Author Topic: Raytheon/Cadillac DeVille camera 3D-printable focusing mod and chopper wheel  (Read 6077 times)

0 Members and 1 Guest are viewing this topic.

Offline Spirit532Topic starter

  • Frequent Contributor
  • **
  • Posts: 487
  • Country: by
    • My website
I recently got my hands on one of the Cadillac DeVille/Raytheon/L3 cameras with the 320x240 BST sensor, and I found two things to be extremely inconvenient:
  • The almost completely fixed-focus, and nearly telephoto lens was focused close to infinity, which made is almost useless in an indoors setting.
  • The chopper wheel on my unit was slightly warped from previous use and replacements are both expensive and unavailable. (the former likely the result of the latter ;D)

To solve these issues, I designed a set of three easily 3D-printable components that require almost no post-processing(more on that later) that I'd like to share with everyone here.

This is how the manual focus assembly looks like:


The replacement lid, transparent in this image, has a large thread cut into it, and allows you to focus anywhere from beyond macro to beyond infinity(though not recommended as focusing beyond infinity also crushes the chopper wheel).
The lens holder, metallic, is designed to accommodate the lens block, which is what you get after you remove everything possible.

A quick guide on removing the lens block from the original front casing:
First, you have to remove the ring that sits on the very top of the lens housing. It has a small notch on the side that is easy to place a prying tool in. Be careful when removing it.
Next, you have to remove the protective window and heater element, which is done by first shearing off the head of the soft black plastic rivet on the side - this is the only destructive operation in this mod. Once the pin is sheared off, the window holder and heater block can be removed by rotating it counterclockwise and gently pulling. Rotating it may be very difficult if the protective window was broken previously and has let dust inside the cavity, so don't be afraid to put a bit of force into the twist.
After the window is removed, turn the top casing around and remove the metal ring holding the focusing mechanism in place, this is done by unscrewing the three self-tapping screws.

Now, this next part is tricky. There is a small gap between the very top of the outer shell and the focusing mechanism, right next to the alignment slot(into which the case fits). You have to place a small flat-blade screwdriver into that slot and slowly pry around the gap, until it's about 5mm wide. This may require a lot of force, so be careful to not slip into the Ge front element.
At this point, you should be able to pull the whole focusing assembly out of the front case.

When the assembly is out, place a hex wrench or appropriately-sized screwdriver into the small brass gear on the back of the focusing assembly, and turn the gear until the back of the lens assembly(the bit with the plastic cover, which is also very fragile) is all the way in, if looking from the back. Once it's in, firmly grasp the lens assembly from the front, and unscrew it.
Note: There is a very tight and angry but short spring holding the lens assembly in place.

At this point you should have just the lens block itself, with all the unnecessary components removed, and it will fit into the lens holder as shown here:

The bottom of the lens assembly should be flush with the bottom of the lens holder.
It's also a good idea to stuff the original o-ring between the lens holder and the lens block.

The replacement lid was not designed to conform to the curve of the device, and therefore is not as rugged, however it does fit the original bolts, and it should sit flush with the casing. A bit of electrical tape around the metal edge of the case is a good idea.

This is how the final assembly looks:

This also shows that the print quality tolerances are very large - the print came out half-broken and it still functions.
The design incorporates a ~0.4mm gap between the case and lens holder, so it should thread in easily.


And here's a video of it in action, focusing on objects closer than infinity(hand, lighter, cracked window in autumn, space heater, cold metal):



Now, a bit about the chopper wheel:


The design may look a little odd, but this was done for a reason - it follows the original Fibonacci-ish exponent-ish curve of the original almost perfectly, and it's balanced to (ideally) within tens of microns. The balance part depends on what you make the wheel out of and how good your printer is, but it's also possible to create it out of almost any sheet material on a CNC router/mill(though the pinch may be a bit difficult to mill out).
You will also have to wrap a bit of tinfoil around one of the legs to make a reflective surface for the closed loop speed sensor.

