Argus 3 (BST) teardown and repairs

[Bill W]

The Argus 3 was a fire fighting camera made by Marconi and e2v mainly between 2002 and 2006, hence the name change.  The version covered here is the most common 'yellow' version which used the Raytheon BST core and was 320x240 pixels.  The other versions were 'blue' with a Raytheon ASi (160x120) and 'orange' with a Lockheed VOx (also 320x240).

As I have repaired a couple of these recently, I took some photos along the way to give a 'teardown' and provide a repair story.  I will post slowly so there is a chance to reply roughly in the time line.  The teardown / tour also appears on www.fire-tics.co.uk/project2

So the camera here was reported as 'not working' which can cover a whole load of actual conditions from totally dead, through black, white or grey screens to 'noisy but grey' screens.  To most users these are all 'not working' although you can probably guess what each is due to.

So it arrived, and although not from a fire service it still needed a good clean before even holding it.  The casing and box mainly stayed in the shed !  Initially I powered it from  croc clips on the battery terminals from a PSU.  A lot can be learnt from the current draw, and in this case all was well (10V, 3A current limit, took 850mA initially and then settled to 450mA).  The picture though was a rolling mess.
On these cameras the screen runs off the external video so  not surprisingly the external feed was the same.  It was however thermally sensitive, a soldering iron could be seen flying up and down the screen.  On the scope the video was clearly out of sync and variable field times.

So I am thinking something to do with the TV line standard switching.  As that is easier for me to fix internally (and it was still smelly) I took it apart.  In the end I could have corrected it from outside with the customer software, but would not have been able to fix the rest.

So after giving it a bath, on with the pictures....

[Bill W]

The first set are just getting into the case and a few pointers to what is accessible on the sealed case.  The screws are all hex head, 2mm or 2.5mm.

First image - shows how to remove front panel to access BNC video and alsos ee if there is a transmitter
Second image - Connectors and functions around the lens window.  That RS232 (Fischer) connector had a dust cap but everyone loses them and they can just fill up with crud.
Third image - shows how to remove the handles.  The cameras also came with side straps in which case there's two big screws each side as well.
Fourth image - these screws only let you remove the bezel and window, so OK to clean it but no use for disassembly.

I have also added the battery and base contact pinout.

The base power contacts are useful as they bypass the on/off switch circuit, apply power and the camera will start.

The charge contacts need some care, as the charger control is in the camera but the power supply was external, so the supply here is 1A constant current, up to 8V.

Bill

[Bill W]

Now the fun starts for proper.  There are 14 screws holding the case together (arrowed in the second photo).  The camera chassis is attached to the front part of the case, so it is best to remove the back.

[Bill W]

The split case.  The LCD and buttons cable is marked with the green arrow, the chassis is held in with the 6 fixings arrowed in purple.  The coin cell is for the clock IC and also memory backup, so a camera that has lost its' time (reset to 1 Jan 2000) or that will not store pictures can be repaired with a new battery. Given most cameras were built around 2005, the coin cells will now (2021) be getting near end of life.

Second image is running the chassis & front on the bench, with the pins for the buttons shown to 'hot wire' the camera.  Just short the end two pins shown to pretend to be the 'on' switch.

These fixing varied over time, from rubber mounts (which let the chassis bounce around too much) to pillars (fiddly) and eventually spacer posts fixed into the chassis plate.  Electrically there is an FFC connection and a power connection from the front case to the chassis to remove (third image) while lifting the chassis out.

Fiinal image, the removed chassis.  This image has the power input connector shown as well as the 6 way 'export test' connector that I will come back to later.

The 6 way is marked up for the wire colours from the front case:

White    1	Unswitched power in +ve from base pin
Red    2	Battery +ve - to on/off switching
Black    3	0V / case / ground
Yellow    4	Charger +ve from base pin
Purple    5	Battery Thermistor
Blue    6	Battery 3rd pin - for AA pack detection

[Bill W]

As for the repair, I got connection for RS232 and could read out the camera settings.   This function is available on the 'public' software that came on the Argus 3 CD, a copy of which is available here http://www.fire-tics.co.uk/Argus3_CD/index.htm

Among all the usual settings, of note was it reported as 625 line but was not enabled to switch between 525/625.  It thought it was January 2000, which means flat backup/clock battery.
 
Once inside I could look at the core video which was fine, and 625 line standard.  Having persuaded the core to be 525, all was well.  So it was running a mix of 625 on the core and 525 on the rest of the camera.


So how did it get like that?
Maybe after switching over the battery went flat as a lot of things only get saved during the turn off, or the user panicked, so the core changed but the camera did not.  Given the flat coin cell more likely seems that on starting sometime the settings storage for the camera might have been lost and so reverted to defaults (525 & no switching). 

