Firefly MV Mono - Is it worth cooling?

[Cath]

Just had a look on ebay to see if they were still around, but prices are currently £ 65-90 + delivery. Wish I'd brought two of these at the original price

The MV camera do you mean? .. if so ..

http://www.ebay.co.uk/itm/Point-Grey-Research-USB-2-0-Digital-Camera-/261170301724?pt=LH_DefaultDomain_2&hash=item3ccef68b1c

£29 + £10 p&p to the UK

[SnakeyJ]

The MV camera do you mean? .. if so ..

http://www.ebay.co.u...=item3ccef68b1c

£29 + £10 p&p to the UK

Thanks again - not sure whey this didn't come up the first time but this looks a lot more like it. I will order a second just in case I blow the first one during modding

[SnakeyJ]

Ok - I finally have all the bits and have assembled the cooling & enclosure to test.

1. External (CPU) Cooler, Peltier, Aluminium Bulkhead, Internal Circulation Fan.

My first test run showed the temperature heading in the wrong direction, so I quickly flipped the tec over to (note to others lettered side is hot, though this isn't stated in the datasheet).

2. Cooling & Enclosure Ready for Assembly:

3. Bench Test after 10 minutes (Not terribly impressive ):

The top lcd is the read out from a probe at the base of the enclosure. The bottom readout is a probe mounted on the circulating heatsink within the enclosure.

Not terribly impressed with the results on this - just 6C below ambient within the room. All the thermal junctions look good and are spread with thermal compound - I was hoping to see more like 20-25C below Ambient temperature which is well within specs of the TEC1-12706. The controller claims to relay up to 30V 8A and there should be plenty of juice from the 15A rated PSU.

Time to go away and think on this to see if I can spot the fault and improve the design.

[SnakeyJ]

OK some revision to the test rig and time spent on learning the quirks of the peltier controller (functions operate in reverse for cooling which is somewhat counter intuitive):

New test rig:

Not optimal as I'm connecting the hot & cool sides with the four steel mounting screws but starting to see far better results. Red LED readout is the coolside temp directly over the peltier, black LED is the temp at the end where the cold tongue will contact the cmos.

Test rig running with some rudimentary thermal insulation - Ambient temp is around 21.5C (must turn heating down).

looks like I'm getting -14C from Ambient, though probably much better if I can mount/secure without the screws.

I'm pretty sure this is the design to aim for as the cooling looks better, lower power consumption and the final weight will be significantly lower

[Cath]

Looks like your getting there Jake

Those screws will definitely be a hindrance.

[SnakeyJ]

Looks like your getting there Jake

Those screws will definitely be a hindrance.

Had to call it a night yesterday as have two poorly kids, though did manage to get 5hrs sleep!

But by midnight I had freezing condensate on top of the cold finger I'm going to do some more redesign this evening, swapping out the CPU Cooler for the much smaller/neater VGA cooler and will make a plastic bracket/clamp to hold peltier and fan to the cold finger. Hopefully this gets the tip to around -5 (-25C from Ambient), which is my new target temprature, then I can shape the tip to fit around the CMOS and do another run of darks for comparison with the originals.

If the results are promising enough I will try and find a suitably rugged 6-8 pin connector (male/female) make up a cable set and get it on the scope - then probably a three week wait for some clear sky for a view of Jupiter/Saturn - Lunar probably going to be too bright (not requiire enough gain) to provide a decent test.

[Cath]

I've just added a function to read the cameras sensor temperature into my software but it's saying it's at around 55 deg C, which sounds way to much but I am using the graph data from the sensor data sheet. Without an external temperature probe on the sensor I can't confirm it's real temperature.

If I put the camera in the freezer for 5 minutes and then try again it's temperature is reading about 20C lower so I guess it's reading OK, and the image looks quite different after I cool it, though it don't take long to warm back up again!

If the sensor really is running at around 55C then I'm not surprised it's noisy.

Also found out need to use flats (pure light frames) to scale the more sensitive pixels (some of them are quite a bit more sensitive) and remove the vertical lines that are left over after using dark frames. Though taking light flats isn't so easy for me.

Without and with dark frame subtraction ..

Hope the kids get better soon!!!