Although the wheel is identical to the original in terms of shape, when installed onto the motor it results in rather poor performance, with a very large non-uniformity, sometimes even image inversion, and awful dynamic range.
I'm not entirely sure what's causing this, and more research is needed, though my suspicion is that the sharp transition between solid plastic and air makes for a poor image, and the original approach of diffusion must be used instead of complete interruption.
The next step would be attempting to replicate the wheel's film print by printing the pattern on a regular transparency sheet using a laser printer and dithering.

This is the result of using the printed chopper wheel, same lens setting as the previous video:




As evident, it needs more work, but I've got other projects on hold that need work, so I'm publishing it as-is, in hopes that someone else may further improve the design.

Update: The unacceptable image quality is caused by a phase shift in the readout of the sensor in relation to the chopper wheel position. This can be fixed by tweaking the position of the tinfoil reflector - try to get it to be as close as possible to the location of the reflector on the original wheel.

Here is a video of the printed chopper in operation:


Note that the phase is still slightly off - this is my mistake, and if you place the reflector in the perfect position, you will get a very clear, crisp image with a better dynamic range than the original chopper.


The files can be downloaded here: https://spirit.re/projects/files/download.php?file=/raytheon/raytheon_stl.zip
Note that the chopper wheel must be printed at 200% size in X and Y, 100% in Z.


If you'd like the original design files, made in Autodesk Inventor 2018, send me a PM.
Any ideas and suggestions welcome.
« Last Edit: June 26, 2018, 06:45:44 pm by Spirit532 »
 
The following users thanked this post: Fraser, ArsenioDev

Offline Vipitis

  • Frequent Contributor
  • **
  • Posts: 867
  • Country: de
  • aspiring thermal photography enthusiast
Looking at the picture.... You might want to set more top layers in your slicer... At least that it what it looks like.

Do 4 or 5 instead of 2.

And regrading the video, it's seems very very noisy, is there any processing to be done to reduce the noise?
 

Offline Spirit532Topic starter

  • Frequent Contributor
  • **
  • Posts: 487
  • Country: by
    • My website
Looking at the picture.... You might want to set more top layers in your slicer... At least that it what it looks like.

Do 4 or 5 instead of 2.

And regrading the video, it's seems very very noisy, is there any processing to be done to reduce the noise?

The prints suffered serious under-extrusion due to bad filament. It was 3 top/bottom layers.

The videos are almost unprocessed - all I did is to boost contrast to the max, otherwise the image looks bland.
Though the noise mostly comes from the ancient camera - the 320x240 BST sensor only has a rated sensitivity of 100mK, and both it and the chopper wheel degrade over time, plus the board is packed with tantalums and some electrolytics that I've yet to replace.
My camera is around 14 years old and has a damaged front lens element, so it's normal.

Still on the hunt for something >320x240 with radiometric output, but no offers and no eBay deals :(
 

Offline Spirit532Topic starter

  • Frequent Contributor
  • **
  • Posts: 487
  • Country: by
    • My website
Have you confirmed that the chopper wheel timing is correct ? You say you have used foil to trigger the position sensor. Is the position of that foil relative to the spray start the same as the original wheel ?

I've tweaked the timing a little more and am delighted to report that I've been able to produce an image BETTER than the almost-perfect(but warped) original chopper wheel:


The piece of aluminium foil has to be placed at the exact same spot as the original, and has to be the same size.
The vertical "gradient" seen in the video indicates that the reflective piece is shifted slightly to one side - it is not, in fact, the shape of the spiral, as I initially suspected.

In conclusion: the 3D-printable chopper wheel is a permanent and easy replacement for the almost unobtainable original, and may even improve image quality.

I've edited the original post to indicate this success.

Regarding noise, these analogue board Raytheon cameras are more noisy than the later Digital board versions like the 300D. The camera should still produce a decent image though. The main cause of image degradation in my experience is the degraded chopper wheel.
This model(the Gen 2), is fully digital as far as I can see - the image is sent out of the sensor straight to an ADC, processed by the FPGA, and pushed out into the DAC to be converted to composite.