So changed the coin cell and did the full factory setup & tests. After this I could switch back and forth as expected.

A couple of other things had also became apparent.  The spot temperature was reading low (37°C for 54°C) and the lens iris was stuck (as this is common I had checked before putting it all back together).   
So onwards.....

Bill

[MapleCoast]

The BST Raytheon core looks different from the 300D Raytheon BST core, this is the first time I've seen a variation of the design as there is no chopper wheel. Thank you for sharing this.

[Bill W]

There is still a chopper wheel. In the pictures it is mostly hidden by the lens mounting but a little sticks out below.

This is the standard BST 'OEM' kit, but without a detector PCB.  When building a whole system in house Raytheon would make their own adjustments to suit, but equallly were happy if camera builders wanted to make their own detector PCB.

We still bought the 'SECCA' processing PCB, detector, some connectors and the wheel.
By the time we had made our board, sent it to Raytheon for the detector and got it back to the UK there were a fair few air miles on those parts.

Bill

[Lambda]

The BST Raytheon core looks different from the 300D Raytheon BST core, this is the first time I've seen a variation of the design as there is no chopper wheel. Thank you for sharing this.

Yes, as said Bill, there is chopper wheel:

More visible indeed here...
http://www.fire-tics.co.uk/project4/index.htm

And from the picture given by Bill.

[Fraser]

Thanks for your excellent informative posts Bill

I have two A3’s awaiting refurbishment and your posts will be very useful 

Fraser

[Fraser]

Bill,

One feature of the BST technology that some users do not like is the halo effect around a hot target. Did you ever see any work done to remove the Halo effect ? I am thinking a totally opaque chopper wheel might improve the image in terms of Halo, but at the cost of dynamic range. I have made solid chopper wheels but sadly did not have time to test them for halo effect before needing to focus on other commitments. If there is a way to reduce the halo effect, I think the BST technology may be more acceptable to users who are used to relatively ‘flat’ images with no halo. The BST core is a lovely bit of kit that I wish had been developed further, but sadly it was not to be.
The current chopper wheel design works well but maybe there is the opportunity to develop different chopper wheels, as E2V did with their more specialist cameras.

Fraser

[Bill W]

One feature of the BST technology that some users do not like is the halo effect around a hot target.

For the 'fire' market it was more the other way round - they loved the BST 'halo' to the extent that some of the current FLIR enhancements are heading that way.  I did not try a P4428 chopper in a BST (maybe I should .....) but would expect it to be halo free.  Remember that fire cameras all had an iris (aka Thermal Throttle' ) which would mitigate the problem if automatic or else used correctly.

The translucent / diffused chopper wheel acts in a similar way to applying a ~20 pixel unsharp mask to the image.

Bill

[Fraser]

Interesting comment on the P4428 chopper wheel Bill 

I think it would be worth comparing the images produced by the translucent and solid chopper wheels. As we have already seen, it is possible to 3D print a replacement chopper wheel that is opaque so anyone buying a camera or core can decide whether to keep the translucent wheel or change to the solid type to remove the Halo I only mention the Halo as it has been raised with me on a few occasions and some users think that it is a feature of the actual imaging array and they do not like its presence. I can see that in a Fire Fighting situations, a Halo around a casualty would be very useful in highlighting them against the background. I have seen edge detection algorithms that are used to artificially add a 'halo'/edge marker to people in order to highlight their presence in a scene. This is desirable in SAR situations where a target may not be an obvious human casualty.

Fraser

[Bill W]

A couple of other things had also became apparent.  The spot temperature was reading low (37°C for 54°C) and the lens iris was stuck (as this is common I had checked before putting it all back together).   
So onwards.....

The spot temperature was as simple as crud on the lens of the sensor.  A clean up improved this enough (48°C on 54 °C, and 83°C on 100°C) for the owner not to want a new sensor.  The Raytek module only has normal coatings so these had degraded in use.

The stuck iris was a bit more surprising, not the usual oil & grime between the two leaves (oil from the motor bearing) but the slot for the leaves had closed up.  See first image
Maybe had a drop or just movement over time of the machined parts & adhesives
The front element block can be removed, so took that off and slightly filed it to sit straight again and the iris ran fine with the factory speed settings. There was certainly some glue in the face along with the screws.
See second image

Bill

[Bill W]

A few more internal images...


First a front case, fully loaded with spot temperature unit and the US version of transmitter (analogue, 2.4GHz, 200mW, FCC 90).

