Fixing Fluke Ti105 - possible upgrade to Ti125?

[mslr]

Hi there,
this is my first post to that forum. Therefor I may introduce myself shortly. I'm doing electronics since 40years and I'm actually responsible for business development in a mid-size company in Germany serving the automotive market and the home appliance market with electronic modules.

I'm doing repair just for fun and private interest without any commercial background. Normally I'm mostly repairing digital mixing devices and power amp's for professional audio which I use myself. Sometimes I'm fixing laptops or phones for my family. Now everybody has an apple laptop and an iPhone I brought back to live....

A thermal camera is a great device for electronic repair. I found a dead Fluke Ti105 and now I'm trying to bring it back to work. Is there anybody out who can support me or has detailed information (service manual, schematics) for that model?

By trying to repair I found that the PCB is the same as of a Ti125 (including microphone for spoken notes). Could be potentially upgraded to Ti125 (of course not for the adjustable optics...). But first of all I have to bring it back to live...

Everything seems to work instead of the thermal sensor. I can only see noise. Optical picture is fine.
Digging into the circuit I found that the power supply providing positive and negative voltage (assuming it is used for the analog circuitry to read out the sensor) is getting hot. It is an LT3471 (Dual DC/DC Converter). The part generating the negative voltage seemed to burn out. Pin6 (Switch output is shorted to ground). Replaced it - same issue after a few minutes. Could not measure a short circuit at the output but I'm still doing some investigations. The negative voltage is used for several components (optical camera, readout thermal sensor, maybe something more).

Has anybody out there access to the service manual or the schematics to support me for the repair? Will make it much more easy. I'm really appreciating any help.


Martin

[Fraser]

Welcome to the Forum Martin.

The schematics for thermal cameras are not made available to the public and the service manuals that are supplied to service agents are normally board level only, no schematics.
The service manuals provide only what is needed to identify which 'module ' requires replacement. In the case of the Ti105, that would be the whole main PCB.

I tend to start fault tracing by downloading the data sheets for the chips used on the PCB and identify all power rails, as you have done. I also map which chips use which power supplies. You appear to have already found a problem in the power supply circuits. A physical low resistance on a power rail should be measurable with power off. If not, you have a component that is drawing excess current. In doing so it may be getting very hot. Sadly this is where another thermal camera would be useful ! Some techs spray IPA onto the PCB and watch where it evaporates quickest as a sort of poor mans thermography technique.

Sadly tracing a component that is drawing excess current can be a challenge on high density PCB's. It could be something as simple,e and small as an ML. Capacitor. Do you know anyone else with access to a thermal camera ? Such would be helpful.

A pretty brutal test method for a failed power rail component is to splice in a current limited power supply set to a high current that is enough to cause noticeable heating of the failed component. This works well but must be done with great care so as not to damage the PCB. Again, a thermal camera hells with this but a IR Thermometer can also work well when scanned across the PCB. You can limit its view with a mask that has a 5mm hole in it.. This ruins accuracy but can show relative temperature change.

If you provide some decent pictures of the PCB and the microbolometer I may be able to provide further comment.

For info, FLUKE thermal cameras are made by the Raytek Thermal Imaging team and the visual thermometers are made by IRISYS.

Fraser

[Fraser]

Further to my last, I always have some mini IR thermometers around the workshop. They are very inexpensive and have a nice small aperture. They are perfect for checking component temperatures on PCB's when hunting an overheating component. Granted I use my thermal cameras these days but a compact IR thermometer is still useful to have in the tool box 

Pictures of units currently on ebay attached.

Expect to pay around 6 Euros or less.

Fraser

[mslr]

Hello Fraser, thanks for the advice.
You describe exactly the path I'm following. Most of the chips are identified. I've an additional thermal camera for my iPhone but I did not have enough time to measure. The linear technology chip for power supply dies in some seconds after switching on, then output voltage is zero and the only heating source is the shorted MosFet in that power supply device.
I'm actually digging the power rails but without schematics it is time consuming to do that with a multilayer fine pitch PCB. But I'm not in a hurry. I've to build a set-up to make all the PCB's accessible in the connected state which is not yet the case. Then I will manually feed the missing voltages with current limiting sources. But first of all I've to identify the needed voltages by checking and identifying the setting resistors.