I have asked my friend for comment on what might be wrong with using the 3D printed opaque, as opposed to translucent, chopper wheel.
It appears that there's nothing wrong, and the wheel can be printed in any kind of plastic. I believe a fully opaque one should work best, as it "drives" the pixels the hardest(ambient of the wheel to target temperature), instead of diffusing it. The camera seems to show a very decent dynamic range, and I believe the original semi-opaque dithered spiral is designed to decrease dynamic range, to make more objects visible on the road.

On the topic of 320 x 240 pixel Radiometric cameras. Sadly all of my 'spare' 320 x 240 pixel cameras are high frame rate industrial models that cost a small fortune, and are not something I would be keen to ship across international borders. Sorry.

160 x 120 pixel cameras and cores are the most commonly found at reasonable prices at the moment. Some are 9fps, others are higher frame rates and only some have documented data outputs.

A decent 320 x 240 pixel radiometric camera is still quite rare and expensive, even on the secondary market. The USA is the best hunting ground for such capable cameras, but sellers of this technology will often refuse to ship it outside of the USA.

Fraser
I've been searching fruitlessly for months now - the eBay sellers seem to completely vanish the moment I mention where I live(Belarus), even though neither the US nor EU have any restrictions on exporting here - the words "ITAR" and "paperwork", and even "Wassenaar"(to which the US has no relation!) seem to terrify them out of existence. Some sellers even go as far as to say "This item is ITAR, I cannot and will not export it"(you've likely stumbled upon this seller), which is just false.
The procedure to export a high frame rate camera out of a country governed by either ITAR or Wassenaar is simple, and only requires a small amount of paperwork, and that is paperwork I'd be happy to fill out for any prospective seller, but alas, none of them want to cooperate.

The thermal sensor used in the Cadillac DeVille camera is BST technology. BST sensors only output a change in their pixels received energy levels. That is to say, the pixels must be reset to a baseline level in order to detect and output the thermal scene data.

If anyone is curious as to how the Barium Strontium Titanate sensors actually work, more technical info can be had here:
https://en.wikipedia.org/wiki/Pyroelectricity
https://en.wikipedia.org/wiki/Pyroelectric_crystal
 

Offline Fraser

  • Super Contributor
  • ***
  • Posts: 13145
  • Country: gb
I have removed my posts from this thread and advised the OP why. Sorry but such was necessary.

https://www.gov.uk/guidance/arms-embargo-on-belarus

Whilst nothing I have said in those posts is truly sensitive, it is risky to appear to provide 'technical support' for such technology to someone located in Belarus when I am a UK national. The post removal is not done lightly.

It just goes to show how tricky discussion of thermal imaging technology can be.

Very Sorry  :(

Fraser
« Last Edit: November 21, 2017, 01:34:08 pm by Fraser »
If I have helped you please consider a donation : https://gofund.me/c86b0a2c
 

Offline Fraser

  • Super Contributor
  • ***
  • Posts: 13145
  • Country: gb
I have printed the Chopper wheel that the OP kindly designed. I am still experimenting with my 3D printer so thought this was an interesting object to print.

The pictures are of the first wheel that I elected to print in the vertical plane. I have another printing at the moment that is in the horizontal plane. Printing in different planes presents advantages and disadvantages. We shall see which looks best for this application. Remember a perfect edge to the spiral is not needed as even the original has a dithered edge due to the semi opaque dot pattern. I would tidy it up a bit using fine sand paper if it was going to be used though.

Print details:

Printer : UP! BOX
Extruder nozzle : 0.4mm
Layer thickness : 0.2mm
Fill : Full (maximum available)
Filament : ABS (UP! Genuine 260C type)
Print time : 100 minutes (Vertical), 64 minutes (horizontal).

Pictures show the print with and without its support scaffold to show how it is printed as well as the end result.

