> > One thing though. No piece of test gear will do more than provide evidence > > of > > what the circuit is doing. It will not tell you 'U38 is faulty' or whatever. > > In principle that was where HP was trying to go with signature > analysis back in the 1970s. The problem has since been solved by
Execpt it didn't really work. Even the HP service manuals which include signatures for various signals have things like 'check signature at pin 5 of U17'. If it's not 1234, replace U17, U1, U18 in order'. It doesn't get you down to a single component. All signature analysis really is is a way of checking if a complex waveform is right or wrong. But if it's wrong it doesn't tell you _how_ it's wrong. Whereas understanding the circuit and knowing what to expect on pin 5 of U17 will let you spot that it's missing a pulse every 5th cycle or whatever, and thus the fault must be... > making the entire product a single FRU (Field Replaceable Unit). :-( > "He fixes radios by THINKING!" > - Someone who had a young Richard Feynman fix his radio, > as quoted in his autobiography _Surely You're Joking, Mr. Feynman! > (Adventures of a Curious Character)_ When I first read that, _my_ first thought was 'Is there any other way to fix radios?' -tony