By the way. To understand how the camera works it is interesting to look into some patents who describe how to get rid of offsets and noise of microbolometers. I was able to indentify some components and the use of them. I'm a Physicist and for me it's really fun to see the thinking of the developers.

Do you have access to the datasheet of the microbolometer? It's naming is BC14573 T65-0341. The bolometer is sitting on a separate PCB which is more or less the analog front end to drive it and to convert the data into the digital domain. It's connected with a flex cable. I can upload some pictures when I'm back home. Up to now I was not able to measure the power rail with connected bolometer PCB due to mechanical limitations. I've to hook up the PCBs to keep them connected.

Thanks again for your support.
 

[Fraser]

Hiya,

Sadly I have no data on that microbolometer number but seeing a picture of its case style may help me identify it as from a particular manufacturer such as ULIS.

Be aware that many cameras will still boot and run without the microbolometer board connected. As you are looking at power supply rails, you can effectively work on the main board without a fitted microbolometer PCB or LCD display.

There are several supply rails in the modern thermal camera. Some are used just for the microbolometer and its specific needs. The Anlaogue and digital supply rails are normall kept separate to avoid digital noise getting into the analogue circuits.

Expect to find the following voltages on the main board of a typical modern thermal camera (not specifically your unit)

5V Analogue supply rail
5V Digital supply rail
3V3 Digital supply rail

Various bias voltage regulators serving the needs of the microbolometer.
Note that it is not uncommon to find more than one DC-DC converter or regulator providing what appears to be the same voltage ! That is to say, you may find 3 digital 5V rails and 2 digital 3V3 rails. There is often a voltage reference that serves the ADC.

In a recent repair I identified all of the Linear Technology power supply IC's and then checked their output voltage looking for issues with them. All were working correctly so they could be discounted from the investigation. You may find a pre-regulator that feeds several DC-DC converters or LDO's. The supply from the battery pack is normall controlled by a power Mosfet and some input filtering/protection is commmonplace.

It should be possible to ascertain the common supply rail that feeds subsequent DC-DC converters and LDO's. It is then a case of mapping the outputs of each DC-DC converter and LDO to see which are connected together in cascade (if any).

You have already identified that a power rail related IC is going into shutdown. It should be possible to trace where it's output pin goes on the PCB by carrying out continuity checks to major IC's and known supply rail points on the PCB. If the overheating component feeds other LDO's etc, are they getting hot ? A failed DC-DC converter or LDO that takes its power from the overheating chip could be your problem.

To see a schematic of a modern microbolometer based thermal camera, take a look at my posting in this Forum on the ULIS 0205 microbolometer sensor array. I provided a schematic for a camera and that shows the various power supplies that are needed.

https://www.eevblog.com/forum/thermal-imaging/the-ulis-02-05-01-microbolometer-an-insight-into-teh-technology-by-fraser/

Correction, I think I removed thevschematic diagram from that thread. I will look for it and see if I can upload it again. The thread does provide an insight into the required bias voltages etc though.

Best Wishes

Fraser

[mslr]

Hi Fraser, thanks again.
I see a LT3471 failing which probably generates a symmetrical supply voltage for the analog domain. It is a dual switching supply and the negative voltage path burns out. Unfortunately it does not shut down and going in some protection mode. The negative switching output always shortens to ground, the positive rail keeps alive.

Everything on the camera is working fine except the thermal image which shows only noise. The camera overlays an optical image with an thermal image. The normal picture is fine.

Your're pointing to other posts. Very informative. I've seen similar structures in patents and on my web search.

