Are most chip problems catastrophic... ie- open or shorted? If so, then can't I troubleshoot typical problems with just a logic probe to test for highs, lows, and pulsing? Since the internal resistance of TTL chips is low, can't I just remove power and test the input-to-output pins? How about hooking up a 3volt battery in series with a buzzer, then negative to ground/positive to input, and listen for the buzzer (ie- current flowing)... or visa-versa? Surely I don't have to understand all the AND/NAND/OR/NOR functions in the practical world of troubleshooting... ie- it's either working or it's not? Request for Question Clarification by markoft-ga on 15 Jun 2002 12:54 PDT Are you looking to troubleshoot individual chips or fixing a large scale electronic device such as PCs, VCRs, or a video game console? Answer There is no answer at this time. Comments Subject: Re: simple, practicle testing of microchips From: quesera-ga on 15 Jun 2002 01:55 PDT To properly answer your question, a little more context is needed. Still, this might help. At the silicon level, yes, most IC failures are opens or shorts. Failure analysis consists of removing part of the black casing and inspecting the circuit under a microscope. But that's rare. It's only done when you're verifying that the fab manufactured the chip in a way that corresponds to the layout files you sent them. So it depends what kind of chip you're talking about. The above is used for microprocessors, for example. Smaller chips that are well-understood and stable designs are tested differently. There are IC testers that will take (for example) a 14-pin TTL logic 74xx series chip, test the ins and the outs, and tell you a) what kind of chip it is (the xx part above) and b) whether all gates are working properly. You *can* do this with a logic probe, and in fact the chip tester works the same way. But you do need to know the inputs and the expected output to know whether you have a good chip. Transistors are frequently tested with an ohmmeter. A single transistor's most common failure mode is shorted. So 0 ohms across the transistor (usually test the outermost two pins) is a bad transistor. A logic gate is made up of several transistors, so although 0 ohms is still an indication of a bad gate, it's a much less common failure mode (several transistors would have to have failed identically for this to occur), so it's usually not a valuable test. Lastly, you won't be able to drive a buzzer with a TTL chip. The current required to make the buzzer buzz is way above the chips driving abilities. At best it wouldn't work (with certain clamping designs), at worst it would toast the transistors inside the chip. A small transistor amplifier could do it, but it wouldn't work any better than a logic probe. (some logic probes have piezoelectric transducers in them, switchable to chirp on logic high or low.) You'd still need to know exactly what you were testing. Hope this helps. Subject: Re: simple, practicle testing of microchips From: jeanluis-ga on 15 Jun 2002 06:03 PDT You should also keep in mind that some errors could be logic errors in the logic programming, i.e. 1+1!=2. Usually these types of errors are not that trivial, and can only really be found with actual use of the chip (i.e. you don't find the problem until you sit down and start programming it) Here is a real life example, at my last company we developed a DSP, and when we got back 1st silicon we found a few problems (like the ones talked about above) before we really started using it, and so we fixed those problems, then produced new chips. When we got 2nd silicon we started to really use it for complex programs. When we did this we found that if an interrupt occurs in the second cycle of a multiply complex (mpycx) the results of the mpycx were lost! Something like that would be very hard to find without actually sitting down and using the chip... Subject: Re: simple, practicle testing of microchips From: pcnetworkctr-ga on 15 Jun 2002 19:30 PDT After reviewing the two most recent replies, one must also consider problems that occur such as propagation delay and "edge" rise and fall times. It would be a little more helpful if there were some specifics to your question. However, to answer one portion of your question; No, logic chips can fail for other reasons than a simple open(s) or short(s). That is to say, that your IC is configured as a trigger or as a simple oscillator/timing circuit, timing errors caused by propagation issues within the devices internal capacitances can occur.... Exactly what type of circuit are you attempting to troubleshoot? Subject: Re: simple, practicle testing of microchips From: sgtcory-ga on 16 Jun 2002 20:03 PDT Hello ignoranttech, I agree with most everything that other users have shared. Most chip (assuming you are talking about microprocessors, flip-flop chips and the like) failures are an open or short case, however like pcnetworkctr stated - there can be numerous other causes, all dependent on the circuit and chips in question. Without knowing the quiescent state of the chip in circuit, trying to troubleshoot it would be a waste of your time unless you could get your hands on something like a Huntron trakker. ( http://www.hammondelec.com/tr200.html ) Essentially this injects a signal, and you read the output of the signal after it has been processed through the IC. You could take a known good chip, and do pin to pin tests, and compare that to the same readings from the suspect IC. There are older industry practices that aren't in regular use anymore, but may prove helpful since you stated you didn't want to get into binary logic. Here is what you *can* try, assuming an everyday cheap IC: 1) Buy some of the same chips you think may be defective. 2) Piggyback. This means place the new chip on top of the old chip while it is still in circuit. Place it on top of the chip in the same pin order...i.e pin 1 on 1, pin 2 on 2 ( Be sure to turn off power though :-) This does many things. If you have an internal open circuit situation, the path of least resistance is through the 'working chip', as a complete open is infinite resistance. The signal ,(logic level - waveform etc..) then is in a completed circuit and *should* work as normal. I say should, because you could be dealing with an IC that has more problems than just the open. If this solves your problem, then you can simply replace the IC in question. This will not work with a short, because the path of least resistance will be the short itself, and the signal in question will once again take this path. This is an old technique, and I found myself using it on rare occassions years ago. I would not use this in a daily routine however. There is definitely the possibility of destroying perfectly good chips if the problem is elsewhere in the circuit. Hope it helps - sgtcory