Reforming caps and CRTs

Tony Duell ard at p850ug1.demon.co.uk
Wed Apr 22 15:07:40 CDT 2009


> 
> On Tue, 2009-04-21 at 20:04 +0100, Tony Duell wrote:
> 
> > The PSU is, of course, a switching regulator run from a mains-frequency 
> > step-down transformer. Strangely there's no crowbar circuit. If that 
> > chopper transistor shorts, the 5V line leaps to about 30V with fatal 
> > effects on all the chips. I don't know why HP cut corners in this way.
> 
> Shades of the Sony sets of the late 70s to mid 80s with the GCS power
> supplies (Gate Controlled Switch, or Ghastly Catastrophic Semiconductor,

Ah yes.. I rememeber those . There were various 'Great Conversions of 
Semiconductors' to replace them with a bipolar chopper transistor like a 
BU208A. 

> depending on who you ask).  Oh, except that clapped full-wave rectified
> mains across all the following regulators when (not if) it failed dead
> short.  Which promptly failed dead short too, with hilarious
> consequences.

This reminds me of the Boschert 2-stage PSUs used in some classic 
computer equipment (PERQ1s, for example). The basic topology is that you 
rectify and smooth mains (giving about 350V DC) and then ring that down 
to about 150V using a non-isolated switching regulator. The output of 
that feeds a couple of transistors running as a free-running oscillator 
driving the main chopper transformer. The output of that is rectified and 
smoothed. Feedback is then applied from the main (often +5V) output to 
the 150V regulator. Controlling the output of that indirectly controls 
the output voltage of the whole PSU of course.

Now for the nasty failuyre mode. The first chopper transistor (in the 
150V regulator) goes short-circuit. So the output of that stage leaps to 
about 350V. The second oscillator keeps on running, so all the PSU 
outputs jump to over twice what they should be. At this point you hope 
the crowbar fires, if notm you have a _lot_ of ICs to replace. If the 
crowbar does fire, it shorts one of the outputs of the PSU. This causes 
the second oscillaotr circuit to work harder and draw more current. This 
is detected by the current limit circuit, which removes the drive from 
(you guessed it) the 150V regulator chopper. Unfortunately, that is 
shorted, so removing the drive doesn't do a darn thing. The 2 transsitors 
in the second oscillator then go short-circuit. Now, across the 350V 
rectified mains supply you have some shorted transistors, some windings 
(both the inductor of the 150V regulator and the chopper transformer 
primary), anf the current sense  resistor (say 0.15 Ohms). That resisotr 
then unrs out, taking the sense transistor with it. And then the chopper 
control IC (a 723 IIRC), a few other transsitors and some small passives 
'join the choir invisible'. Oh, and a couple of PCB tracks melt.

Yes, I have seen the aftermath of this. I also had to repair it....
 
> These would come in with the customer saying "It went bang and then the
> picture and sound went off.  It's probably just a fuse, it won't be too

Oh, the fuse proaly has failed. But as we all know, 'a transistor 
protected by a fast-acting fuse will protect the fuse by lowing first'. 
And the same applied to GCS devices.

-tony



More information about the cctalk mailing list