(new pic posted, please help) Cairns Viper, self-heating, some other questions

[Logan]

Thank you Fraser!
I do have another working battery, I ask that question was because I don't want to use the original battery pack.
I actually just found the manual about the same time as you
It seems the "signal" pin on the battery is actually using a protocol to communicate with the camera (instead of some kind of simple analog signal), also the bq2050 is actually monitoring the battery in/out current (instead of just voltage).
So I think there's no way around that, unless I can mess with the host firmware which I can't, I have to live with the annoying huge flashing warning sign if I don't use the original battery pack.
I hate this design 

[Bill W]

All you need is a battery pack with a BQ2050 in it - or maybe just something that responds like a BQ2050 would.

You'd need a scope to see what goes back and forward over the DQ pin.

This looks very like the ancient BQ2010 protocol and the device updated for lithium cells.  As such the datasheet for the BQ2010 (and its' evaluation board) may be better explained and give you a few more ideas.  Not so sure something as 'new' as SMBus would be able to cope.

I'd expect (based on the BQ2010 - as per Argus2 and Argus4) that the camera sends a read of (REGx) and collects the return data to display battery state.
If you think it would help I have the BQ2010 low level software (DOS) to send read commands and display the response, and the EV2010 kit has the circuit to turn DQ into RS232.

Bill

[Fraser]

Bill_W,

Thanks for the correction on the communications port type on this early gas gauge  Another that uses that DOS program with fussy serial port timing ! Thankfully the required interface is simple though I bought the versatile Texas Instruments EV2400 interface for SMBus communications with gas gauges but sadly most I have tried (from more recent battery packs) are protected by a password and require a custom (Manufacturer controlled) version of the software, so battery re-celling becomes a real challenge At least with the bq2010 and bq2050 you have a chance of succeeding

Fraser

[Logan]

Thank you Bill and Fraser.
I know I can replace the cell in the old battery, but I want to be able to power it from a DC supply without warning.
Maybe I should look for the direct output from the core later, but now I would just solve the heating problem first.
 

[Fraser]

You need to spoof the cameras battery monitor into thinking it has a battery fitted, as Bill has suggested. The battery monitoring firmware will expect a response from a bq2050 on it’s data line. To my mind, there are two ways to do this….either build a small microcontroller board that mimics the bq2050 in terms of challenge and response, or a simpler approach would be to rebuild a battery pack to satisfy the cameras firmware that such is present and then provide another power feed to the camera so that the battery is not actually providing the camera power and remains charged. This idea would be simpler but would need some thought.

Basically you need the camera to see a valid data output from what it thinks is a bq2050 gas gauge, even if in the real world it is a spoof or ‘dummy’ containing two small Lithium cells to power just the bq2050.

And remember, your working battery will be on borrowed time now so it is worth looking at rebuild options for that pack as well.

Fraser

[Bill W]

Bill_W,

Thanks for the correction on the communications port type on this early gas gauge  Another that uses that DOS program with fussy serial port timing ! Thankfully the required interface is simple though

The pure DOS program (ap10a.exe) - is a straight boot to DOS and it is purely a command line, and seems OK on any PC with a serial port.

It is the 'nicer' one (EV2010.exe) that we have been trying to get going on old windows PC's.  ev2010.exe would not be any good for Logan and the BQ2050

regards
Bill

[Bill W]

The 2050 had an evaluation kit - and sounds the same as the BQ2010

https://www.ti.com/lit/ug/sluu040/sluu040.pdf

note the warning:
The EV2050 uses the PC-AT real-time clock to provide
the proper bit timing for serial communication with the
bq2050. The modem control lines are used as the
single-wire serial interface to the bq2050. Any TSR that
uses the PC real-time clock affects the operation of the
EV2050. For proper operation, the EV2050 should not
be operated from a DOS shell program


Not been able to find the software though.

Bill

[Logan]

Thank you guys.
I finally got my multimeter and did the test.
I don't have a suitable power supply yet, so I use the one in the Cairns charger, rated 12v 3.3A, with a knob on the board, I set it to 11.6v. It's regulated very well, voltage not changing even 0.01v with variable load. I took the risk of burn the camera, but it didn't , so the camera surprisingly can work 5-11.6v at least.

