OpenIRV. ISC0901B0 (Autoliv NV3, FLIR E4/5/6/8) based opensource thermal camera

[tmbinc]

VGN, thanks! I'm deeply impressed by your reverse-engineering, electrical design, industrial design _and_ project presentation skills. I can't wait to use your design and play around with it!

I think there are great benefits of this project that go far above "getting a cheap thermal camera" (which is a very great goal in itself!). It will bootstrap open source efforts to operate on raw thermal sensor data and I see a realistic chance of building an image processing pipeline that matches (and exceeds) that of commercial products. The math is not trivial, but also not super hard, and it is about time that we don't rely on companies that tend to intentionally cripple image quality (noise generator, mandatory fusion and cropping - WTF) on the affordable range of thermal imagers, and take this back into our hands.

[VGN]

Possible configuration without the focus board. In this case we should change the bottom enclosure part. It has the same outline, but different height (13mm to 7mm).

[VGN]

Welcome to my GitHub repo!   

https://github.com/OVGN/OpenIRV

All mechanical parts STEP models are avaliable. You can try to print some of them for a test. That will be cool if you show your attempts to us, especially gears. I have used SLA printing, but didn't try any other technologies. Keep in mind that this parts are not release version, I'm going to make some changes, even not backward compatible.

[ajw107]

That is wonderful news, it's quite late here now but just got my SLA today, so will definitely have a play when I get round to assembling it later this week.

Welcome to my GitHub repo!   

https://github.com/OVGN/OpenIRV

All mechanical parts STEP models are avaliable. You can try to print some of them for a test. That will be cool if you show your attempts to us, especially gears. I have used SLA printing, but didn't try any other technologies. Keep in mind that this parts are not release version, I'm going to make some changes, even not backward compatible.

[VGN]

Thanks to Treehouseman, looks like even forum users could not email me, that's strange.

So, here is my email both for registered and non-registered users. I hope this "light" obfuscation will protect me from spam somehow 
Consider to email me if you would like to be a beta tester, developer or just a backer.
It would be great if you keep mail header like this: "OpenIRV.Developer/Tester" or "OpenIRV.Backer" or etc...

I promis not to publish your email anywhere else and do not send mails, that are not relevant to OpenIRV project.

[bicycleguy]

I just received an e-bay NV3 that seems in near new condition with just a plastic tab on the window holder broken.  Opening the case reveals a ribbon cable from below going up to the window that is not shown in your pictures.  What is that for?  I can probably get a picture, but its pretty tight.  How do I unhook it?

[VGN]

I just received an e-bay NV3 that seems in near new condition with just a plastic tab on the window holder broken.

This window holder locking plastic tab breaks very easily. Mine was broken too)

Opening the case reveals a ribbon cable from below going up to the window that is not shown in your pictures.  What is that for?  I can probably get a picture, but its pretty tight.  How do I unhook it?

This is actually an internal heater with integrated SMD thermistor. Original camera activates this heater in cold weather conditions to melt the ice at the front window, as ice is not IR transparent. There is a FPC connector at the other end of this cable. The heater itself is glued to the plastic housing by double-sided tape. You don't need this part with OpenIRV.

[Treehouseman]

I just received an e-bay NV3 that seems in near new condition with just a plastic tab on the window holder broken. 

So you're the one that bought that camera! I went to go grab it the other day because 5% ebay bucks and it was gone. Ended up going for the Mercedes one instead.

[bicycleguy]

Here's an attempt to print the worm with a regular printer, Prusa Mk3s.  Looks bad but in my experience  it will probably still work.
" alt="" class="bbc_img" />

[ArsenioDev]

Damn you lot snapping up all the good deals on them, back to monitoring the alerts smh

[VGN]

Here's an attempt to print the worm with a regular printer, Prusa Mk3s.  Looks bad but in my experience  it will probably still work.

Nice try!) I counted 31 layer per 5.1mm or looks like your layer height is around 0,15mm. I think you probably can get much better quality with FDM if you make layer height 0,05mm, exact this layer height I used while printing my parts on SLA.

[bicycleguy]

Nice try!) I counted 31 layer per 5.1mm or looks like your layer height is around 0,15mm. I think you probably can get much better quality with FDM if you make layer height 0,05mm, exact this layer height I used while printing my parts on SLA.

Yes on both.  Never tried smaller than 0,15.  I have a 0,3 nozzle to try too.
Here's an uncleaned-up gear for test.


Just curious, is it necessary to use black?

[VGN]

Yes on both.  Never tried smaller than 0,15.  I have a 0,3 nozzle to try to.  Here's an uncleaned-up gear for test.