My explanation is the following: The reason I can only see noise is the missing negative voltage rail and then the analog path has a non symmetric supply unable to handle signals near to zero voltage. After some time it happens that the positive voltage rail also shuts down because of the overheated device. Then the noise disappears and I can only see a homogen colour - clearly the complete analog supply for the bolometer is tight to zero and the readout is zero which is then calibrated to room temperature. The structure you can then only see in the image is linked to the stored calibration data.

I've attached some pictures of the microbolometer. In one picture you can see the failing device on the bigger PCB - it is the one in between the four coils just behind my magnifier. Hopefully the upload works.

[Fraser]

Have you tried to run the main board without the micro bolometer board fitted. If the main board will boot you can determine whether the fault lies on the main board or microbolometer board. Just try to switch the camera on and monitor the failing rail for behaviour.

The microbolometer looks like an older design with its round window. Similar to tha format used in easily Indigo cameras. I will see if a I can identify it.

If you can trace the failing supply line to a significant chip that it powers you may find the fault along the route. Negative supplyrails are not very common in my experience so it may feed just one chip ? All tantalum capacitors on that supply rail should be checked as such are a common point of failure as I expect you know.

If I were approaching this fault I would look to find a point where I can isolate the negative rail in order to check it without its load. Such is not always easy though. I would then substitute a current limited power supply set to the voltage that the DC-DC converter is set for. Your little phone thermal camera can then be used to monitor the PCB as current is increased.

I have a tendency to always reverse engineer a section of PCB that I am investigating. It is time consuming but I suspect your errant negative rail has a very limited 'customer base' on the PCB.

Fraser

[mslr]

Hello Fraser, thanks for the comments,
I do it in a similar way when I do a repair and I do not have the schematics. First of all I try to identify the main function blocks by identifying the components. Normally it is not that difficult to link them to certain strange behavior if you have a principle understanding of the device. This led me to investigate the analog signal path of the microbolometer because anything else worked fine. The hot DC/DC converter was easy to find. Reverse engineering of the converter gave me the assumption that the converter may do a negative voltage. Which matches perfect to my assumption of a failure in the analog signal path. Couldn't measure any short up to now in output because of limited accessability when everything is mounted and connected. I can only say the main PCB is fine. I'll do that measurement next weekend. As a shortcut try I just replaced the converter chip (the output switching MosFet is mostly a weak point) but without success.
Up to now I only know that this power rail is used in the optical camera, the bolometer PCB and in the main PCB. Since in some application notes this part of the LT3471 is often used to drive power LEDs in a similar configuration and the connector to the optical camera uses 3 pins for this rail I assume that this rail is a power rail and serves also the lightning LEDs of the optical camera and is not only serving the negative voltage of the analog path. But this is an assumption and not yet verified.
My biggest fear is that the microbolometer needs a negative voltage and is the reason for the short circuit. It is a big and heavy device an I have seen some broken screw holes of the outer housing which is often a sign of an impact. Hopefully this impact did not affect the bolometer. Therefore the datasheet of the bolometer will really help.
I'll keep you updated.

[mslr]

Just some more findings: Unable to find any shorts in the power rail I digged deeper in the circuit. By reengineering the DC/DC converter I found that the feedback loop of that power rail is broken. The voltage divider which is setting the voltage is not connected to the feedback pin of the device. May be a defective trace in the PCB. That missing link will lead to a continious 'on' of the switching MosFet shorting the input inductor to ground. Obviously the Mosfet will not survive that. I'll replace the chip and give him the correct feedback tomorrow. Lets see if that brings everything back to live.



Gesendet von iPad mit Tapatalk

[Fraser]

mslr,

Excellent work. This is looking hopeful 

Fraser

[Chanc3]

Fingers crossed this end!


Sent from my iPad using Tapatalk

[mslr]

Fingers crossed this end!


Sent from my iPad using Tapatalk

One step ahead. Repaired the power rail and the feedback. But for some reason the converter is now shut down. But fortunately it is not getting hot, reference voltage of the chip is ok. The thermal picture is just some colours but no more noise (and no thermal picture...).