TEST LOG:
Boot up
0 min: 0.75A/8.7W
10 min: 0.81A/9.4W
16 min: 0.86A/10W, white line begin to appears in picture
21 min: 0.92A/10.7W
30 min: 1.03A/11.9W
35 min: 1.11A/12.9W, NUC can no longer maintain uniformity, large foggy area appears
40 min: 1.16A/13.5W, temperature warning icon appears, current reading stop increasing
50 min: 1.16A/13.5W
Torture Test ended.
(The current value is for the whole camera)

More information: The suspected TEC header is 2-pin, 6.1v regulated(no matter the main power voltage, temperature, or load). I don't know if it supposed to be so. Seems like the controlling part is on the sensor board, or maybe mine is malfunctioning.
Any ideas?

[Bill W]

That is a flat out Peltier.  Should be down to around 5W by no more than 3 minutes.

Pull the Peltier header and see what the current is.

The current rises a bit as the Peltier gets hot, the 6.1V regardless is the Peltier controller power supply maximum voltage.

Ok, so the mystery now is why ......... 

Bill

[Logan]

That is a flat out Peltier.  Should be down to around 5W by no more than 3 minutes.

Pull the Peltier header and see what the current is.

The current rises a bit as the Peltier gets hot, the 6.1V regardless is the Peltier controller power supply maximum voltage.

Ok, so the mystery now is why ......... 

Bill

Without TEC and under same 11.6V, the current is 0.67A,(7.8W), and remain unchanged in the 10-min test.

[Fraser]

This document may be of interest. Note the comment on thermal runaway but I do not think we are seeing that here.

https://www.teamwavelength.com/download/applicationtechnotes/an-tc09.pdf

For the heatsink to be getting hotter and the current drive to the Peltier element to be max’d out you are seeing a failure in correct temperature reporting from the microbolomter, a failure in the TEC or its output Peltier driver H bridge. If the TEC is working and it’s Peltier driver is OK, then the sensor circuit providing the temperature of the microbolometer must be reporting an erroneous higher temperature than the nominal ~30C . The TEC is desperately trying to cool that microbolomter and by the looks of the symptoms the Peltier is managing to get that microbolomter very cold, too cold to operate correctly in fact. You can view the Microbolometer die temperature by looking at it through the microbolometer window with another thermal camera. An IR thermometer will also work if held up against the window. If the temperature sensor circuit fault is suspected, look and test for connectivity failures in the circuit and monitor the TEC behaviour as previously discussed. Sadly you cannot just guess at the fault in a closed loop system due to interactions between the component parts of such.

[Fraser]

I just had a thought, is it physically possible to accidentally reverse the polarity of the wires going to the Peltier element of the microbolometer ? I ask because such a polarity reversal does no actual harm but the following occurs….

Microbolometer temperature sensor ‘reports’ that the die is below ~30C.
The TEC needs to warm the microbolometer a little with the Peltier element so applies a heating drive polarity to it.
Because of the Peltier wiring reversal, the Peltier would actually be driven into cooling mode and cools the microbolomter !
The die temperature sensor keeps telling the TEC that the die temperature is too low so the TEC keeps applying what should be heating via the Peltier element.
The reversed Peltier wiring keeps using the TEC  drive current to cool the microbolometer die and it has no hope of ever reaching +30C but, instead, likely drops to -10C or so ! The microbolometer would be so far out of its ‘comfort zone’ that it would not operate correctly. The poor TEC would continue to command heating but would in fact be causing cooling !

Just a thought and worth checking maybe ?

[Fraser]

In case anyone is wondering how a connectors polarity can end up reversed….some cameras are sent for repair, declared B.E.R., and effectively scrapped. If reassembled as a scrap unit connectors may, or may not be correctly fitted by the tech. I would argue that the camera should leave in the same, or better condition than it arrived in at the service facility but people can get lazy if a unit is declared B.E.R.

These Viper cameras were not known for their reliability and some spent a lot of their lives in service centres being repaired. There appears to be a video related fault on this particular camera so it may have gone in for  that issue but come out with B.E.R notice and an additional fault created during investigation/reassembly. All conjecture on my part thought.