[SnakeyJ]

Cath,

Fahrenheit to celcius conversion? 12.7C possibly, though sounds too cold for indoors unless the heatings packed in.

I've downloaded, but haven't had a chance to play with your firefly MV Capture as the cameras still in bits awaiting the cooler - may get my spare through from Canada before the week is up. It does look like your flying ahead with this and hopefully I can give it a test run very soon. The one burning question that remains is where do the lasers fit in to the plan (from your original thread)? Are you doing/planning something clever with adaptives ?

I have used Darl Frames is Registax, but never attempted Flats or Bias - Getting good Flat frames seems to involve a fair amount of art and time to achieve even illumination and I think needs to be done each session, or if you change orientation of scope/camera. Both of these are stacked / averaged to give master flat and bias references for subtraction from the final image. The bias subtractions to eliminate read noise should be relatively straightforward though, and from my limited research you just need to set up a library for the different frame rates you intend to capture at, updating once or twice a year. Not sure if these offer the same levels of improvement as the dark frame subtractions and may be a case of diminishing returns - however definitely worth a couple of hours experiment to have a look.

Actually the white lines may well be read noise, so bias frames would definitely help here.

Kids ok - the usual colds/fevers but one with bronchitis!

[Cath]

Think I got the temperature reading OK now, just need to adjust the offset and scale a little bit.

I can confirm the camera improves a LOT with cooling Jake, even 4 degC change makes quite a difference.

The sensor appears to heat itself 15 degC above ambient. So if you can counteract that well enough you'll be surprised with the difference it makes!

[SnakeyJ]

Think I got the temperature reading OK now, just need to adjust the offset and scale a little bit.

I can confirm the camera improves a LOT with cooling Jake, even 4 degC change makes quite a difference.

The sensor appears to heat itself 15 degC above ambient. So if you can counteract that well enough you'll be surprised with the difference it makes!

Thanks Cath, sounds very encouraging and hoping to get a good 20-25 down below ambient so looking forward to some proper testing tomorrow night.

I've remodelled the heatsink and benched this with the VGA cooler fan tonight, but this did not provide sufficient cooling to the hot side which was running somewhere around 44C! Swapped back over to the CPU cooler and reduced the hot side heatsink to to around 24.2C (3.5C above Ambient). The tip of the cold finger heatsink is running at just above 2C, though I need to clamp this all up properly and get some thermal compound in.

Unfortunately I left the needle files at work, so will have to complete and mount this tomorrow for some new dark frames.

[Cath]

The one burning question that remains is where do the lasers fit in to the plan (from your original thread)? Are you doing/planning something clever with adaptives ?

lol, no not adaptive correction or anything like that, wish it was.

The lasers are for distance measurement intended for UAV's, the camera monitors the spot position and with a little bit of trignometry you have your distance to whatever, but have left the guys to it for now, want to see if the camera can be helped in the astro world with software.

I'm surprised at the shear size of heatsink you need, although I've never used peltiers before. Very interesting. Maybe it'll dissipate the heat much better outside. The camera will be tiny compared to the heatsink lol, or at least by the looks of it.

[SnakeyJ]

lol, no not adaptive correction or anything like that, wish it was.

The lasers are for distance measurement intended for UAV's, the camera monitors the spot position and with a little bit of trignometry you have your distance to whatever, but have left the guys to it for now, want to see if the camera can be helped in the astro world with software.

I'm surprised at the shear size of heatsink you need, although I've never used peltiers before. Very interesting. Maybe it'll dissipate the heat much better outside. The camera will be tiny compared to the heatsink lol, or at least by the looks of it.

Maybe the diy adaptive optics could be the next project!, though 5mW lasers are probably going to be a little small and the cost of the optics probably a little beyond reach... Always fancied the arduino pilot UAV hexacopter, maybe in 3 or 4 years when my lads a little more capable and I can watch, offer sage advice and teach him how to use the tools properly.

Also surprised on the difference between the two heatsinks, if I can maintain the hotside at 3.5 above ambient with the CPU cooler I struggle to see why the VGA cooler is failing so badly @ 23.5 above ambient. I need to achieve a weight saving here so will have to source something which is a good median - I'd rather have lighter with a more powerful noisier fan than have a massive heatsink hanging off the camera at the end of the focuser

[SnakeyJ]

Another thought on the cooling fans - I may not be running the fans at full speed so may need to gnd or raise the voltage on the blue control header.