Yes, this one looks much better! But I'm not sure about that four tips. Anyway, FDM looks promissing.

Just curious, is it necessary to use black?

Do you mean plastic color? What's the difference?


BTW, I haven't still tell about the lens internals. If you unscrew the lens holding ring, you can find a lot of interesting parts. We are going to use only the lens and the lens holding ring (first two positions on photo). Also we need some spring to compensate the thread backlash. This is extreamely needed for proper autofocusing, as backlash causes an error when the motor changes rotation direction. The lens movement distance from end to end is 2mm, and we also have 2mm of space from the bottom of the housing. Generally we need a 4mm height light spring with 2mm working stroke. Unfortunatelly, we cannot reuse original wave spring, as it is too stiff, because it was designed only to lock the holding ring from being unscrewed.

Frankly speaking, that was a great headache, as manucturing a new special spring could be too costy. Manufacturers just hangs up the phone if you are going to purchase less than thousands of spring... But luckily I found a solution - stencils! The stencils, that are usually used to transfer solder paste to a bare PCBs. I decided to design a flat spring, based on this technology. This is going to be reliable and very cheap! Another good feature is that stencils are stainless.

I haven't finished the spring design yet, I should define the number, the width and the length of "legs", the stencil thickness. There are some early examples in attachments. Maybe you noticed a mistake on renders, that legs ends should be bended too.

[bicycleguy]

Just curious, is it necessary to use black?

Do you mean plastic color? What's the difference?
The inside of the gear beveled surface is nearly a lens shade.  On a visible light camera this would all be black.  Is this necessary?  Almost seems like white would emit less IR.  But I'm ignorant of this IR stuff.

I have designed a few stainless steel springs however.  The material is critical.  Not sure about stencil material.  If you can bend it sharply and it doesn't spring back its the wrong kind.

[VGN]

The inside of the gear beveled surface is nearly a lens shade.  On a visible light camera this would all be black.  Is this necessary?  Almost seems like white would emit less IR.  But I'm ignorant of this IR stuff.

The camera FOV is 24° h x 18° v, according to this documentation: http://www.safetyvision.com/sites/safetyvision.com/files/FLIR_PathFindIRII_User_Guide_1.pdf
The angle that this inside of the gear beveled surface forms with main optical axis is wider, than FOV angle of 24° h. In other words, camera do not see any presence of this gear, as it doesn't cross the FOV. That's why I think there is no sense to change the plastic color.

I have designed a few stainless steel springs however.  The material is critical.  Not sure about stencil material.  If you can bend it sharply and it doesn't spring back its the wrong kind.

I believe stencil material and thickness should be stable, as this parameters are very critical for PCB manufacturing. Anyway, this is a matter of experiments.
Looking for good ideas..

[Ruhkukah]

Barev, VGN! Hi all!

I took a stab to print these two challenging focus parts (worm and wheel). Part of the worm broke off after my kid played with it (to speed things up I used 5% infill, so it is pretty fragile)  Now reprinting it
IMO the quality is good enough for the purpose, both parts are very smooth to the touch. For the worm I used black PLA filament, 0.05mm layer height, 0.15mm TriangleLab nozzle and 0.15mm extrusion width, 2 perimeters. For the wheel I tried 0.025mm layer height. My printer is Prusa MK3s stock. I didn't use any supports, so the cross and the circular plate needed a little bit of cleaning.
See attached.

VGN, I applaud your work - it looks very professional and top notch in every aspect. Looking forward to become a tester!

Pavel

[rockwell]

Can we use with this interface the Autoliv NV4 (12µm, 640x480, 50°x39°) sensor in the future ?
Would be cool when these sensor become available and we could use same interface.

[tmbinc]

1. Is the FLIR PathFindIR II _identical_ with the Autoliv NV3? I know it looks pretty much spot on, but I wonder if it's the same thing. Is that an AutoLiv OEM product then? If so, is there also an NV2-based PathFind device?

I would be _really_ curious about firmware dumps of these devices, or any specification that go beyond the two-page summaries.

2. NV4 - I've heard about it, but haven't seen it. Do you have any details? Is the NV business still at Autoliv or now at Veoneer?

[rockwell]

Veoneer is an Autoliv spin-off.
Unfortunately I don't have more information on the NV4 sensor.

[VGN]

Barev, VGN!

I took a stab to print these two challenging focus parts (worm and wheel). Part of the worm broke off after my kid played with it (to speed things up I used 5% infill, so it is pretty fragile)  Now reprinting it
IMO the quality is good enough for the purpose, both parts are very smooth to the touch. For the worm I used black PLA filament, 0.05mm layer height, 0.15mm TriangleLab nozzle and 0.15mm extrusion width, 2 perimeters. For the wheel I tried 0.025mm layer height. My printer is Prusa MK3s stock. I didn't use any supports, so the cross and the circular plate needed a little bit of cleaning.
See attached.

