FLIR Tau? Wildly varying brightness / gain issue.

[CRC]

I obtained a PTZ security camera that has a regular visual camera on one side and a thermal on the other. I first thought the lens on the thermal was a zoom but it's auto-focus.

When powered up the optical side puts out a decent image. The thermal camera seems to have some sort of issue that is causing the gain to vary wildly and randomly. It will display an image to my monitors but then will darken and lighten rapidly and randomly. It NUCs normally and when I manually focus the lens I can see a very high quality image when the brightness comes up briefly between random intensity changes.

It looks to possibly be a Tau core with a Photon Replicator Board. https://www.thermalvideo.com/flir-tau-photon-replicator-board.htm

I have attached some photos of it. Does anyone have any idea where to start with this?

Video of what it's doing posted at the link below.







https://drive.google.com/file/d/14bFZsKvl2e6GKTXFRMhDwzTWNzVYMSjs/view?usp=sharing

[CRC]

After finally being able to connect to the camera I am able to get it stabilized by doing a reset and saving the configuration. However,  after cycling the power it's back to the same behavior. It's as if the calibration and settings are not remaining in memory. It's there a battery or something that is supposed to retain settings?

[DaJMasta]

No battery, not sure as I haven't taken one apart myself, but probably EEPROM or FLASH.  Have you connected to it with the GUI utility?  You may be able to get some info about the stored calibration (even just a timestamp would tell you if it's being correctly retained), or if there's some flag set that's reverting the intended behavior.  Should just be a serial link, if using the VPC module, just over a USB port, but I think through the Hirose connector it's just a Tx/Rx pin.

[CRC]

I am able to connect to it with the utility.

Here is what happens. Upon powerup the camera image looks like a rave party. It's wildly flashing/varying up and down the gain spectrum. The image itself is decent, it's just the brightness and intensity that are out of control. Calibration and what not appear to be way out of order. Once I do a reset command it settles down and works normally. However, if I save the now reset parameters it does not retain them and does the same thing again the next time it powers up.

Physically, I found that a tiny ribbon cable with a bit of woven tape at the end fell out when the back board was removed. I assume this is a thermistor that contacts the FPGA as it seemed to just lay against it once the back board was in place. I noticed the little plastic retaining clip was broken and when I tried to see if I could get it back in place with some tweezers it launched into orbit somewhere never to be seen again. I placed the ribbon into the connector the best I could and held it in place with a little dab of hot glue.

Could a bad thermistor cause it to lose track of the thermal state its in and act this way? Also, the diagnostics in the GUI show the current temperature to off. It is indicating 41 degrees F when it's actually about 73 F ambient in the room it's in.

Here are some more values it reports in advanced controller GUI mode:
Filter Gain: 0
Filter Control: 0
Spatial Control: 17
Plateau Value: 150
ITT Mean: 127
Max Gain: 8
AGC Filter: 64
Dynamic DDE: 17
ACE Threshold: 0
Information Threshold: 100
Gain Switch Values
Hi-Low Thresh: 140 C
High-Low POP: 20%
How-Hi Thresh: 100 C
Low-High POP: 95%

Spot Meter Temperature -437 F   

The camera shows to be a Tau 2.4 (640x512) Part# 46640019H-FPNLX
It has a very nice motorized large autofocus lens attached which gives a beautiful image once its stabilized (reset) and working normally.

I suppose I could code up an Arduino Nano or similar to fire off a reset command each time it boots but was hoping to find a more legitimate fix.

[CRC]

I was able to get sine comms with it using a USB to 3.3v TTL adapter but now nothing. I guess I'll try a real serial port to TTL adapter again on my win7 PC which seemed to be reliable previously. This thing is really pissing me off.

[CRC]

Does anyone have a firmware file or any ideas of how to totally reset this thing or jtag it or whatever?