There is a control reference for the four wire fans at http://www.formfactors.org/developer/specs/REV1_2_Public.pdf. Older three wire fans are just +ve, -ve and tach so just requrie voltage speed control.

[SnakeyJ]

Sussed it - late so will post proper results tomorrow, but temps right down to -5C (25C below ambient) on idle, rising to +3.5C at highest gain under test (though only shooting single frames). Noise 25% of my original figures back on page 1 and massively better dark frames. Think I can improve this further, but was a little scared to put thermal compound around the shoulders of the cmos chip as its very likely to get spread all over the place during re-assembly. Certainly looking a whole lot better

Late now so turning in.........happy!

[Cath]

ooh you've actually put your cooler onto the camera now then!

Close up photo time!

Have you measured the camera sensor temperature with one of your probes I wonder?

I'll take a look at bulb exposure for the camera in the next few days. If an external hardware trigger is needed then maybe just set one GPIO pin as output and wire it directly back onto the trigger GPIO input pin and toggle the output pin (through software) to trigger a bulb exposure.

[SnakeyJ]

ooh you've actually put your cooler onto the camera now then!

Close up photo time!

Have you measured the camera sensor temperature with one of your probes I wonder?

I'll take a look at bulb exposure for the camera in the next few days. If an external hardware trigger is needed then maybe just set one GPIO pin as output and wire it directly back onto the trigger GPIO input pin and toggle the output pin (through software) to trigger a bulb exposure.

Hi Cath - you do like your early starts!

Cheers - I've just collated all the details and cropped the dark frames and will post in a few minutes.

For last nights tests, the probe was running on the cool side of the peltier - I will try this tonight, but was planning to Beta your software as I can calibrate with the cameras sensor and put a probe on the tip of the cold finger. Earlier tests (prior to assembly), I have about 3C gradient along the cold finger and some thermal conduction lag/delay, though these were done prior to insulating and installing and do not take the thermal output of the camera in to effect.

I ordered the GPIO header/cable from ClearView Imaging (Pt Greys UK Distrib) for £ 7 + VAT. They initially wanted £ 20 courier, though managed to negotiate to £ 2.50 for RM Signed and they were very helpful. Silly money really for such a minor part, but does save a lot of time messing around.

I will try the Bulb exposure as suggested, though it is not clear from the Tech Reference whether this will only trigger up to maximum exposure time - however I will soon find out!

[SnakeyJ]

Ok, its taken a while to get here, but after much tinkering I managed to get the whole setup running as a prototype and some pretty good results - though still some more development/refinement ahead to reduce weight and I need to seal the camera body and introduce some dessicant to prevent internal condensation.

I've managed to retain the standard plastic case, with a small slit cut through the side for the cold finger. I have put a thin plastic gaskit around the CMOS to ensure the heatsink does not short out any board components and also covered the base of the cold finger with insulating tape. I've then cut a small slit out of the base of the lens holder on one side, such that the tip passes around through and is clamped in place. The tip is cut to fit around inside the lens housing, with a small square cut out to allow the CMOS sensor to see. Much time spent with needle files and bench magnifier to achieve a fair job on this (could be better with some practice). Due to the risk of mess I did not use thermal grease around the cmos/cold finger junction due to risk of spreading during reassembly. I will try and source a less messy TIM alternative - thin pad or similar.

The test setup is shown here with the cooling active - Cooldown from Ambient was approx 8 minutes to -5C (much quicker than previously). This was switching in an out, but did not play with the set point settings/sensitivity - LL was -5C, UL may have been 0C, but not certain. The temp sensor is located at the centre of the peltier on top of the cold finger, so there will be some variance and lag along the length of finger. Max temp displayed during testing was 3C, so this is definitely achieving something! The black LED on the RHS is not connected to the rig, but displaying ambient room temp.

For ease of reference here are the darks shot back on 12th Feb with the unmodified firefly mv mono:

Here are the darks shot with the modded camera - showing 50% improvement from the passive heatsink at ambient temperature:

And here are the results from the first run with cooling (-5C to +3C):

I think this can still be improved with a decent thermal interface between the cold finger and CMOS and a few minor tweaks - but at max gain there's something like 75% reduction in noise from the original unmodded results a very noticeable improvement. However, these are all single image captures, so will have to test how quickly I can dissipate the heat during continuous video capture.

