Experience servicing NUC shutters (FLIR Boson more specifically)?

[DaJMasta]

I've got a Boson core that was in sort of rough shape, some dings in the lens, flaking coating, some damage on the lens body, but which seems mostly functional.  One immediate peculiarity was that when I got video out of it, I would get a negative image of what it was facing, and I soon realized that while the NUC was triggering and the shutter was making noise - it wasn't actually closing.  I removed the lens and manually triggered the shutter a few times and saw no movement, and when I put an even temperature target over the whole frame and trigger the NUC, I get a normal looking image out of it again.

In any case, while I expect FLIR will probably sell me a shutter, it will probably way more than I'm willing to spend, so do you have an idea what could be wrong with the shutter or if it's possible to really service.  By all accounts, it sounds exactly as I would expect a shutter to sound, which suggests to me that the solenoid is working and its action is intact, but I assume the parts it moves are quite small and specially sized for the device, so they may be very difficult to fabricate a replacement for, but I was wondering if someone had some experience that could shed some light on it before I try to crack it open.

[Vipitis]

I had a shutter not fully open and not fully close on my older core. It was a matter of just fine tuning a screw and carefully cleaning a bit of debris off the flag that caused friction.

Images could help here.

[Fraser]

Sadly I have never had a Boson pass across my bench so cannot advise on the FFC flag fault.

There are some nice pictures of a Boson teardown here that may help with disassembly.....

http://7400.me/2020/05/11/flir-boson-320-60hz/

As you state, the fact that you can hear the FFC flag solenoid suggests that it’s armature is being driven which us a good start. The FFC flag assembly is a development of that used on the LEPTON is is not normally an item that is serviced. This may make its disassembly a little challenging if FLIR chose to heat stake the casing. I suspect your BOSON is the victim of a drone crash so you may be lucky and find that the armature has just disconnected from the flag. These assemblies are pretty simple so not much that can break 

Good luck with the repair.

Fraser

Update: Photo of FFC shutter assembly added from provided teardown link.

[Fraser]

This excellent teardown of the Lepton FFC flag assembly may help you to understand its design and simplicity. Expect similar inside the BOSON flag assembly.

https://www.richis-lab.de/Opto04.htm

Basically a solenoid armature acts upon a plastic lever arm that in turn moves the flag. Not much to break due to the low weight of the parts involved but plenty of opportunity for the flag mechanism to become disconnected or stuck. The armature (donunt shaped magnet) could have just detached from the plastic lever arm so turns without transferring the movement to the lever.

Your challenge will be the disassembly stage as FLIR tend to use heat staking in place of fasteners on these parts. A sharp scalpel should remove the melted plastic ‘overhang’ though. Reassembly may be done using a suitable adhesive to avoid trying to reuse the heat stake pin remnants.

If plastic parts like a pivot has broken, I use plastic weld delve to to repair them as it is superior to adhesives. The plastic needs to be solvent dissolvable though.

Fraser

[Fraser]

I am attaching pictures from the above teardown in case the page ever disappears.

Fraser

[DaJMasta]

Took the lens off to have a look today, the shutter on my Boson is a considerably thinner module that lacks the step down area around the sensor (and the thickness that goes with that).  You remove the lends thread and front cap with four hex bolts (same size and thread as the M1.6 VPC mount) and then the shutter is exposed, with a small round nib on the lower right in a curved channel that's the actuation end of the shutter.  I pushed it up and the shutter closed, with a bit of grime on the back.

It's been working ever since 


I guess I didn't really need to worry, no need to even take apart the shutter module, but it was only 2-3mm thick.

[Noopy]

I am attaching pictures from the above teardown in case the page ever disappears.

Fraser

That won´t happen the next decades... 

[DaJMasta]

Been playing around with image settings and palettes for a couple days and I think I've got a feel for it, while you can get some good images out of it, it definitely doesn't has the same performance as the Tau 2, even with digital video instead of analog.  I was able to get rid of most of the line artifacts just by turning down the Digital Detail Enhancement setting, but I wasn't able to fully remove the vignetting at the top of the sensor (could be that the lens calibration needs to be redone, but this has been consistent with most Boson stills I've seen).  I had some issues with very quick AGC autoscale transitions which sometimes made a sort of flickering effect, even when the smoothing for it was set very high (software/firmware bug?), but they generally happen less with time after changing a setting.  The one really notable pitfall that I don't think processing can overcome seems to be the dynamic range of the sensor.  Setup to look at my face, when choosing a high contrast palette I can see temperature detail on a Tau 2 and an E60 that simply doesn't show up on the Boson, and with the slightly higher background noise and vignetting, the dynamic range of the whole thing seems notably lower.  My comparison was using a "consumer" grade Tau 2 and a "professional" grade Boson, and in my experience the supposedly higher end Boson performed worse.

It's a nice looking image, for the most part, and turning up the digital detail processing does bring out some features of the image that could be useful to see (and the detail enhancement looks vaguely higher resolution or slightly similar to MSX to me, given the sharper edges), but I hope FLIR can get some more sensitivity and a lower noise floor out of their next small size cores.

I made a probably too long video comparing a Boson 640, a Tau 2 640, and an E60, going through a few palettes and settings and showing off some of the quirks of the Boson that doesn't seem to be part of either of the other two.

[Vipitis]

For the boson the gain control is quite sophisticated; perhaps it needs the fine tuning. I don't think dynamic range will be an issue with temperatures of human skin.

