Firefly MV Mono - Is it worth cooling?

[SnakeyJ]

Ok - A quick and dirty comparison TEC Off vs TEC On on Saturn 24/04/2013.

There was about 60 minutes time lapse between the two captures and the seeing was definitely worse for the TEC on Capture. Both captures are made through Astronomik PP 742 IR-PASS filter at F15 on my SW Exp 200P.

The stacking process removes much of the noise, but no doubt in my mind that the cooled camera gives a better result.

Examining the firecapture log files, the uncooled exposures are slightly shorter (53.76ms v 64.82ms) and Noise (avg.deviation) is reported as 13.19 and 9.36 respectively. Not a very scientific comparison, but even given the relatively short exposures there is a visible improvement. If conditions allow over the weekend I will re-run the comparison with more control and longer exposures!

[Cath]

The cooling makes a vast difference by the looks of it Jake !

The AS!2 camera is so much better isn't it, surprised you haven't put the FireFly permanently to one side looking at those photos.

You've managed to get FireCapture working with the AS!2 camera then I presume?

[SnakeyJ]

The cooling makes a vast difference by the looks of it Jake !

The AS!2 camera is so much better isn't it, surprised you haven't put the FireFly permanently to one side looking at those photos.

You've managed to get FireCapture working with the AS!2 camera then I presume?

Whoops - All capture done on the Firefly MV Mono, but stacked in Autostakkert2 (AS!2) rather than Registax as I find it much quicker to use that Registax (though I still get better lunar results in Registax6). The big advantage of AS!2 is that it re-orders the frames in quality order, which I find much easier than trying to hand select the best frame in R6.

[Cath]

oh!

Sorry, my bad lol

Very impressed with what you've done with the FireFly then Jake!

[SnakeyJ]

oh!

Sorry, my bad lol

Very impressed with what you've done with the FireFly then Jake!

Thanks, though I know I can squeeze a lot more out of it in better seeing and there are still improvements to be made with the cooling and also optimising capture settings and technique. I'm really hoping to get some good seeing, so that I can barlow up to F20-F25 - this will really push the exposure and gain settings, where hopefully the cooling will yield more visible improvement.

Here's Piotr's (Riklauni) test image on Saturn taken from his excellent review on the firefly (stock firefly mv mono, not sure if this is the C11 or 14 - but much more light capture and focal length)

Think I've still got a lot of room for improvement of results!

[Cath]

Amazing really!

The only thing the camera is lacking is long exposure (and a hole behind the sensor chip for easy cooling ability). Maybe if PointGrey were pushed they would add a long exposure ability into the firmware.

[SnakeyJ]

The MT9V022 will definitely never make first choice of sensor for LX, but it would be nice to push this up to 3 seconds or so where I think it would make a really good entry level guide camera.

As it stands for around £ 30 on ebay, it's an absolute bargain choice for Lunar, Jupiter, Venus and Solar. Saturn is really pushing the limits with anything below 10" aperture, though will depend a lot on apparent magnitude and seeing. Though I'm not sure I could ever have justified paying the list price of around 295CAN$ on this!

There are other Industrial cameras out there with much better sensors - the e2v ruby sensor looking very sensitive and interesting, with great products available through IDS, Basler and Point Grey - however they are so expensive you could be buying dedicated astro ccds with decent astro focused software for the price. I'll just have to keep an eye open for a bargain s/h or clearance item for the next project.

Can't help thinking it would be nice to get something modular, so you could change sensors depending on target, but run common electronics/drivers and cooling - I would have thought there was more than enough processing grunt nowadays to cope with the hardware abstraction that would be required. Failing this a common body with standard hardware and flash programable firmware to suit a range of existing sensors. Though obviously not a winner financially for the manufacturers

[SnakeyJ]

Actually an important point to make on the test images - Even with the TEC turned off, the camera is still modded with a substantial (albeit passive heatsink). My previous dark frame benchmarks suggested this gave a 50% reduction in noise over the unmodded camera and this effect should be higher with the lower ambient temperature outdoors.

There have also been a few 'improvements' since those darks were shot (not least some TIM between the heatsink and CMOS), so I will try and get the camera outside this evening and re-shoot comparable dark frames to see how the it all stacks up!

[Cath]

Would love to see the results! .. what you've done so far is well good I think

A PointGrey request has been put through (by a PT employee, I just asked one of them the question) asking for a firmware feature to allow longer exposure on the MV camera. I doubt it'll go any further but you never know!

[SnakeyJ]

Would love to see the results! .. what you've done so far is well good I think

A PointGrey request has been put through (by a PT employee, I just asked one of them the question) asking for a firmware feature to allow longer exposure on the MV camera. I doubt it'll go any further but you never know!

Just off out to shoot the darks now so hope to get posted in an hour or so.

You never know - it would make it a very useful guidecam, though I have never tried the ascom driver out - I won't hold my breath as this cams pretty much bottom of the range and must be close to end of life for Pt Grey, but definitely worth the attempt!

[Cath]

Still writing my software bit by bit. Got it tracking selected stars OK now, but still working on the algorithm.