Still someway to go to get this running - but hope to get a chance at Saturn if conditions allow during the IOW Star Party next weekend. The dimmer target should give a pretty tough real world test and I have a 3 mornings of images from the unmodded firefly for a rough/ready comparison (they look very grainy to me). Hopefully if the bits arrive in time I may also get a go at some LX exposures if I can make this work via the GPIO interface.

I'll do some more testing tonight and hopefully get some more temprature data from the cold finger and the CMOS temp sensor (via Cath's new programme). Might also try for some IR Images of my darkened lounge.

Things still on the to do list:

CLS filter for the camera nosepiece to seal and prevent condentsation.

Dessicant package.

TIM between CMOS and cold finger.

Smaller/Lighter CPU Cooler pref with variable speed control.

LX control via GPIO - here's hoping.

Darken/paint the upper side of the cold finger around the CMOS.

Clean the CMOS, lens housing, nosepiece and camera internals.

Check that there's no vigneting or obstruction caused by the slightly increased focal length and the copper heat sink.

Investigate shielding/passive cooling for other pcb components.

[Cath]

Well done Jake!!

The cooled noise images look much more even/cleaner than the uncooled ones. The white peaks are gone on the cooled sensor - much better.

If I'm typing messages at silly 4am times then that only means I've woken up and having a cuppa before trying to get back to sleep again. Seems to be an 'age' thing

You can get double sided sticky thermal pads, they use them on PC graphics cards a lot instead of thermal paste.

If you do try the software then let me know if you have any problems or whatever, I've only been using it in Windows XP. Only really doing it out of interest and so never really intended to release it but it's there if anyone wants to play with it. I'll add a continuous noise level reading to the software as it sounds like it would be useful. Just been adding star tracking into it (you draw little rectangles around the stars your interested in) but haven't yet finished that bit.

What Peltier device are you using?

If you want to make the camera pcb water/moisture proof then you can paint it/dip it in Plasti-Dip ..

http://www.plastidip...astiDip&stage=1

It's a rubber plastic coating, and it can be pealed off if you ever had a need too later on.

or their Liquid Tape ..

http://www.plastidip.co.uk/eStore/index.cfm?Plastidip_Liquid_Electrical_Tape_-118ml&stage=3&colour=Black,White,Red&pid=PDL-0019

[SnakeyJ]

Cath,

Thanks again - I hadn't come across the plastidip products before but they look really useful so will give this a try and am trying to source the best TIM - it wants to be quite thin and sticky but not too adhesive so things can be stripped down for cleaning - the vram pads may be perfect if not too thick and I may even have some knocking around the office somewhere.

The peltier is the TEC1-12706 which is a 60W 40mmx40mm single stage ceramic faced peltier - commonly available via ebay for a couple of pounds including postage from China. The specs are @ http://www.hebeiltd.com.cn/peltier.datasheet/TEC1-12706.pdf

Probably this is a little keen and power hungry for the final production model and I would think the 40W package is probably going to be sufficient in a small and neatly engineered package.

If you want to get really aggressive with the cooling then you can stack the peltiers for a two stage cooler, which should give you 50C below ambient - but more risks and heavier power requriements. I've read a fair bit (though not understood all), but understand that once you get below -20C the benefits of deeper cooling are a diminishing return at least for CCDs.

Software may be interesting as I'm on Win 7 x64 (Laptops) and Linux at home, but won't hold you to anything on this - though keen to have a look as the package, interface and features looks really useful.

If this won't play I may be able to read the register directly from Firecap or FlyCap - sure I saw something about manualling accessing the registers:

Register 82Ch gives the temp in Kelvin (according to the tech reference for the firefly/fea).

[Cath]

Well the software seems to work OK on my Windows 7 64-bit, but if you do end up trying it and theirs any problems let me know.

That 82Ch register won't return the sensor temperature value, Firefly seem not to want to allow it for the MV, so you have to read the sensor register directly.