The chassis connects to the front board by the 1mm ribbon, then all wiring is from that PCB for connectors:
BNC for Video out
Fischer 102 series for RS232
'ambient' temperature yellow wires (a thermistor)

By comparison, the 'no options' basic version with the various blanks annotated



Bill

[Bill W]

PCB's in more detail

1 - Video Processing PCB takes the digital data from the BST core and adds colour, graphics and data to it.  Collects signals from spot sensor, battery voltage and internal temperatures, also controls the display.
Output analogue video (for BNC or rear LCD display)
Output controls to BST core and iris settings

2 - 'Motherboard' PCB for power switching (on/off and 10V supply), battery charge control and interconnections.  Also has a humidity sensor to alert the user to extreme wet - hopefully before the BST sensor is damaged.  It can be a nuisance over time as the humidity rises anyway and can give a false alarm.

3 - SECCA PCB.  The Raytheon processing board which carries the sensor calibration so is paired to the BST detector.  It outputs both analogue and digital video, in the Argus3 though the digital is used.  The SECCA has an 80 way 'Airborne' WBS80 connector, which seems to be obsolete now.
A couple of faults seem common, one is failed tantalum capacitors (mainly on the earlier version) and damage to the inductor '223' by the PCB holder.  Can happen if the camera is dropped and prevents operation with low battery volts as it is the step up part of the main voltage SMPS

4 - Detector PCB.  Although the BST core did offer a Raythen Detector PCB, Raytheon were happy to release their design to allow camera builders to make up a shape to suit their camera and include any other features that they wished.  The row of pins below the detector are where the SECCA is connected via a transition PCB from the 80 way connector.
Various functions are annotated on the PCB and in the image.  The main interest for hobbyists is the ability to use just this as a stand alone camera and access the BST video and control the core direct via the BST RS232 and software that has become public.
The PLD gives a phase shift on the chopper signal (as the sensor is in a different place to the Raytheon default) and operates the lens iris to close if the digital data overloads on too many pixels.

[dennis1573]

Regarding the opaque chopperwheel and "Halo" effect, i did some tests with my Argus 3 BST core.

I made a chopperwheel, using shrinkwrap and graphite paint (Graphit 33 by Kontakt Chemie) for the Spiral.
I designed the Chopperwheel frame in sketchup 3d.
I have attached the .stl file and some pictures, for anyone who is interested.
It has to be printed 2 times, and the shrinkwrap is then "sandwiched" between the 2 halfes.
I put the original chopperwheel on a photocopier, to get an exact copy of the Spiral.
I cut the transparent part out, and used it as a mask to spray the Spiral on.

Here are 2 test videos i made, recorded with my ImmersionRC Powerplay DVR

Original Chopperwheel (With "Halo" Effect):
https://youtu.be/DVV6sw-Xg-E

DIY opaque Chopperwheel (No "Halo" Effect):
https://youtu.be/enszw6B4wGQ

[ArsenioDev]

WOW that's an EXCELLENT demo of the chopper and it's masking. I wonder how hard it would be to print a custom thin monolithic chopper wheel. Prob pretty trivial

[Bill W]

Another option is thin (0.5mm or less) unclad FR4.  It does not need painting.
It is what was used for P4428 and Argus1 choppers.

Those however would benefit from a balance weight, whereas the thin film ones are OK without.

Bill

[dennis1573]

I had some time, and did some more tests.
I redesigned the Chopperwheel, the Spiral is now also printed so this Chopperwheel has to be printed only once.
The Spiral is only 0.3mm high, and so light that balancing isnt needed.
Ive attached some Pictures, and the .stl file for the Chopperwheel V2.
Unfortunately my 3d printers extruder is made of PLA and a bit broken...
So the Chopper i made, looks a bit sad... but it works fine.
I printed it using: PLA, 0.3mm layer height, 100% infill.
I waited for the heated bed, to completely cool down before removing it.
You really dont want to bend it, and potentially scratch your sensor...

Here is another testvideo, this one is a bit longer:

https://youtu.be/Ren3Ua61UEU

I forgot to mention, all videos were made with the Iris stuck in the fully open position.

[ArsenioDev]

WOW that is some VERY nice video from an older BST stack, the loss of the black haloes really does a WORLD of help.

[Fraser]

I agree, for general use the solid chopper wheel improves the image tremendously 

Dennis, thank you for sharing your design. My only concern is the 0.3mm thickness used. Maybe a little thicker would be safer ? As to balancing, if such is required it just needs a small piece of lead tape adding to the wheels centre 

Forum member Spirit designed a 3D printable chopper wheel as well. His design included balancing as I believe his design software provided a weight distribution compensation feature.

I have always found the Raytheon BST core used in Argus fire fighting cameras to produce very nice imagery, albeit with the Halo present. The Cadillac DeVille is less impressive thanks to decisions made regarding the image presented to the user.

