Photonic signature:new way to troubleshoot?
mklrobo
Posts: 420
:cool: In my efforts to troubleshoot electronic boards, I have come up with an idea to use different frequencies of light,infrared to ultraviolet, to troubleshoot
suspected electronic boards. I would record these frequencies with the basic stamp, and put in a file. I would power the board, then,
I would put the suspect board in a "dark box", and run the photosensors across the board, from left to right, then decending
down the board, (like a CNC) until the board was totally recorded. Recording a good board first, then a bad board. A defective part that was not
working, or partially working, might show up as no frequency or a different frequency. The stamps would control the motor moving the arm holding the
photosensors, as well as reporting the voltage levels of the photosensors. This technique seems logical, and was one of the tasks that I had in mind for
my HP calc/propeller "tricorder". Comments?
suspected electronic boards. I would record these frequencies with the basic stamp, and put in a file. I would power the board, then,
I would put the suspect board in a "dark box", and run the photosensors across the board, from left to right, then decending
down the board, (like a CNC) until the board was totally recorded. Recording a good board first, then a bad board. A defective part that was not
working, or partially working, might show up as no frequency or a different frequency. The stamps would control the motor moving the arm holding the
photosensors, as well as reporting the voltage levels of the photosensors. This technique seems logical, and was one of the tasks that I had in mind for
my HP calc/propeller "tricorder". Comments?
Comments
Sounds like an interesting experiment.
Basically, you took a working board and set it into a special 'diagnostic mode', then registere the bit pattern on selected pins, and generated a CRC code from it.
Here's the box I learned to 'test' boards with:
http://www.home.agilent.com/en/pd-1000001381%3Aepsg%3Apro-pn-5006A/signature-analyzer?&cc=NO&lc=eng
But my teacher told us that he only taught us this because 'it was in the book' and that it had no real practical use.
(That way, if a manager was stupid enough to ask about it at a job interview, we could say that 'yes we know how to do that'. It was in the 80s... Managers were still allowed to read PR leaflets from instrument makers... )
It is much better to know what kind of signals you're supposed to find on the different pins and having an idea what can cause them to misbehave.
For your idea to have any practical use, you'll need to get a very good resolution, so that you can spot if part of a die embedded inside a plastic encapsulation is hotter/colder than normal.
Just registering that a chip is slightly warmer/colder than normal may not be enough.
Of course, reading off light(IR) isn't completely unknown in the IT business, but then it has been more of a hacking tool...
(And it generally assumes that you're free to shave off the top of the IC package)
Thanks for the info. From what you describe, it looks like it would be accurate to predicte it would work. How precise the reading is, would indicate the efficency of the
device. Since light power disapates at the square of teh distance, the sensors will have to be close to the board. Then, I will have to have a conversion of
lumens per mill/mico watt. This could give an expectation of the correct power consumption, of the working board. Thanks again.)
Awesome advice!
the board with power on, or off, and it would take voltage "signatures" to be compared to suspect defective boards.
I did not know about the hacking tool, Hmmmm, may have to look into it.
I will try it when I get my tripeller, or propcorder working. "tricorder via propeller".
Thanks!
They shaved off the top of a mCU and observed which parts heated up when they ran sample programs, then matched that to the patterns generated by the protected device.
cool!
is unreal what they will do to get data!
If anyone had done it to get to the data, no one would have heard of it.
Listening to the CPU from space is a bit of an exaggeration...
But picking up the sweep from an old CRT monitor from across the road, though...
And from what I understand, some electric 'ball headed' typewriters were very 'noisy' on the grid...
(You could pick up the spin and actuation easily if it was the only one in use)
I believe that a lot of the old 'listen to something from space' stories were fabricated by the CIA or others to stop people from thinking about how they REALLY did it.
You CAN see large troop movements from space... But to ID them?
In the case of the USSR, they sent in agents afterwards, to dig through the latrines...
Toilet paper was always in short supply, so the soldiers used whatever paper they had on hand...
But if you let it slip that you picked up their comms from space, you not only took the pressure off of your agents, you also forced the Red Army to continually enforce stricter and stricter radio procedures, until it becomes almost useless.
Anyway, some components have an audio output also, which it may be interesting to scan for.
Particularly 100Hz and 20KHz are good to look for.
100Hz(or 120Hz) is from rectified AC. (If this drops to 50/60Hz, you have a rectifier malfunction)
20KhZ or thereabout is popular as the switching frequency in a Switch-Mode PSU
This frequency i chosen because it's beyond human hearing.
So it shouldn't be too difficult to pick up a weak 20KHz hum from some components in a computer. But if it's above a certain level, there's a big possibility that there's a fault in the output stage of the PSU, or that it's being overloaded.
The problem may be 'far away' on a PCB, maybe coming through a buffer circuit, or it could be a decoder that causes the incorrect IO port to be read...
You have just as good a chance of finding the problem with a dowsing rod as with a signature analyzer.
Instead of making a 'signature program' for a computer board, people should take the time to make a proper diagnostics program that can output some sort of status.
I disagree. The essence of locating a fault is simply to follow a faulty signal back to locate where it changes from being bad to being good. A signature analyzer may not be a panacea but it can be a powerful tool in a well designed system. The parallel version (logic analyzer) is even better.
No argument here. Having a good diagnostic program can be of tremendous help.
Oh, I agree 100%. Problem is that one cannot take an entire ATE system with them into the field, much as we would like to at times. That leaves us with multimeters, scopes, digital signal analyzers, logic analyzers, and such other equipment that is portable enough to go with us.
Ihad Arduino and Simplepin running well before Mint Crashed. I could unplug a drive to protectt this Ubuntu ,but idnt threre a graphic tool than can be magic to someone like me ?
Ill mis some knowledge wtitten on thise OSes nut i might end up installing everything on the first drive. >>????
'
Well, good luck!
Each bearing in the jet engine was intentionally given a slightly different metallurgical composition, so by analyzing particulates in the oil, you can determine which bearing may be going bad without tearing down the whole engine.
As for the OP's project, it sounds great! Go for it. Good resolution (to see spots on individual chips, as suggested by Gadgetman) is probably a good idea. I'm not sure how useful it will be in practice, but I'm sure _something_ would come out of it - there must be differences to look at. I'm sure it'll be an interesting exercise.