Fraser
« Last Edit: November 21, 2017, 09:25:31 pm by Fraser »
If I have helped you please consider a donation : https://gofund.me/c86b0a2c
 

Offline Spirit532Topic starter

  • Frequent Contributor
  • **
  • Posts: 487
  • Country: by
    • My website
That print looks pretty nice! It looks like the mounting hole is a little too big though. I designed it to have no tolerance on the hole so as to allow a more precision way of expanding the hole(i.e. a drill bit or reamer).
100% infill is required, but the part is pretty low in volume so it doesn't make much of a difference to print time.
It should be printed horizontally to make sure the hole plane is parallel with the wheel. I've experienced minor warping when printing and subsequently drilling out the hole - it can be fixed by heating up the middle section of the wheel with a lighter and aligning it on the shaft.

The width of the spars was also chosen to allow you to wrap the tinfoil piece completely around it and not worry about the width. However, you do have to worry about at which radius it was placed - it has to be start and end at the radii of the original reflective disk.
 

Offline Fraser

  • Super Contributor
  • ***
  • Posts: 13145
  • Country: gb
Here is the chopper wheel that was printed in the horizontal plane.

It is a better print  :-+

Fraser
« Last Edit: November 22, 2017, 12:57:21 pm by Fraser »
If I have helped you please consider a donation : https://gofund.me/c86b0a2c
 

Offline Fraser

  • Super Contributor
  • ***
  • Posts: 13145
  • Country: gb
I just printed another chopper wheel and I am pleased to say that the result is repeatable and very easy to print. This time I managed to separate the wheel from the raft by directly 'attacking' the joint with the separation tool, rather than peeling the raft away.

This is just another example of how useful a 3D printer can be. I had not considered printing a chopper wheel before ! My thanks to Spirit532 for the effort he put into the creating the 3D files  :-+

I was going to print the new lens holder and front plate but it will take 11 hours to print the front plate with 100% fill  :o

Looking at the new lens mount design, I do wonder whether the thread helix gradient is a little too steep in terms of lens movement. From experience the focus distance adjustment on most thermal camera lens blocks is very small. The thread is quite a shallow gradient helix to provide fine movement.

Fraser
If I have helped you please consider a donation : https://gofund.me/c86b0a2c
 

Offline Spirit532Topic starter

  • Frequent Contributor
  • **
  • Posts: 487
  • Country: by
    • My website
I was going to print the new lens holder and front plate but it will take 11 hours to print the front plate with 100% fill  :o
Fraser

The helix is not too steep, and you don't have to print it with 100% infill.
I printed mine with 10% and it worked, though around 20-30% should be better.
 

Offline Fraser

  • Super Contributor
  • ***
  • Posts: 13145
  • Country: gb
Thanks Spirit532.

I am still learning about 3D printing and what infill is useable for different objects  :)

Fraser
If I have helped you please consider a donation : https://gofund.me/c86b0a2c
 

Offline Vipitis

  • Frequent Contributor
  • **
  • Posts: 867
  • Country: de
  • aspiring thermal photography enthusiast
Hey Fraser,

I would like to recommend the r/3dprinting subreddit or even better this Discord: https://discord.gg/r7hdHP for all the advice and help you will ever need.

Infill is needed to support the top layer of a print. Infill is also needed to make printed parts stronger for mechanical application.
For figures are display pieces infill of 15% or less is all you need.
More infill means longer printing times and more filament which both costs more money.
100% infill is only needed for very small parts or some special case mechanical applications like Devin Motes' Springo.

Most parts use less then 30%. There is also different infill patterns.
 

Offline Fraser

  • Super Contributor
  • ***
  • Posts: 13145
  • Country: gb
@Vipitis,

Many thanks for the help  :-+

I was very unsure how much infill was required for decent structural strength so 'played safe' on initial prints.

The official 'UP! Fila' filament is not cheap so any reduction in the amount used is welcomed. Fortunately Maplin Electronics UK has just stopped selling the UP! printers so they sold off all their 500g reel twin pack ABS 260C UP! filament  :) I bought a total of 7kg at £18/kg. That was half normal price. I already have two new UP! Fila 1kg black ABS reels from my stock for the UP! Mini printers that I also own  :) That lot should last me a while  ;D

I also looked into your advice on CD software. Sadly AutoCad is no longer free for home users. It is free for those in educational institutions only and you have to apply using such an institutions email account. Obviously this is an attempt to convince students that AutoCad is the best software to use when they leave University and start working for companies. That will be payback day fro Autodesk. Private, non commercial  individuals like me have to take out a monthly subscription  :(  I just do not do enough CAD to justify it.