Barev, Ruhkukah!) Thank you for your test! Looks really good. You'd better try to use 0.025mm layer height for worm too. In this case the worm surface will not be so stepped. Looking forward your results)

VGN, I applaud your work - it looks very professional and top notch in every aspect. Looking forward to become a tester!

Thanks much! Don't forget to email me to become a tester. You can find my email in attached images of the first topic post.

1. Is the FLIR PathFindIR II _identical_ with the Autoliv NV3? I know it looks pretty much spot on, but I wonder if it's the same thing. Is that an AutoLiv OEM product then? If so, is there also an NV2-based PathFind device?

I don't exactly know, but I think, that NV3 is identical to FLIR PathFindIR II with some light changes, that each car manufacturer make to fit this camera in their car.

Can we use with this interface the Autoliv NV4 (12µm, 640x480, 50°x39°) sensor in the future ?
Would be cool when these sensor become available and we could use same interface.

I'm trying to do the best to provide as wide support of different IR cores as it is possible: more RAM, higher RAM throughput, more IOs at the sensor connector, more various adjustable power rails at the same connector, etc. In case of best scenario, we will just have to make a cheep adapter board to connect current hardware with a new sensor and develop a HDL module that will interface with it. Also, probably, a new sensor covering enclosure.

[horstmannhid]

Is it maybe an idea to use a ring magnet instead of the spring for the lens thread compensation? I think for 2mm there is always enough force available.

[VGN]

Is it maybe an idea to use a ring magnet instead of the spring for the lens thread compensation? I think for 2mm there is always enough force available.

You are right, this is definitly a good idea! 

Four 1.5mm height magnets generate enough force, pulling a metal ring, mounted on the lens. I glued magnets directly to the housing, but I think some special thin plastic frame, that will hold magnets properly is strongly needed. We will also have to glue the lens to the focus wheel gear. I used a hot glue gun instead of cyanoacrylate for a test.

Advantages:
> no additional friction!!!
> still cheap and reliable solution

Disadvanteges:
> a bit more complex
> two more gluing points (the lens, the magnet frame)

Have something to think about..

[VGN]

I'm glad to say that PDF schematics of the M-board (the most complex one) and B-board (the most trivial one) are avaliable on my repo!
https://github.com/OVGN/OpenIRV/tree/master/docs/schematics

This is alpha version v0.1.0, I probably will make some changes, but without changelog. First stable version will be v1.0.0.

Looking forward your constructive criticism, bug reports, recommendations, ideas, etc.

In attachments you can see how M-board and B-board looks like now. For M-board I decided to replace JTAG connector, FFC/FPC connector is quite thin and can be easily passed between enclosure parts out of the device for debugging purposes.

I hope other schematics will be finished by the end of this week. I'm going to purchase components for new revision and start porting HyperRAM controller to new FPGA this week. I have a special board from another project that utilized Spartan-7 FPGA and a single HyperRAM IC for testing.
That will be fun)

[horstmannhid]

Four 1.5mm height magnets generate enough force, pulling a metal ring, mounted on the lens. I glued magnets directly to the housing, but I think some special thin plastic frame, that will hold magnets properly is strongly needed. We will also have to glue the lens to the focus wheel gear. I used a hot glue gun instead of cyanoacrylate for a test.

That sounds great for this little test. What I meant is using a ring magnet with the right dimensions. They can be bought via ebay. Have a look at the attached picture. If there is a fitting magnet no additional plastic frame for the magnets is needed. Just glue the ring magnet to the housing.


Had a look at the schematic. Debouncing of the buttons will be done in VHDL via a filter in the FPGA? In the microcontroller unit it is an annoying task to debounce something.

[bicycleguy]

In the original design they seem to have gone to great lengths to keep the lens assembly from rotating when focusing.  There are 4 sliding teflon surfaces between the focusing ring and the lens.  Was that to keep wear particles from the threads off the sensor?

Looks like the foam gasket that seals the bottom of the lens to the housing is not thick enough to seal for close focusing if it requires more than about .5mm expansion from the stock setting. 

A popular way to get around lead screw slop on 3D printers is plastic nuts. They apparently last long enough for that extreme task.  Why not replace the threaded focusing ring with a combined focusing ring + ring gear version.  The SLA printer may be up to the task.  The threads don't have to be perfect, the errors control the slop.  And you get rid of a glue joint !  The threaded portion could be backed with a circlip if required. edit, not enough room