I note quite a bit of dynamic noise in the images from your camera Dennis. I do not recall seeing that on my Argus 2 and 3 cameras. It may be worth further investigation but Bill W will be able to advise if such random noise is at the expected levels.

People should not consider BST an old and therefore poorly performing technology. It was actually pretty good  No FFC frame freezes in a time when an FFC flag was deemed essential for microbolomter based cameras  plus, of course, 30 frames per second imaging 

This work has encouraged me to dig out some of my BST cored cameras and get the faulty units running again. I have a fair few !

Fraser

[Bill W]

I note quite a bit of dynamic noise in the images from your camera Dennis. I do not recall seeing that on my Argus 2 and 3 cameras. It may be worth further investigation but Bill W will be able to advise if such random noise is at the expected levels.

I agree it does not look quite right - I wonder if the peltier is running OK ?  Dennis, can you monitor the current consumption for the core as it turns on ?

In ambient temperatures the detector can get lucky and be near 30°C anyway.

Also of note in the video is the sun-safe nature of the BST (thick lump of ceramic) compared to a microbolometers' thin slice of silicon


Bill

[dennis1573]

Hi Fraser, i increased the Spiral height to 0.9mm (so 3 layers at 0.3mm layer height) for you.
I did not test it, so be careful...
I tried it once, with the Spiral at the same height as the Chopperwheel frame (3mm)
but it was completely out of balance...
I have attached the .stl file.

Bill, I think my camera is fine, i use it without the Argus3 control board.
So no Iris control (Iris is fully open), and i have set contrast and brightness to Auto in the GUI.
But i just measured the current, it draws around 550mA at 8V and after a few seconds it drops to around 350mA.
And yes i wanted to highlight the "sun-safe feature", i would never do that with a Vox or A-Si or an old Pevicon camera.
I have also noticed, that the argus3 camera can see the sun through glass. But my A-Si and Vox cameras cant, i thought BST is only LWIR?
here is a video: https://youtu.be/hV0hQQB1D9Q

This might be slightly off topic, but i have a question regarding the Raytheon BST GUI.
I decompiled it, and it looks like there is another password.
It accepts "user" and "tech", but i cant see any difference.
There is a Menu in the GUI, which looks like it might allow me to mark bad pixels.
I see it with the decompiler, but there seems to be no way to get to it...
Maybe someone has an idea? I have a PalmIR250 with 12 bad pixels...
i have attached pictures of the menus, i cant access...

[Bill W]

Bill, I think my camera is fine, i use it without the Argus3 control board.
So no Iris control (Iris is fully open), and i have set contrast and brightness to Auto in the GUI.
But i just measured the current, it draws around 550mA at 8V and after a few seconds it drops to around 350mA.
And yes i wanted to highlight the "sun-safe feature", i would never do that with a Vox or A-Si or an old Pevicon camera.
I have also noticed, that the argus3 camera can see the sun through glass. But my A-Si and Vox cameras cant, i thought BST is only LWIR?
here is a video: https://youtu.be/hV0hQQB1D9Q

That power sounds about right, must just be the result of auto-gain that I am not used to seeing.

As for any MWIR sensititivy, it is not intentional.  The camera is designed for LWIR, however in comparison to an ASi / VOx:

The BST pixel is a simple 'black' absorber of radiation, the structures in ASi/VOx are tuned to the LWIR wavelengths
You have only got 'internal' lens coatings on the sensor and Argus3 lens which are broader than the durable / carbon ones used on an external surface.
May also depend on whether some windows have any extra coatings with IR properties


Bill

[Bill W]

I have also noticed, that the argus3 camera can see the sun through glass. But my A-Si and Vox cameras cant, i thought BST is only LWIR?
here is a video: https://youtu.be/hV0hQQB1D9Q

As for any MWIR sensititivy, it is not intentional.  The camera is designed for LWIR, however in comparison to an ASi / VOx:

The BST pixel is a simple 'black' absorber of radiation, the structures in ASi/VOx are tuned to the LWIR wavelengths
You have only got 'internal' lens coatings on the sensor and Argus3 lens which are broader than the durable / carbon ones used on an external surface.

Just tried this with a ULIS ASi based camera, and it too can see the sun through double glazing.  Highly attenuated of course and the spot temperature reads about 50°C.
That is with germanium optics and 8-14µm internal grade coatings.  Adding an external 'durable' coated window drops the spot temperature of the sun to 28°C, so that alone must be around a 70% loss.
With a hard carbon window, you see nothing.

A point to note is that the ULIS sensors are not filtered against 3-5µm and do have some response (10% or so) there.

Bill