[SnakeyJ]

OK - a new set of darks. Ambient temp outside was around 11C. Tests done to same settings as previously:

Peltier/TEC Disabled (Passive Cooling Only):

Peltier/TEC Active (Setpoint -3C)

These tests are based on double frame TIFF capture at 752x480x16bit - I have centre cropped a 300x300 pixel square out for the comparison images (as per previous tests).

Something I didn't do before was to test the heat/noise build up on longer captures - so here are the results for 100 frame captures at the same settings (centre cropped from final image for comparison).

Little variation at the extremes, but obviously some additional heat/noise for the middle captures - sorry to miss the penultimate test!

Overall Pretty pleased and the first two test runs are better than back in February - noise reduced by around 80% from unmodded camera. However, I wasn't achieving the same cooling on the camera today - which is probably due to not cleaning and putting new Arctic Silver between the TEC and output Heatsink on my last clean/rebuild. Not too fussed at the moment at my watercooling components should be arriving in the next couple of days and this is definitely the way ahead for the firefly (and subsequent cameras)!

[SnakeyJ]

Still writing my software bit by bit. Got it tracking selected stars OK now, but still working on the algorithm.

Cath,

Looks like you've been burning the midenight oil on your Astrocamcap development - it looks very impressive and I like the trails/positions and think you might be on to something for nifty to assist with drift alignment. The star tracker 3D wire diagram looks very nice and techie and looks like a modular plugin. It looks like you are tracking all 8 trailed items simultaneously, with the focus (and pop info) for the highlighted trail. Certainly seem to be no issues with sensitivity here.

Just one query, noted that the ADC vref is set up to 2.1V and was thinking the normal value was 1.4V - is this to improve the sensitivity and is it pushing the hardware?

[Cath]

It looks like you are tracking all 8 trailed items simultaneously, with the focus (and pop info) for the highlighted trail.

Yes it'll track as many stars as you like, unlimited really.

What I'm thinking is maybe guiding with multiple stars rather than just one. Using multiple stars overcomes noise (like stacking) and improves accuracy (not that the algorithm needs any more accuracy) and also gives you field rotation data and field distortion data. Also, if you loose one tracking star you have more as backup.

Just playing at the moment, just trying to think of new algorithms (and test them out) that might be useful for astro imaging.

Certainly seem to be no issues with sensitivity here.

It's not to bad, but the problem is as the temperature of the camera changes through the night the background noise changes level, which is pain. Definitely need your cooling

Just one query, noted that the ADC vref is set up to 2.1V and was thinking the normal value was 1.4V - is this to improve the sensitivity and is it pushing the hardware?

I was thinking maybe using a higher ADC reference that any circuit noise before the ADC would appear smaller to the ADC, although it doesn't appear to make a great deal of difference. If you want maximum gain you need to go the opposite way as set it to 1.0v reference, but without cooling on the MV it's a bit hopeless to do that.

I don't think it's pushing the hardware, it's just one of those built-in features of the sensor chip really.

[SnakeyJ]

The multistar tracking will be excellent, especially under our often less than perfect patchy skies - I see that MaximDL have a plugin for something similar, providing a positional/guide average from a number of stars, but your idea looks rather more interesting and might just be worth a quick patent search!

Re cooling, I would definitely consider the passive heatsink and fan even if you didn't want to go the whole hog on setpoint TEC - I've got lots of the copper sheet left over if you would like me to knock a heat sink up, though I can't promise the ulitmate in precision engineering but it would certainly help.

Re the ADC VRef voltage I think I am doing this with the camera register (2C) control now added into firecapture (following excerpt from CN post by Pippo)

1) "Gain2x". This controls the digital-gain. Namely which are the bits not considered in the 10bit to 8-bit conversion. The button puts F8 in the register 80 of the sensor. This corresponds to an increase x2 of the signal (independently of the actual value of the analog-gain).

2) "SignalBoost". This controls the reference potential of the AD converter. The button puts 00 in the 2C register of the sensor. This corresponds to an increase x1.2 of the signal (independently of the actual value of the analog-gain).

3) "GainBoost". This sets the analog-gain beyond the 12dB standard limit. It puts 7F in the 820 register of the camera. This corresponds to a 24dB gain level. For the moment the intermediate values of the analog-gain between 40 and 7F are not implemented. If you need these you should use Flycap2 as described above.

Notice that the off-state of this third button (of just this one) does not work, i.e. it does not restore back the standard level of the analog-gain. However, to restore the standard values, you have just to move the standard control of the gain.

The advantage for me is to get more gain out of the system, but without cooling it gets horribly noisy - though I haven't tried this setting without the Gain x2 register, which would provide a more subtle boost from 12 to 14dB gain. I hadn't thought of going in the other direction, though imagine (naievely?) that a higher reference would decrease the max available gain and reducing both sensitivity and noise.

[Cath]

Just to match up what FireCapture does with what I've written ..

1) "Gain2x". This controls the digital-gain. Namely which are the bits not considered in the 10bit to 8-bit conversion. The button puts F8 in the register 80 of the sensor. This corresponds to an increase x2 of the signal (independently of the actual value of the analog-gain).

That's the same as the contrast slider.

