Re: Power supplies

[Rubywand <rubywand@swbell.net>]

JCA31@MAINE.MAINE.EDU writes ...
> 
> After trying all the "obvious" checks, it looks like it may indeed be time
> to desolder some caps. Here are the current facts:
> - since that 'globar' resistor is in series with the main AC line, and
>   since the resistance across the line looks the same as in good supplies,
>   it seems a safe assumption that that piece is good.

     Not necessarily. A cracked globar or one with a popped-away lead (which 
still retains some contact with the element) may not exhibit 
discontinuity. It may exhibit a high resistance. Current flow could, then, be 
small enough to avoid further destruction and an obvious break. Measuring R 
across the AC input, especially if you are using a digital meter, might not 
reveal such a condition. (Besides, checking the globar itself is easy. You just 
measure R across its leads.)

> - since the IIe's in question operated just fine when other, known-working
>   supplies were connected, it seems that the motherboard is fine, and the
>   problem must be in the power supply
> - all the caps, when checked with an ohm-meter, looked like caps should look,
>   but these were checked *in circuit* (yes, after being suitably discharged!)
> - like the caps, the main switching transistor measured the way one would
>   expect an in-circuit, functional transistor to look.

     This is kind of a dubious check, especially if you are using a digital 
meter.

     By the way, the reason a digital meter may not be very useful in such 
checking has to do with the low voltages and currents which may be employed. 
These may be below normal semi-conductor "break-over" levels and, so, you may 
detect high R values which are not characteristic of the circuit at normal 
operating voltages. A way to see if a meter is susceptible to this kind of error 
is to measure R both ways across a silicon diode. R should be low (roughly 10-100 
Ohms) one way and very high when the meter leads are reversed. 

     Even if you are using an old analog meter, reliably checking a power 
transistor in-circuit is often difficult unless you know what specific values to 
expect for which leads touching which points.

> 
> I'm not so sure about these last to checks being terribly valid, though,
> with things in-circuit. Since the supply is divided roughly in half by
> the transformers (one large, one small) in the middle of the board, a
> logical thing to do would be to isolate which side of the supply the fault
> is on.

     If, when switched-ON, there is no sound (a thunk, a high-pitched whine, 
clicking, etc..),  you can be pretty sure that the problem is on the input side.


> One way to do this would be to check the activity at certain points,
> say, the transformer taps. This was the sort of arcane technical detail
> i was hoping to find. (Stuff like "18kHz ramp from 0 to 20V from ground to
> tap 4".. or something. The example is pure fiction)
>

     Something like this may be available via a Sams Photo-Fax.

 
> The only way i know that i can get this sort of info is to (*shudder*) open
> up a known-working unit and get the probes ..
>

     Good idea! Using an analog meter, you could do a number of power-OFF R 
measurements -- say, Collector-to-Emitter, Emitter-to-Collector, ... for the 
power transistor, etc.-- and use these to spot problem components on the bad 
supplies. 


Rubywand