Agilent E3634A Power Supply Teardown & Repair

[noodlebear]

Gerry Sweeney just released an interesting teardown and repair of a power supply:

Video is here:
http://youtu.be/a24Fv45FuOc

His blog also has some interesting stuff on building his own power supply: www.gerrysweeney.com

[dimlow]

I found that interesting , thanks for posting.

[free_electron]

Good video but the fault analysis is wrong...
The reason for having that pulsing effect is because the current output opamp was fried. Under normal operation the output of that opamp doesnt come near 5 or -5 volts... So the 4051 will never exhibit this latchup problem.

[noodlebear]

Good video but the fault analysis is wrong...
The reason for having that pulsing effect is because the current output opamp was fried. Under normal operation the output of that opamp doesnt come near 5 or -5 volts... So the 4051 will never exhibit this latchup problem.

Right, but that's pretty much what he said isn't it? That HP didn't need to consider that under normal conditions.

[gerrysweeney]

@noodlebear - yes thats what I said (thank you for making that point), if they did anticipate it, which is highly unlikely IMHO, I expect it would not have bothered them too much - but its worth knowing that the HC405x switches do this under that condition.

@free_electron I did say that under normal conditions its not a problem.  However, the point I was really making is that under any condition where the output of the op amp falls below -5v that the protection added (i.e. the clamp diodes) while they do protect the 4015 from damage it does does drive the chip into an unexpected condition which is clearly outside the spec of the device, the data sheet is pretty clear that the switch I/O's should not be driven above or below VCC/VEE, so the design [mod: of the HP supply in this part of the circuit] from a pure academic point is flawed. Of course clamping diodes are common practice and its really not obvious or intuitive that the chip would do what it actually does under these conditions, and for this design really does play havoc with the rest of the circuit.  Anyway, the real point of the explanation was to alert others to the behaviour of a HC405x switch under those conditions - I bet that the original designers did not foresee that behaviour because if they had I have no doubt they would have designed the problem out - speculation on my part of course....

Gerry 

[linrenzhi]

Thank you very much, by watching video I repaired my E3634A, thanks again!

[Greybeard]

@noodlebear - yes thats what I said (thank you for making that point), if they did anticipate it, which is highly unlikely IMHO, I expect it would not have bothered them too much - but its worth knowing that the HC405x switches do this under that condition.

@free_electron I did say that under normal conditions its not a problem.  However, the point I was really making is that under any condition where the output of the op amp falls below -5v that the protection added (i.e. the clamp diodes) while they do protect the 4015 from damage it does does drive the chip into an unexpected condition which is clearly outside the spec of the device, the data sheet is pretty clear that the switch I/O's should not be driven above or below VCC/VEE, so the design [mod: of the HP supply in this part of the circuit] from a pure academic point is flawed. Of course clamping diodes are common practice and its really not obvious or intuitive that the chip would do what it actually does under these conditions, and for this design really does play havoc with the rest of the circuit.  Anyway, the real point of the explanation was to alert others to the behaviour of a HC405x switch under those conditions - I bet that the original designers did not foresee that behaviour because if they had I have no doubt they would have designed the problem out - speculation on my part of course....

Gerry

The first thing I would do against such a catastrophic burn-out failure would be to add a 7A (6.3/8/10A whatever) fuse in the main current path (e.g. between secondary winding and bridge rectifier).

Is there a way to change the circuit to basically prevent the problem?