Thermal Camera Teardown - The FLIR TAU 320 by Fraser

[Fraser]

I recently purchase a faulty FLIR HS324 thermal camera/scope in the hope of restoring it to full operation. I traced the failure to the TAU core which is a bit of a nightmare on the repair front. The Tau core in the HS324 is a TAU320  that produces a 320 x 240 image. This particular implementation use a standard TAU with some additional modules attached to its rear. I will not be covering those modules in this post as they are being included in my HS324 thread.

The good news is that a failure in the TAU means that you will get to see inside this camera core. In order to investigate the fault I have used another TAU 320 as a working reference core from which I can the readings. This thread will include pictures of PCB's from both cameras and a comparison of them as they are differing revisions.

Pictures follow. I am providing the normal VGA resolution images plus some slightly higher resolution images of the more intricate parts. There will be plenty of pictures for those interested in the TAU and I will also be commenting on my fault finding progress.

On with the show.....

Fraser

[Fraser]

The interior views begin.......

[Fraser]

More......

[Fraser]

More.......

[Fraser]

More....

[Fraser]

More.......

[Fraser]

More.....

Good PCB on Left, Faulty PCB on right

[Fraser]

The Microbolometer.

Note that it is a high quality encapsulation and superior to that found in such cameras as the Ex series.

Fraser

[Fraser]

More......

[Fraser]

Fault tracing ......

I hot wired the TAU interface PCB that is used in a FLIR PS32 in order to power the TAU cores for testing. The TAU needs no other support in order to produce an image. The indication of normal operation is the twin FFC event just after power is applied. Failure to produce these two events indicates a critical error or fault in the core. My faulty core does not produce that twin FFC event and no video is output.

Access to the PCB test points in the TAU is made more difficult by the close proximity of the adapter PCB that I am using. I have been using PCB track repair wire to extend the test points to a more accessible position.

The components on the TAU are small in order to keep the unit compact. 0402 discrete passives are used all over the small PCB. Good magnification is needed to work on the PCB.

Comparisons with my working TAU PCB revealed that the faulty TAU PCB has a problem with its 12V supply rail which is reading only 9.3V. The 12V rail is generated from the main 5V supply using an LM27313 Boost regulator. testing revealed that the potential divider in the Boost converter feedback path is in trouble. One of the resistors was way off value and so has been removed. I will fit a replacement resistor and its associated parallel 220pF capacitor tomorrow. This may, or may not, be the only fault on the unit. We shall see. I am getting too old to be working with 0402 components !

In the attached picture there are some bare pads in front of a larger brown 4.7uF SMT capacitor. These are where the removed components once sat.
 
Fraser

[Chanc3]

Great teardown as always - saving me from my curiosity as always! I don't think my boss would appreciate me taking out Tau apart!

Looking forward to the repair results!

[Kilrah]

Nice, beautiful construction!
And I'm wishing for more 1600px photos, the 640px postage stamps are always a bit deceiving!

[Fraser]

I will certainly include more higher resolution pictures in the future. The 1Mb file limit does constrain image resolution and I can only upload posts every 60 Seconds so two images per minute.

Fraser

[Fraser]

It should also be noted that the TAU is not a FLIR design. It is an Indigo product that came to FLIR when they purchased Indigo.

With modern integration there is not much to be seen inside the TAU camera.  A small double sided PCB is all that is needed. Impressive. The LEPTON is another impressive example of thermal camera miniaturisation. Miniaturisation of thermal cameras will always be hampered by the physical requirements of the microbolometer and it's lens though.

Fraser

[Kilrah]

Interesting, is that the reason for your 2 PCBs being different, i.e one being the original Indigo design, and the other a Flir-designed minor rev?

Instantly looks even more modern/advanced knowing that the design must then date from before 2003...

[Fraser]

Indigo were specialists in thermal camera miniaturisation. That is why FLIR bought them and their expertise.

Both my Tau cameras are from the FLIR owned era, but there are different revisions of this platform. It's predecessor was the Photon. A similar camera but a little less integrated.

Fraser

[Fraser]

I am investigating the 'faulty' 12V rail on the failed Tau PCB. I want to be sure that this revision of board is supposed to have 12V on that rail.

Time to reverse engineer that part of the PCB.To help me, I put the PCB in the high resolution X-Ray machine. As you can see, this is a reasonably high component density board with several layers. The X-Ray image gives me the information I need in the area around the 12V boost converter though. I include the X-Ray images for your interest.