[Logan]

Hi Fraser.
I just watched the sensor rapidly cool down with TEC. We were correct, it's cooling down as soon as power up.
I found another 3-pin temperature sensor glued on the metal frame, it seems change resistance with temperature, but seems affect nothing, even I unplug it, there's no warning.
About the polarity reverse, it seems impossible, because the plug on both side is fool-proof, thus cannot insert reversed.
Thank you.

[Logan]

Err, I just saw your last post.
So that's why it looks so homemade?
Many loose or fragile wires(I already have to rewire a couple), strange patches on cables, some unused wires or pins... Maybe some are just by design...
So should I just invert the TEC header and power on to test? Will it have risk of damage something?

[Bill W]

Err, I just saw your last post.
So that's why it looks so homemade?
Many loose or fragile wires(I already have to rewire a couple), some unused wires or pins... Maybe some are just by design...
So should I just invert the TEC header and power on to test? Will it have risk of damage something?

Only if very careful and as a last resort.  I would find a way to just touch the Peltier on in reverse and see if it stabilises or runs riot again.  The problem is that the 'hot' side gets hotter a lot faster than the cold side as it has pump energy plus dissipation going into it, and it will only be the sensor die to take that power not a heatsink.

Someone did that on a ULIS sensor and it was never the same again.  It did however nearly double the sensitivity !

Bill

[Logan]

It did however nearly double the sensitivity !

Seriously? How?
Okay I'm doing it now.

[Logan]

No, it doesn't work.
The camera don't boot with TEC wire inverted, the 6.1v pins stay at 0.27v, seems like a short circuit.
Maybe the controlling system is indeed on the sensor board, and the 6.1v pins is just power.

[Fraser]

In this situation I would definitely begin reverse engineering of the TEC controller circuit to establish the location of key components and suitable test points. Such reverse engineering has been detailed previously on this forum but basically your need to build a ‘net’ of component interconnections. This is done both visibly and by continuity checks across the PCB. There are tricks to aid the continuity checks such as losing one probe on one ‘node’ and browsing the PCB for continuity with that node using the other probe tip or attaching the other probe to your finger with aluminium foil wrapped around it to create a broad probe head. I use an excellent Wavetek short finder that has a steel brush and a pointed probe for continuity checking. It is fast and effective. Basic reverse engineering is sadly an essential skill when repairing many thermal imaging cameras as there are no schematics available for this technology.

Fraser

[Fraser]

The short finder thread……

https://www.eevblog.com/forum/testgear/wavetek-sf10-short-finder-a-very-useful-tool-if-you-can-find-it/

[Fraser]

Thread detailing reverse engineering of PCB’s, including comments from me on the topic.

https://www.eevblog.com/forum/eda/pcb-reverse-engineering/

Fraser

[Bill W]

Maybe the controlling system is indeed on the sensor board, and the 6.1v pins is just power.

The 6.1V is some power rail (maybe 7V ?) less the saturation of the H bridge driver transistors and maybe a series resistor for current measurement.

Looking closely at the image where Fraser noted them, seems a bit suspect that the bottom two are legible and the top two might look a bit cooked ?
Not necessarily a cause, likely still just an effect.  Scope or DMM all pins on that op amp.

[Bill W]

It did however nearly double the sensitivity !

Seriously? How?
Okay I'm doing it now.

It did indeed, we never knew why but that Ulis 03191 A-Si was known as 'Christine's magic sensor', must see if it ended up in the 'scrap'  .  It was too weird to calibrate normally.
It was run for a good hour or two and was down to a short on the temperature feedback voltage making the sensor seem 'too cold'. 

The heat may have cooked the pixels or die amplifiers or perhaps done a bit of extra gettering.  But it was a stand-out image.

Bill

[Logan]

In this situation I would definitely begin reverse engineering of the TEC controller circuit to establish the location of key components and suitable test points.

Thank you. But I have never done such thing before, I will try to do it later and report back.

[Logan]

Scope or DMM all pins on that op amp.

I don't have a scope... How should I DMM the pins? I guess test the resistance to GND when power off, and voltage to GND when power on?