Anyway, your busy enough at the moment with your cooling, which I'd like to see the real world image results from

[long_arms]

I've been following this thread with interest (although I don't understand anything thats going on ) but those recent dark frames sure are impressive!!

If you find a lightweight solution thats within my capabilities I would love to do the same to mine.

Are there noticeable vibrations from the fan?

Dan

[SnakeyJ]

Dan - its neither difficult or expensive and not too much work - though I started on an over complicated idea which is a bit of a dead end. I would say £ 40 for parts if you sourced everything new. I also spent another £35 on a regulated psu - but this was something I wanted for the mount anyhow so have discounted.

The most expensive bit was the peltier controller, which was £10 pound + £ 20 shipping from hk - a luxury but nice to have set point control on this - if ur reasonably aufait with electronics and a soldering iron, you could build your own and save a few quid.

Not sure if it would be worth the effort at outset, though looking quite viable now. Most of the bits will get reused on other projects and for less than a tenner in parts I can add active cooling to another camera.

Looking at the results even a passive hs & fan would make a real difference.

Above all else Ive learned a bit & kept busy on some cloudy nights

[SnakeyJ]

Dan,

Sorry missed the bit about vibration - wouldn't have thought this would be an issue unless by fluke I hit a particular resonating frequency - but will try once I can get this all mounted and tested. Looking clear for tonight so I will probably stop tinkering and get out to use the scope.

[SnakeyJ]

Just checked the weight of the modified camera - some 400gms (though most of this is the oversize cpu cooler!)

[SnakeyJ]

Cath - I managed to have a first run with your app last night. I was hoping to do some proper imaging, but a fantastic start to the night prove very short lived and the clouds had rolled in by 21:30hrs.

Instead I did a little indoors testing - took the nosepiece out and replaced with the supplied lens, downloaded your MFC and tried to run on my Win 7 x64 spare laptop. The only slight hitch was the following dependency error:

FireFly_MV_Camera_Capture.exe - Ordinal Not Found. The ordinal 12384 could not be located in the dynamic link library mfc90.dll.

Quickly resolved by Google & installing the Microsoft Visual C++ 2008 SP1 Redistributable Package.

Like how you've laid this out and the layout is very logical and clean - leaving a good clear space for the captured image. It detected my Firefly straightaway and started displaying the live capture. As this was the frist time run with a picture I was a bit scared I'd damaged my camera as the default gain is very high and I had lots of grain/random noise but precious little image. A quick play around with the settings and I was able to get recognisable landmarks, but as I've pushed the lens housing forward I can't quite achieve focus with this lens. Out of interest the initial temp reading for my ff was 8C, with the preview capture this soon rose to 15C (about 5.5C below ambient room temp) with cooling switched off.

I did 3 captures of 500ish frames to get a feel and see if the temprature moved around a lot, but seemed very stable unless I maxed the gain or played with the settings.

I couldn't work out the dark frame function - it said it was capturing darks, but no clue where they went and then I could not work out how to use the dark frames - prob my stupidity, but in defense it was getting late!

to get anything sensible output wise I had to stack the frames in registax, but got the following results:

Ambient

Cooled:

Camera temp sensor displayed -4 constant through the capture.

Not good images and out of focus to boot, but the amount of noise is considerably reduced by the cooling, though there are still scan lines present which seem more obvious in the cooled version.

I also did the same tests in fire capture and got the following results:

Ambient:

Cooled:

I've lost my screen shot of the settings in you app, but the fire cap settings are below (from cooled session):

Camera=Firefly MV FMVU-03MTM

Filter=L

Duration=26s

Frames captured=200

ROI=752x480

FPS=7

Shutter=36.47ms

Gamma=0 (off)

Gain=1200

Exposure=7

SignalBoost=0 (off)

GainBoost=0 (off)

Brightness=178

Gain2x=0

Histogramm(min)=472

Histogramm(max)=4092

Histogramm=99%

Noise(avg.deviation)=255.26

Conclusion - less noise when cooled, but results in this test not quite so impressive as the darks, but probably due to the much shorter exposure times. I will have to try something through the telescope to achieve focus and do some more serious tests and could also do with cleaning the sensor and finishing off/sealing the camera prior as per the to do list above. I really do like playing and Cath's application excellent once I figure out all the controls/options.