Fraser

[Fraser]

Using the physical magnification available in the MX-20 I closed in on the 12V Boost Regulator area. The MX-20 also provides a zoom feature an further image enhancement but this is good enough for my needs.

Fraser

[Fraser]

For comparison, a Raytheon PCB from an EEV Argus camera that basically does the same job as the TAU pcb. There are two versions, Analogue, and Digital, so I have included both types fro interest.

Fraser

[Fraser]

A couple of PCB dimension reference images to show how compact the Tau Core PCB is. The tiny 0402 passives are barely visible to my 48 year old eyes so I use a microscope to work on this sort of kit.

Fraser

[Fraser]

For those interested in the chips used on the Tau PCB, here they are:

KLB = BAT54BRW Schottky Barrier Diode array
6210 = BD6210  DC brushed motor driver for FFC flag drive
EP3C25U256 Cyclone III  FPGA - The 'brains' of teh TAU
SRPB = LM27313  Boost converter for 12V rail
25P64 = M25P64  64Mbit serial Flash SPI
PDII = TPS3828-33   Voltage monitor & supervisor IC
BYJ = TPS62240 300mA step-down converter
BQE = TPS62400 400mA step-down converter
ADV7127 10 Bit 240MHz D to A converter
D9JRN = MT46H8M32LFB5-6   256Mb SDRAM  2Mb x 32 x 4

Fraser

[Fraser]

The 12V rail on the working TAU Core is making me scratch my head.

The chipset on the TAU PCB requires the following voltages +1V2 , +1V8, +2V5, +3V3 and +5V0

There is no obvious requirement for +12V that I can find on the PCB. Even the 'RS232' data link is 3V3 only.

The 5V supply rail is up converted to +12V and this is fed the Microbolometer PCB. On that PCB, the +12V feeds a LT1761 (marked LTGC) LDO +5V regulator. That regulator is intended to provide a well regulated low noise supply of +5V. No great mystery there, but why boost to +12V ?

If that is the only purpose of the '12V ' supply, then the +9V3 supply on the faulty unit would be more than enough to meet the needs of the LT1761.
Hmmm more investigation needed.

Fraser

[Kilrah]

On that PCB, the +12V feeds a LT1761 (marked LTGC) LDO +5V regulator.

Are you sure it doesn't also go straight to the sensor?

If not I'd check the output of that LDO, there could be a problem downstream of it that loads 5V too much, in turn loading the 12V too much... in which case the problem would not be the 12V rail itself, but whatever draws too much.

[Fraser]

@Kilrah,

Thank you for the comments.

I tested the voltage of the power supplies with the Microbolometer disconnected so thankfully the microbolometer is not loading the '12V rail'. I have yet to find any components using the '12V rail' on the PCB and can find no low impedances on that rail. It also worth mentioning that i monitored the current draw from my PSU when powering the good and faulty PCB's. Both drew 134mA. this suggests no excessive loading on any rail WRT to a known good board. There are some minor differences between the two PCB's including the FFC Flag motor driver IC running off of the 3V6 rail on the faulty PCB and the 5V rail on the good PCB. I am wondering if the 9V3 reading  found was indeed legitimate. No harm dome as I have not changed the rail to 12V yet   

I was tired when I was looking at the supply voltage lines and now I look at the Boost converter again, I doubt that the potential divider was faulty as a resistor normal cracks, going O/C or high in value. That would wildly change the output voltage, not just change it by 3V. There is always a risk when using a different revision of PCB as a voltage reference unless you are certain that the supply rails are unchanged between revisions.

Fraser
 

[Fraser]

When working on a faulty PCB without schematics, I always draw up a simple block diagram showing the interconnection of the various modules, PCB's and major components. This provides an easy topology reference and diagram on which to annotate test results.

The block diagrams are usually pretty rough and always hand drawn. I attach my present diagram for the TAU320 for others reference. Please note that this is a sketch and is not supposed to conform to any schematic or block diagram standards, it is just my shorthand when repairing kit 

As you can see, the camera PCB is actually pretty 'simple' in terms of the number of IC's used. The FPGA is the brains of the outfit and enables such integration.

The Video DAC produces the composite video signal for direct feed to the Hirose I/O connector with no buffering.


UPDATE: The Block diagram has been updated and corrected. You will find the latest version here:

https://www.eevblog.com/forum/thermal-imaging/thermal-camera-teardown-the-flir-tau-320-by-fraser/msg1004215/#msg1004215

I am removing the old version from here as it could cause confusion.

Fraser