I think there is another document to be found - but this one explains a few parts already. https://flir.netx.net/file/asset/15755/original/attachment

Offtopic: since you are working with multispectral imaging - are you aware of the concept of chromatic order? It's used in astronomy when you combine 3 channel narrowband images back to RGB. I wonder if a similar idea works with broadband UV, Vis, IR.

[Fraser]

The Boson is disadvantaged on two fronts compared to the TAU2 core...

1. It uses 12um sized pixels that are challenging on the signal to noise ratio front. Remember, it is always better to improve the S/N ratio at the ‘front end’ than to try to digitally process a signal with poor S/N ratio in an effort to pull more signal out of the noise. 12um pixels are pushing the OEM’s to the limits of their current image processing abilities. This is one reason why the 12um pixel DRS Tamarisk has enhanced area pixels to improve the S/N ratio.

2. The Boson uses the Movidius video processor and FLIR had to learn how to use its extensive capabilities just as the Movidius chip manufacturer was bought by Intel. There is more to that story but I will draw a veil over that.

The BOSON had the potential to be an excellent imaging core that had the size advantage seen with the Lepton yet the image quality of the larger TAU2. So much promise but sadly, from what I have seen, it has not lived up to the expectations of end users. Such a pity.

I do not see the next generation of mini core using the Movidius chip. Looking forwards, FLIR have limited options and likely want to recover their investment in the BOSON project before embarking on the release of a replacement.

Fraser

[DaJMasta]

Not that I ever expect to hear, but I'd be interested in how financially successful the Boson has been for FLIR.  Given the popularity of it for drones and thermal scopes and such, I wonder if just the fact that it is small and capable in a way that the consumer market has had access to has meant that it's a runaway financial success, even if it hasn't quite been up to snuff in terms of image quality.


Thanks for the note on the Boson image adjustments, indeed one of their example gain pictures seems to be able to differentiate temperature gradations smaller than I'm seeing on mine.  I haven't dove into the SDK, just the control app, and it could be that just the combination of adjustments needed hasn't been clear to me to choose and try out.  I also see that the highest level row of pixels in most of the installed palettes look pretty similar, so maybe there is detection but the chosen palettes just aren't up to the task of making it that visually distinct.

As for chromatic order, I hadn't heard the term but am aware of the idea - the reason I haven't looked into any compositing yet is that all of my cameras are discrete (and usually different FoV) so the different images would need to be transformed somewhat before being laid over each other to line up, though that kind of approach could certainly convey a lot of information at once (I'm interested in doing the same with single high dynamic range images, since a monitor generally displays 8 bits a channel, it would be interesting to map a 14 bit scientific camera to a color space so you could see all the gradations on a standard display).  I recently got a couple of old filter wheels and I'm hoping I can get a fairly broadband camera behind it to do similar kinds of experiments, but without the geometric adjustments for multiple cameras, though I have to actually check what filters I've got (or would want) to know what I could actually see.

[Bill W]

Not that I ever expect to hear, but I'd be interested in how financially successful the Boson has been for FLIR.  Given the popularity of it for drones and thermal scopes and such, I wonder if just the fact that it is small and capable in a way that the consumer market has had access to has meant that it's a runaway financial success, even if it hasn't quite been up to snuff in terms of image quality.

Some commentary on the Teledyne - FLIR tie up
https://optics.org/news/12/1/4

It suggests to me that FLIR 'retail consumer' lines are as good as dead from a new product standpoint.  FLIR were already heading away from it and are now becoming part of Teledyne.  Could be more of interest in the shareholder documentation when it comes out.

Boson was surely intended to take over a lot of the Tau volume (but be cheaper to make) and also improve performance where the Lepton is being used in professional applications (eg Fire).
I can only see that currently it fails on both counts and that it is only of use where size matters above all else and imaging conditions are fairly benign.

Are Teledyne likely to put up the cash to rewrite the Boson software so it works ?
Will there be anyone left at FLIR that knows how to ?

Bill

[Fraser]

Bill,

With regard to FLIR moving away from consumer grade products, that is exactly what I would expect to see. I was very surprised to see FLIR entering that market when their core competence was making expensive professional products for ‘intelligent customers’. The term ‘intelligent customers’ is not intended to be an insult to consumers... it is just that consumer products tend to have quite a high overhead in terms of customer support when compared to products sold to professionals who know what they are buying and how to operate it correctly.

Teledyne will likely do a Profit and ‘Total cost’ (including support) analysis on the products made by FLIR and I would not be at all surprised to see the consumer products either dying off over time or being sold off as a separate entity, much like Indigo used to be. Core and microbolometer licensing agreements would be needed for such to happen but that is not a barrier to such a move.

If the BOSON proves to be any sort of embarrassment to the FLIR brand I feel sure that Teledyne will kill it off as they did not experience the financial investment and attachment to the project that some at FLIR possibly did. Teledyne is all about quality and reputation for high end systems and I would expect them to pull FLIR back up into the ‘professional thermography only’ arena and away from the budget end of the market.

Fraser

[tske]

Those sensor package looks so cost-cutting, I wonder if they can hold the vacuum as long as a proper packaged one, such as in Flir Photon.

[Fraser]

They are MEMS devices that provide excellent reliability. There is no reason why the vacuum should fail more easily than the traditional soldered and bonded ceramic/metal modules. The greatest issue is poisoning of the vacuum via component outgassing as the vacuum volume is so small. Modern MEMS technology microbolometer have proven to be very reliable in service.

Fraser