Fraser
If I have helped you please consider a donation : https://gofund.me/c86b0a2c
 

Offline Spirit532Topic starter

  • Frequent Contributor
  • **
  • Posts: 487
  • Country: by
    • My website
Autodesk Inventor is free for students and (somewhat) free for home/educational use.
Fuision 360 is completely free for-profit unless you make <$100k/year IIRC.

AutoCAD is getting slowly outdated with parametric design and digital manufacturing getting used more and more.

Theese parts were designed in Inventor.
« Last Edit: November 22, 2017, 09:04:49 pm by Spirit532 »
 

Offline Vipitis

  • Frequent Contributor
  • **
  • Posts: 867
  • Country: de
  • aspiring thermal photography enthusiast
Yeah,

Autodesk offers education accounts. I am luck enough to have a 3 year licence to the big package through just signing up and mentioning my school. Sadly Autodesk has move their photogrammetry software from 123D to Memento to ReMake and now ReCap and it's only cloud based. It was the only ly Programm that suppported thermal images for photogrammetry with radiometric data. And doing it manual in VisualSFM is something I still struggle with since I am not great with imagemagick yet.

Back to the topic. Fusion360 is all you need and more. You got CAD workspace for modeling and patching, even sheet metal if you're into that.

You can sign up as a hobbyist and use it for non enthusiast and startup work up to 100k$/year(if you make that much you can afford a subscription and the 5% royalty).

This type of license needs to be renewed every year but cost nothing and maybe takes a day or two to get approved. It's also available for other Autodesk software. You might have to get a license for some add-ons if you plan to do physics simulation or renders.

There is a lot to learn in 3D printing but most comes from just doing it and gathering own experience. For example ty no infill, 5 or 6 top layers and 4 shells. Changing you bridging to rather fast and also turn off fan for bridges. You would be surprised how sturdy some parts can get without infill. The durability of parts has different factors. Infill is just one. Orientation, material, temperature, print quality and perimeter settings are also factors that go into durability.

ABS is rather strong, but also a challenge to print and new ASA or PC composites are combining the strengths of ABS with the simplicity of printing PLA.

Next to Discord or Reddit a good source of 3D printing knowledge is YouTube. There are really great YouTubers that provide advice and reviews the top channels every 3D printing enthusiast should now are.

Joel - 3D Printing Nerd
Angus - Maker's Muse
Tom - Thomas Sanladerer
Devin - Make Anything

You should be able to find them by a simple search, I can't put links right now.
 

Offline Fraser

  • Super Contributor
  • ***
  • Posts: 13145
  • Country: gb
Thanks Vipitis  :-+

I am somewhat spoilt by my new 3D printer. It is an 'as new' Tiertime UP! BOX that forum member 'Toploser' very kindly sold to me recently.

The Tiertime printers use a 'in-house' driver software and slicer. They are a very well formed 3D printing solution that is very user friendly, whilst providing excellent results. My UP! BOX prints ABS beautifully thanks to the nice warm enclosure and filament that is matched to the machine. I can use 3rd party filaments as the extruder and bed temperature may be set as needed. I will experiment with other makes of filament when I am used to the printers normal quality.

The UP! BOX is really designed for Office and Developer use and it is normally an expensive beast, but thankfully I got a great deal from Toploser. The software that drives the printer is very well tuned to the printers needs so it does not have masses of system fine tuning options. They are not needed in this printer. All the usual 3D printing options are present including layer heights as small as 0.1mm.  The bed is auto leveling and the nozzle height calibration is also automatic. So far the printer has been faultless and just prints whatever you ask it to. Perfect for me as I do not have the time to continually tweak settings to achieve a good print.

I got lucky  :) I could never have afforded the UP! Box when it was almost £1500  :o

I also own a pair of UP! Mini printers but neither has had much use really. The UP! BOX is so user friendly and reliable that I have been printing many items that I had waiting for the UP! Mini's on my long to-do list.