Now I have an other very strange behavior. The clock of the camera is running very fast. On minute lasts some 10seconds. That was not the case before. Had to stop the work for the next days due to other tasks. I'll keep you updated.
Martin


Gesendet von iPad mit Tapatalk

[Fraser]

Have you any history for the camera ? Multiple faults can occur but are not that common on thermal cameras. Such usually occur when an overvoltage is occurs at the charger input socket. An RTC running so fast is very unusual. Drop damage can also result in multiple faults. Corrosion due to fluid ingress can caused many failures with very unusual symptoms.

The shut down switching regulator is worthy of further investigation around its pins. Many such chips have a chip enable pin. If it is being deliberately shut down by the microprocessor it may be a response to another fault and so a safety fallback on that power rail is activated.

Can you identify the cause of the failed feedback line I. The DC DC regulator? A broken track may be the cause but what broke the track, Corrosion, physical drop damage ? Or just bad luck ?

it would be worth checking the PCB to see if there is a dedicated Real Time Clock IC as that should be pretty much independent of other circuits and running at thirty two KHz.

Fraser

[mslr]

Have you any history for the camera ? Multiple faults can occur but are not that common on thermal cameras. Such usually occur when an overvoltage is occurs at the charger input socket. An RTC running so fast is very unusual. Drop damage can also result in multiple faults. Corrosion due to fluid ingress can caused many failures with very unusual symptoms.

The shut down switching regulator is worthy of further investigation around its pins. Many such chips have a chip enable pin. If it is being deliberately shut down by the microprocessor it may be a response to another fault and so a safety fallback on that power rail is activated.

Can you identify the cause of the failed feedback line I. The DC DC regulator? A broken track may be the cause but what broke the track, Corrosion, physical drop damage ? Or just bad luck ?

it would be worth checking the PCB to see if there is a dedicated Real Time Clock IC as that should be pretty much independent of other circuits and running at thirty two KHz.

Fraser

Hello Fraser,
of course the inspection in a microscope for any visible damage was my first step, even before analyzing the schematics. But nothing  was visible. Even in the microscope at high magnification. The feedback trace is completely buried in a middle layer. The vias to the traces are part of the solder pad and covered by tin. No sign of humidity but, as I mentioned, broken screw holes as a sign for a drop.
Yes the chip has enable pins but I had no more time to do some measurements. I only measured the reference (1Volt) and the supply and feedback loop. Everything was ok. Since it is a fine pitch package it is not just easy to measure without the risk of making a short leading to a blow out of the chip.
One thought:
A wrong RTC device running too fast is very unusual. It looks like the camera part is behaving like a hight frame rate camera instead of 9Hz but the software is not aware of it. Could be a path to tweak the frame rate? Today I do not have the time to go on. Maybe Wednesday or Friday.
Question: Has anybody some experience how the frame rate is reduced. It may help me to look to the right part of the PCB.



Gesendet von iPad mit Tapatalk

[Fraser]

Frame rate........

1. In your camera the frame rate of the sensor cam be anything it can cope with, even 120fps.

2. The frame rate of the LCD / video output can be equal to the sensors frame rate or it cam run at a slower/faster frame rate. That is just a software function.

3. The frame rate has to be limited to less tha 9fps for worldwide distribution without the complexities of dual use technology paperwork. As such the OEM limits the frame rate in the image processing stages of the camera. This is done in such a way as to make it extremely difficult for anyone to change.
The FPGA code is normally responsible for this frame rate control.

If someone has tried hacking the FPGA code you will likely have a bricked camera.

Fraser

[mslr]

Still searching for a time slot to do deeper investigations. Up to now no additional progress.

@Fraser: Did you find any spec/datasheet for the microbolometer? Maybe you have access to a camera schematics using the same or similar sensor?

[Fraser]

Sorry, no to both of your questions I regret to say.

Best Wishes

Fraser