Greg Mayman wrote in a message to Roy J. Tellason:

 -=> Roy J. Tellason said to Greg Mayman
 -=> about "POWER SUPPLY" on 11-05-03  12:06.....

 RJT> I'm thinking that this is a startup problem of some sort.  I keep
 RJT> reading posts about higher-value resistors that go bad,  stuff like
 RJT> that... 

 GM> Yes, it could be a start up problem.

 GM> I went back and read that Silicon Chip article in more detail. It
 GM> is concerned with modifying an AT power supply for a single high
 GM> current +13.5v output, but the description of the operation is
 GM> mainly for an unmodified power supply.

Sounds handy,  for any number of uses.  I'll bet that you could make one
heck of a battery charger,  for one thing.  And it'd run cooler than those
linear setups.

Come to think of it,  I'll bet that dead power supplies would be a good
place to snag some real high-current diodes,  for those chargers I have
that need them!  I'll bet that dead power supplies would be easy enough to
come by,  too.

 GM> First hint: scan the track side of the board and print it out as
 GM> large as possibe -- or photocopy it and enlarge it to the
 GM> maximum. Then draw in and label the components. This will help in
 GM> tracing the circuit.

Not a bad idea,  though it might be a little awkward to actually do it.  I
wonder how my scanner would cope with that idea?  I wouldn't be able to
print it out,  though,  not having a working printer here at the moment.

 GM> Second hint is as follows:

 GM> Pre-test Before Modification
 GM> ============================

 GM> Some more recent PC power supplies derive their control circuit 
 GM> power from the +12v output. This feature allows the control 
 GM> circuit of these supplies to be powered and checked without 
 GM> connection to the 240vAC mains.

Hmm.

 GM> Connect a 33 ohm 5w resistor between the +5v output (red) and 
 GM> ground (black) and a 47 ohm 5w resistor between the +12v output
 GM> (yellow) and the 5v output (red). This will maintain an 
 GM> approximate 5v to 12v ratio between the respective outputs.

I've been wondering what specific loads were recommended for this sort of
thing.  Does this refer to any particular wattage supplies,  or mention AT
vs. ATX supplies?

 GM> A variable DC power supply (8 to 14v range) is connected across the
 GM> +12v output and ground. Check for power at pin 12 of the TL494. It
 GM> should be about 0.6v less than the supplied voltage.

Hmm.

 GM> In the absence of power a jumper needs to be placed between pin 12
 GM> and the +12v line.

 GM> An oscilloscope is used to view the waveforms and operation of the
 GM> TL494 and LM339 as the applied voltage is slowly raised from 8v to
 GM> 14v (no higher than 14v). A 30KHz (or higher) sawtooth waveform
 GM> should be present at pin 5 and square waves should be visible on
 GM> the unbgrounded output pins 8 and 11 (or 9 and 10). 

I don't think there's any 339 chip in this supply I have here.

 GM> [MY COMMENT: The block diagram shows that these pins are connected 
 GM> to the two output transistors: 8 = Q1 collector, 9 = Q1 emitter, 
 GM> 10 is Q2 emitter, and 11 = Q2 collector]

I just wrote down the URL for that chip again,  and need to go get the data sheet...

 GM> These oscillations should stop as the voltage is raised to a level
 GM> equivalent to the design output. The waveform should reappear as 
 GM> the voltage is reduced. If the over-voltage circuit has been 
 GM> activated, the waveforms will not reappear until the circuit is 
 GM> reset by removeing the power.

So this is in that chip?  I wondered how they handled issues like that. 
That and over-current...

 GM> Careful adjustment of the power supply is necessary to demonstrate 
 GM> these two very similar voltage levels.

 GM> If no oscillations are observed, pin 4 of the TL494 [block diagram 
 GM> shows this is associated with "dead time control"] will need to be 
 GM> isolated and connected directly to ground (follow the track from 
 GM> pin 4, desolder and lift one leg of each component connected to 
 GM> this track, the track can then be grounded by a jumper wire). The 
 GM> overvoltage protection circuit will now be inoperative.

 GM> Reconnect the power and a sawtooth waveform should now be visible 
 GM> at pin 5 and square waves at pins 8 and 11 (or 9 and 10). Do not
 GM> exceed 14v in an attempt to demonstrate the over-voltage 
 GM> protection mode -- you have just disabled this circuit!

 GM> .....

 GM> If your PC power supply cannot be tested with an external DC
 GM> supply, you can still modify it but it will be far more difficult
 GM> (and dangerous) to do any initial testing.

 GM> [ Why? It should be possible with a second DC power supply for ] [
 GM> the TL494... but that's just a guess.                        ] 

 GM> Further Reading: Making use of an old PC power supply, Silicon
 GM> Chip, Dec 1998.
 GM> ===================================================================

This article is about modifying one to give you one single output voltage? 
I saw one article on the web someplace where they took one and put a set of
binding posts on the top of the case,  giving you use of all of the outputs
you wanted.  A load resistor was added to keep the supply happy, 
tie-wrapped to the grille inside of the case and cooled by the fan and some
thermal grease.  Not a bad idea,  actually.

I'm also thinking that these boxes might be handy for some other projects. 
If I don't leave the existing stuff in there I can have plenty of room for
other boards to be mounted,  and use all sorts of alternatives for actually
powering the circuitry inside.  And dead power supplies ought to be easy
enough to get a hold of...   Beats going out and *buying* metal boxes for
stuff.

 GM> There is quite a bit more about testing the circuit with the mains
 GM> input disconnected, but the above should get you started. 

 GM> If you want the whole article, I suggest you contact Silicon Chip
 GM> magazine and see if you can get a copy of the October 2003 issue
 GM> with this article.

 GM> Their telephone number from outside Australia is 061-2-9979-5644.
 GM> Normal office hours would be 2300 to 0600 UTC during the present
 GM> daylight saving.

 GM> Price of the magazine is Aus$7.25. I don't know what the postage
 GM> would be. They do take Visa and probably most major cards.

I don't think I'd care to go quite that far...  :-)

---