From: A.R. Duell <>
Date: Mon Jun 23 21:26:40 1997

> > The 5V regulator seems to be a switching one, which is what I'd have
> > expected.
> Huh? That would make sense because the linear regulator in T220 case
> is limited to 1.5amp which it requires measly 3.6amp on that 5v
> lines.

In the PC-jr, I think the 5V regulator was a step-down switching one using
that large inductor on the PSU card as the energy-storage device. The 12V
supply is (AFAIK) a linear one. I am working from memory - my PC-jr
Techref is about 70 miles away,

> > Yes, is that suprising? After all, a switching supply is nothing magical.
> > You still need to transform the AC output into DC and regulate it. Of
> > course you can always perform some regulation on an SMPS by varying the
> > drive waveforms on the primary side.
> Hmmm...I fix PC powers and modern SMPS for the computer all the time
> to save $. Around here in canada costs above $60cdn for a new
> 200watt PSU.

Yep, so do I. Most SMPS faults are pretty simple (at least if the
transformer has not got shorted turns...). And it's certainly worth doing
for the more obscure supplies (i.e. not generic PC ones...)

> >
> > Talking of bad design (I was earlier...) I've seen some _crazy_ SMPS's in
> > my time. In a Zenith monitor there's a supply that combines the
> > reliability of a switcher with the efficiency of a linear. It rectifies
> > the mains, feeds it to a free-running chopper (no regulation applied
> > here), shoves that into a little transformer, rectifies the output and
> > feeds it (about 18V DC) into a linear regulator that uses the power-on LED
> > as the reference. It's the only monitor that _requires_ a green power-on
> > LED. When my blew up (chopper failed, etc), I replaced the entire mess
> > with a little torroidal transformer. It was cheaper than a new chopper
> > transistor and a lot simpler.
> >
> Ewwww.... that is bit overkill to design that! But closed loop

Do you meant the redesing of using the torroidal transformer (at 50Hz)? It
was trivial - jumper over the chopper transfomer, rip out the chopper
board, use the mains rectifier as the LV one and feed 20V AC up the
'mains' cable. A lot less work than rebuilding the SMPS in my case
> switchers is very efficient and compact but that transformer is
> certainly larger than the orignals.

Actually, the 50Hz 30VA transformer was no larger than the original

> Are you talking about this beige tiny 12" green or amber monitor with

That's the one.

> angular back? In this, chopper trans pops and kills one resistor.

Well, I had shorted turns in the chopper transformer (tested on a homebrew
'ring tester'). It didn't seem worth rewinding.

> They are usually either TTL or composite.
> I have a working zenith Eazy PC with hd in it. Slow and buggy. :)

I still have mine on a much-hacked PC/AT

> > Then there are the Boschert 2-stage supplies that are used in Sanders
> > printers and PERQ 1's (and probably elsewhere). A shorted chopper in one
> > of those (which is a very common failure mode) will blow up 2 more
> > expensive power transistors and then take out a number of small
> > transistors, the chopper control IC (a good old 723), a few passives and a
> > couple of PCB tracks. I had to sort out such a mess once - I can provide
> > the full story if anyone's interested.
> Hmmm... you could email me privately with this account of yours. :)

It may as well go here - it's just about on-topic....

The basic circuit design is :

1) Rectify/double the AC mains input to give 340V DC
2) Feed that into a non-isolated 'flyback' converter to step it down to
3) Feed the output of that into a free-running push-pull oscillator using
a couple of big power transistors.
4) That oscillator drives the primary of the main transformer
5) The outputs of that transformer are rectified and smoothed. The 5V
output is fed back via a regulator/optoisolator to the oscillator that
drives the 'flyback' chopper mentioned in (2).
6) Oh, there's a crowbar triggered from the 5V output that shorts the 24V
output to ground in the event of overvoltage. There's also a current-sense
resistor in the return line from the oscillator (3) that shuts down the
drive to (2) if it overcurrents.

Now, here's what goes wrong.

1) The chopper (2) shorts. The output of that stage now leaps to 340V
(since there's a DC path through the shorted transistor and the inductor)
2) The oscillator (3) continues running, so the 5V line leaps to about 12V
3) The crowbar trips, shorting the output to ground
4) The overcurrent trip tries to work, but as the chopper (2) is shorted,
it doesn't do a darn thing.
5) The 2 oscillator transistors short as well. The total load on the 340V
line is now the 3 shorted transistors, a few low-resistance inductors, and
the 0.12 Ohm sense resistor. The latter explodes, feeding 340V into the
sense circuitry, which dies expensively, often taking some PCB tracks with
6) The fuse fails...

> By the way, where I could buy this real techref and how much?
> Soooo, I could fix up it with addons to XT standards.

I got mine from IBM about 2 months ago. I'll try and find the part
number/form number to order it if you like.

> Jason D.

The gates in my computer are AND,OR and NOT, not Bill
Received on Mon Jun 23 1997 - 21:26:40 BST

This archive was generated by hypermail 2.3.0 : Fri Oct 10 2014 - 23:30:30 BST