - just wondering if I can source a supply of replacement ribbon cables (just in case).

[tonedeaf]

yes inefficient it will be...

i could insulate the casing a bit

and there is always the posibility to "glue" a 6x6 peltier on the cmos, and use the adapter with a cpu cooler...

also i don´t need to drive the fan at 12V and so on full rpm...there is always the way to go with lets say 5V and it won´t have the power to go to full speed -> so less rpm

but you are right, the direct method will be better...

on the other side, with a small peltier on the cmos, i could drive a 2-stage system

but if everything goes wrong...

...let´s face it, some hours of fun in my garden-shed, w/o hearing complaints from my wife

[SnakeyJ]

yes inefficient it will be...

i could insulate the casing a bit

and there is always the posibility to "glue" a 6x6 peltier on the cmos, and use the adapter with a cpu cooler...

also i don´t need to drive the fan at 12V and so on full rpm...there is always the way to go with lets say 5V and it won´t have the power to go to full speed -> so less rpm

but you are right, the direct method will be better...

on the other side, with a small peltier on the cmos, i could drive a 2-stage system

but if everything goes wrong...

...let´s face it, some hours of fun in my garden-shed, w/o hearing complaints from my wife

Very good points and not trying to be dimissive on this as I would really like an effective solution/mod that doesn't involve dissasembling the camera, especially as I may well treat myself to the colour version of this cam at a later point. However, I worked on a very similar approach with my firefly mv which did work, but considerably aided by having a plastic bodied case. Though without effective insulation of the cold finger I was really struggling to achieve deltaT above 5-6C, once the cold finger was properly insulated I was getting 15-18C running a TEC1-2706 at 13V 5A. The fan vibration was visible at high magnification / focal lengths F20+, but more of a nuiscane was the torsion movement caused as the fan cut in and out around setpoint (though a proper variable speed/PWM peltier controller) would eliminate most of this. I'm just about to modify this old setup with for closed loop water cooling which should remove a lot of the weight and vibration issues.

Haven't got a garden shed at the moment, but do have a nice little utillity room/workshop if I can only get rid of a large train set and some of my other junk - I think we can all appreciate/benefit from a few hours of undisturbed project work!

[SnakeyJ]

Peltier sourced - I'm going to use a larger 15x15mm 2V unit (TEC101706) which is cheaply available via ebay - http://www.ebay.co.uk/itm/TEC101706-Cooling-Thermoelectric-Cooler-Peltier-Max-DC-2V-3-79W-/190615123958?pt=LH_DefaultDomain_0&hash=item2c618bf7f6 I paid £ 10.60 ish for two units (2 ordered due to previous bad experience of dropping/blowing one TEC12706s mid project).

My plan is to turn a copper spigot with a 7mm diameter stud 3mm long and 16mm diameter plate 2mm thick (though this may be cut and soldered). This will give me approx 9mm depth to acomodate between the bulkhead and rear of the CMOS sensor (which may require some shims to bring the sensor board forward 1-2mm). This will effectively use the bulkhead and body as a heatsink, which will require some external sleeve, fin and active fan cooling. Add in some internal insulation of the sensor bay and a convenient dessicant system. I'll have to mull this somewhat as the dimensions of the existing enclosure leave little free space - I think at this stage a bigger case would be helpful at least in testing.

I will start work an arduino based PWM controller now I have Vmax and Imax parameters and try to source a 2V 6A DC to provide the TEC supply - mains or 12v-2v DC reg in the arduino control box? Plenty of things to consider....

[lennylovit]

keeping an eye on this thread , i have a QHY5L-II lovely little cam

Lenny

[SnakeyJ]

keeping an eye on this thread , i have a QHY5L-II lovely little cam

Lenny

Certainly is and I shall do my very best not to break it

[tonedeaf]

so mod is made:

40x40 12705 peltier element ~17€

arctic cpu cooler ~10€

manual fan steering 5-12V ~6€

2 component heat glue ~19€ (good that you can close the injectors -> usable for the next mod -> eos450D)

here you can see the ice-build-up!! at full power of the fan (@ lower power no ice)

now the darks made with apt

center crops!

95% gain 300s room temperature, without cooling

95% gain 300s 5V fan low power

the full power dark went somewhere, can find only that one

100% gain 120s 12V full power fan (thats where the second photo was made -> on ice)

oh the whole setup will look like this

[SnakeyJ]

Well that certainly seems to deliver a very impressive improvement/reduction in noise - and very impressed with this over 5 minute exposures - though I'm going to need another camera for guiding purposes! Electrical tape and low density packing foam or perhaps laminate floor underlay would make quite a good insultating layer around the camera body and collar/heat sink. Do you get any histogram/noise information from your capture logfiles?

Certainly proves this camera has a lot more to give, perhaps even some of the brighter deepsky objects will yield decent results.

[tonedeaf]

ok made historigrams from the crops...

...also have seen that the "dark" cap let in some light on the upper right corner (which was cut away, but maybe got there somewhere else)

w/o cooling

lo power

full power

hmm the foaming seems to be a good idea!

also the heat-"foam" (or what this is) between the back of the cmos and the "finger" in the casing, is disturbing me...

...i don´t think that this element transfers the heat that well....

[tonedeaf]

also have seen the error...

the peltier-element was for 27€ and not 17

hmm i´m real new to ap, and don´t have so the experience on the gain settings...

...will be a 90% gain be sufficient?

if so i think the result will be even better...

tomorrow when i´m sober i´ll try a foaming and maybe a aluminium bridge instead of the "heat-foam"

[tonedeaf]

as normal...

can´t stop it...an idea in my head and...

some heat-paste and alu-plate (...between the cmos and the casing-finger)

300s 95%gain full power cooling (ice is building up again...mmmh maybe time to make a mojito )

here´s the historigram of the crop

600s 95% gain full power

600s 90% gain full power

i have to admit i´m satisfied

just waiting for clear skies...can´t wait to light up some DSO´s with this cam

[SnakeyJ]

Those results look stunning - enjoy you Mojito, I'm all set up ready for some Saturn imaging and the clouds have just rolled over - curses

Its a thermal pad BTW - I should be cautious with standard thermal paste as it will flow when warm and most are electrically conductive - if you get it flows between the sensor and PCB you may short out some of the CMOS pins. There are special non conductive pastes and better pads that may be a safer bet.

Looking forward to seeing some real time images - perhaps some of the globs could be nice targets without too much exposure to get nice results.