2) "SignalBoost". This controls the reference potential of the AD converter. The button puts 00 in the 2C register of the sensor. This corresponds to an increase x1.2 of the signal (independently of the actual value of the analog-gain).

That's the ADC reference voltage.

3) "GainBoost". This sets the analog-gain beyond the 12dB standard limit. It puts 7F in the 820 register of the camera. This corresponds to a 24dB gain level. For the moment the intermediate values of the analog-gain between 40 and 7F are not implemented. If you need these you should use Flycap2 as described above.

Notice that the off-state of this third button (of just this one) does not work, i.e. it does not restore back the standard level of the analog-gain. However, to restore the standard values, you have just to move the standard control of the gain.

The little "Max" button by the camera gain slider does just that.

The advantage for me is to get more gain out of the system, but without cooling it gets horribly noisy - though I haven't tried this setting without the Gain x2 register, which would provide a more subtle boost from 12 to 14dB gain. I hadn't thought of going in the other direction, though imagine (naievely?) that a higher reference would decrease the max available gain and reducing both sensitivity and noise.

I'd stay with an ADC reference voltage of 1.0v then for maximum gain as your cooling has reduced the noise level some what

FireCapture is a really nice program, I like it myself, he just needs to give the user some easy buttons etc to use rather than the user having to access the camera registers directly themselves.

[SnakeyJ]

FireCapture is a really nice program, I like it myself, he just needs to give the user some easy buttons etc to use rather than the user having to access the camera registers directly themselves.

They are a little tucked away, but Torsten has added some nice tick boxes in a drop down menu (from version 22beta37 and now in the release version 22). It is a very useful bit of software, which I find nicer to use than Sharpcap (though haven't revisited this for a while) - but it with all these things the more you add in, the more complex the UI becomes. It would be quite nice to float all the camera controls in a single separate window and this could then be customised for each of the supported interfaces more readilly. With the drop downs you often forget what is ticked/set and then wonder why the standard controls are locked or have restricted ranges. My only other minor annoyance is that it doesn't add a letter in to the capture file filename to denote which filter is being used - it gets boring after a long night to manually check the logs and rename 50-100 captures!

And thanks for the clarification on your controls!

[Cath]

I've got lots of the copper sheet left over if you would like me to knock a heat sink up, though I can't promise the ulitmate in precision engineering but it would certainly help.

ooo wa .. thank you Jake!

The only trouble is the camera is also being used with another project and so can't really attach anything to the camera to make it bigger. Not that I have tools to do any metal work anyway etc lol.

[SnakeyJ]

I will have to remake one to be optimal for the watercooling, so no sweat to knock two up at the same time while everything is out - I'll keep a similar design to original, but will make it shorter and bend it 90deg, so that it exits the bottom of the case along the join line and folds back towards the rear of the camera - leaving sufficient material to attach a 40mm square TEC underneath if required. It will make the camera 30mm deeper in the same direction as cable connections, and perhaps add 5-6mm to the height where it exits the case and turns - but otherwise fairly minimal impact on physical dimensions. I don't want to risk fouling the focuser / filter wheel by bending the h/s forward of the camera.

[SnakeyJ]

Not directly related to cooling the firefly, but just found a very good guide on cooling the QHY5, using very smally TECs connected directly to the back of the cmos sensor - some very useful info for anyone considering a little more finesse than my somewhat brutal approach:

http://thx8411.over-blog.com/pages/Cooling_the_QHY5_imagerautoguider_step_1-3170312.html

http://thx8411.over-blog.com/pages/Cooling_the_QHY5_imagerautoguider_step_2-3239743.html

Of course the back of the cmos here is directly accessible through the PCB which makes life somewhat simpler - but some great info and links.

[nightvision]

Cath that is

Still writing my software bit by bit. Got it tracking selected stars OK now, but still working on the algorithm.

Fascinating software you are putting together, do the tracked stars deliver different outputs, if so do you plan to average them or do something more sophisticated?

[Cath]

It's work in progress at the moment Tony, so not not sure what path it'll take. Anyway, this is Jakes FireFly cooling thread

[SnakeyJ]

No complaints from me!

[Cath]

No complaints from me!

noooooo!

Anyway, I've added a video streaming option to the program now, needs improving but it works and lets me see what's being captured inside the house while I'm out at the telescope by watching it on the Android tablet.

[SnakeyJ]

Very impressive add in Cath - is this being compressed on the fly as H264 stream? I imagine that would require quite a lot of system resource not to mention put some load on the wireless bandwidth (dependent on FPS and Resolution).

I had a couple of bits through the post yesterday Silicon tube (though too stii, heavy and cumberson for final use) and my cooling radiator - still waiting for the pump and fluid, but hope this will come in time for the weekend and I should at least be able to test.

I also managed to get the new QHY5L-II mono out for a couple of test runs - it just blows the socks off the firefly in terms of speed/sensitivity, comfortably giving me twice the frame rates on a bigger 400x400 ROI and a whole lot less noisy! OK it did cost almost 10x what I paid for the firefly (though around the same retail list price) - will be nice to see if this results in Pt Grey dropping and clearing some of their older but better models like the fleas.