[hap2001]

https://www.eevblog.com/forum/thermal-imaging/how-to-upgrade-flir-tau-firmware/

[DaJMasta]

Don't know how helpful it is, but I think the FPA or shutter assembly has an integrated thermistor for self-calibration and FFC, it could be that a broken thermistor is throwing off the camera's reference point and making the readings/AGC settings wonky.

Maybe a couple things to check:

Lock the gain settings/temperature range - does it still happen?

Force an FFC - does it change?

[CRC]

Good points. Once the unit is sent a factory reset command it settles down and works pretty much normally other than radiometry values being way off.i will try to open it back up and try to solder the thermistor to see if that helps.

Even if values are set to save as default upon boot up and they are static not dynamic type settings it always acts up when power is cycled. It seems nothing is ever retained by its memory and any GUI applied settings after boot up are running in RAM and not from flash memory.

[CRC]

Don't know how helpful it is, but I think the FPA or shutter assembly has an integrated thermistor for self-calibration and FFC, it could be that a broken thermistor is throwing off the camera's reference point and making the readings/AGC settings wonky.

Maybe a couple things to check:

Lock the gain settings/temperature range - does it still happen?

Force an FFC - does it change?

Locking gain settings still causes it to happen.
I need to look for a second thermistor but haven't had time yet.
Inspecting the thermistor on the flexible board with the ribbon cable again causes me to have additional questions. When testing it with a DVM I get no reading across the leads (4 lines on a flex but 2 on each side converge to make only two circuits up the thin thermistor flex). The tip is encapsulated in a bit of fabric tape. I can see through it when holding it to a light. I don't see any sort of device at the end of the flex inside the tape. It looks as if the flex is just cut off and taped over. I don't know what a thermistor for these is supposed to look like and maybe Fraser has some images in some of his teardowns, I hope. I was able to seat it back into its connector seemingly fairly well but the camera's behavior did not change. It just seemed to me that the device should have indicated some reading on the DVM and ohms readings should have changed as it was subjected to temperature differences on the test bench.

I recently did another factory reset and was able to dial in a good looking image playing with different settings. However, right now it's indicating a reading of -62C so I can't help but think that the thermistor is bad or missing at the end of the flex or the gain tables are incredibly out of range.

[nidlaX]

Maybe sub in a resistor temporarily that would put the sense temperature in a sane region?

[DaJMasta]

Don't know if I've got the description right, but could the thermistor be configured for four wire measurement?  Could be that the flat flex traces are variable in resistance or just a little high, so four wire measurement is a benefit (though not nearly as simple as just a divider.)

If you see an EEPROM chip in there, you could try to dump the info or just replace it and see if it's just data retention/corruption.

[CRC]

Don't know if I've got the description right, but could the thermistor be configured for four wire measurement?  Could be that the flat flex traces are variable in resistance or just a little high, so four wire measurement is a benefit (though not nearly as simple as just a divider.)

If you see an EEPROM chip in there, you could try to dump the info or just replace it and see if it's just data retention/corruption.

So, like a little Kelvin clip? Hmm hadn't thought of that possibility.
As for the memory I will have to take a look again. The camera came from a large PTZ assembly with optical on the other side. The back of the thermal housing contains a PC board that the Tau plugs into. After studying the board a while it appears it's the character generator to overlay text on the analog image as the camera video goes into the board to an overlay chip then on to a video generator chip. It also has a microcontroller on board that looks to talk to the video overlay chip to, I assume, choose text from the eeprom. The serial port on the microcontroller goes to a MAX232 UART then out to the Tau serial lines. What's bizarre to me is the thing appears to be setup to talk to the Tau using the true RS232 output not TTL. The Tau has what appears to be a Photon replicator board on it but I didn't think that did 232 to TTL conversion.

It made me wonder if the thing was setup to get all configuration commands from the addon board with the microprocessor and video overlay but that seems overly complicated and not necessary but obviously it does talk to the camera in some fashion.