Another 'win' I have just had is my finding a French 3D Printer supplier selling the upgraded UP! BOX+ CPU module kit that I can fit in my printer to provide Wi-Fi connectivity, Filament monitoring and auto print recovery after a power fail. It also uses an upgraded microprocessor that is better protected against memory corruption by static electricity. It was priced at 99 Euros instead of the usual 400 Euros ! I suspect a pricing error but they dispatched it to me yesterday without comment.

I purchase a 3rd party CPU daughter board for one of my UP! Mini printers. It converts the printer into a standard Smoothie based printer compatible with G-Code. The same HK based seller also provided me with a couple of excellent bed heater PID's and driver boards to convert the two Mini's to a variable temperature heated bed, rather than fixed at 60C with a mechanical thermo-switch.

All in all, my experience of 3D printing has been very good when using the Tiertime UP! printers. I still have much to learn however.

Fraser
« Last Edit: November 22, 2017, 11:59:42 pm by Fraser »
If I have helped you please consider a donation : https://gofund.me/c86b0a2c
 

Offline AKMac

  • Newbie
  • Posts: 6
  • Country: us
Hey guys, I just found this post, but I also had a chopper wheel posted up! Haven't had a chance to print it yet, but what do you all think of this?

https://myhub.autodesk360.com/ue29b9825/g/shares/SH7f1edQT22b515c761e4cbb2447926bcf77
 

Offline Spirit532Topic starter

  • Frequent Contributor
  • **
  • Posts: 487
  • Country: by
    • My website
what do you all think of this?

Two things:

1) The shape of your spiral looks a little off in the beginning and end - I got caught by that during my first attempts.
The original wheel doesn't follow the log exactly - it tapers off, which had quite an effect on my image.

2) Your wheel is unbalanced and will vibrate.
Check out the centre of gravity/mass feature in Fusion, it'll help eliminate the issue.

Other than that, it looks pretty good.
Did you design it from a picture reference, or just by eye and measurements?
 

Offline Fraser

  • Super Contributor
  • ***
  • Posts: 13145
  • Country: gb
AKMac,

What camera is your chopper wheel designed for ? It has the optical timing notch so I am suspecting it is not for the Cadillac DeVille Raytheon camera.

For those unaware, the chopper wheel commonly found in the fire fighting cameras and Thermal Eye 300 series is different to that used in the Cadillac DeVille camera. The Cadillac cameras chopper wheel diameter and spiral form is different to the standard Raytheon BST kit wheel.

Fraser
« Last Edit: December 18, 2017, 11:32:08 pm by Fraser »
If I have helped you please consider a donation : https://gofund.me/c86b0a2c
 

Offline AKMac

  • Newbie
  • Posts: 6
  • Country: us
As far as shape and the optical timing notch, this is an exact copy of the chopper wheel that came with the camera. I was not aware the Cadillac version didn't have one. To clarify, this is for the Cadillac camera. As for the shape being off, I'll have my designer revisit the original image again and retrace it.
 

Offline Spirit532Topic starter

  • Frequent Contributor
  • **
  • Posts: 487
  • Country: by
    • My website
I'm not entirely sure why you would want to replicate the part I already designed, since you specified that it is for the Cadillac camera.
 

Offline AKMac

  • Newbie
  • Posts: 6
  • Country: us
Spirit, I had said I had that designed before I saw this post. If your design works better than the one I had drawn up, then great! I just wanted to show what I had done up as well. So, why would my chopper wheel have the optical timing notch if the Cadillac cameras didn't use it?
 

Offline Spirit532Topic starter

  • Frequent Contributor
  • **
  • Posts: 487
  • Country: by
    • My website
Which version of the Cadillac camera is it? Does it look like mine, with a single board, or is it the older, stacked arrangement?

I believe the older, first revision had a different chopper, but the spiral should be the same for both. The timing on my unit is done via a reflective piece of foil on the bottom of the wheel and an optical reflection detector, the notch indicates it's an interrupter around the wheel(not under).

Try printing both designs, yours and mine, see how they compare.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf