Look for faults on cards

[elettronica52]

Does anyone know which thermo camera I have to buy to look for smd components that are broken based on temperature?
I try to repair an instrument board but when it is inserted it is not possible to take measurements due to lack of space for the oscilloscope probe. What resolution should the thermo camera have?

[Ultrapurple]

If there isn't enough space to squeeze in an oscilloscope probe then it sounds unlikely that you will get even a small thermal imaging camera anywhere near the board.

TICs work best as PCB diagnostic tools when they can have a clear look at the board from a distance of several inches, used when the DUT is powered up. Sure, you may sometimes need a closer look from nearer but most of the time there are advantages to seeing as much of the board as possible.

Resolution doesn't have to be high. Even 160x120 is good enough to spot hot bits on a PCB, though I have found the next step up (320x240 or 384x288) are a better compromise between cost and resolution. Once you've identified 'there's a 'hot bit over there' you can move the camera and on the better models re-focus (or add a close-up lens) to see more details in a smaller field of view.

Ultimately it all depends on what you are trying to look at. A photo or two of a typical thing you're trying to do fault finding on wouldn't be a bad thing to post.

[elettronica52]

I try to repair a measuring instrument from Rohde & Schwarz that has this broken card: it does not generate the frequency. It is mounted vertically and there is about 10 cm of space with the other card. With the probe it is impossible to track the signal: the tracks are small and there are no test points, only mmcx connectors but without references.
I thought that a thermal imaging camera might see hot spots where to narrow down the search such as a lossy capacitor or a warmer component.
Otherwise an extension of the 96 way futurebus connector must be made! but it could alter digital signals. As a compact camera which would be ideal for these applications with an acceptable cost? (Flir or Seek)

[DaJMasta]

Look through the instrument's menus to see if the test points are accessible from there, I've worked on some other R&S gear which will give you a readout from various places, but it doesn't list an optimal value.  You can often get some information on what's working by looking at what output there is (what harmonics, power level, etc) with a spectrum analyzer, and can pick up something by tracing over the signal with an EMI probe or similar.


While you may get some information from a thermal camera, my guess is the fault won't be so obvious with that alone, and unless you make an extender or have a couple slots next to it free, it will not be easy to monitor with a thermal camera anyways.  I'd start by looking for rails on the chips with datasheets you can find, since it will give you their power requirements and pinouts.  Also helps to monitor the biasing network for amplifiers in the signal path - or even measuring current through it with a multimeter/scope on each side of a resistor in it, for example, you can probably get an idea of what's good or bad by comparing like parts on the board.

Also helps to know where the main signal path is, so looking at a block diagram and plotting it out visually on the board can be helpful in showing you where to be probing next.  For these R&S board in particular, they tend to keep the RF stuff on one side, so if there are parts on the back, they're generally power/control/monitoring (like ALC) parts that go to parts directly on the opposite side of the board.  The little silver parts (like on the very bottom of the card) are power supply line filters, so they can be good test points for rail voltages.  Also the MMCX (I thought they were different connectors, but maybe they are that) test points need to be "engaged" by moving a passive into place - usually a capacitor or something - and they are good for testing purposes, but using them is sort of in-depth - usually you have to disable the normal path to measure the point.

For me, a thermal camera probably wouldn't be a tool to look towards at this point, though I'd probably consider it anyways just in case if the view of the board was alright when installed in the chassis.  Don't know what an "acceptable cost" would be in your case, but you don't need particularly high resolution, sensitivity, or special features to find faults on a board - just have to scan around for irregularities and have a reasonably close minimum focal length.