Date   : Wed, 06 Dec 2006 06:32:00 -0600
From   : julesrichardsonuk@... (Jules Richardson)
Subject: [OT-ish] A3020 won't turn on

Richard Gellman wrote:
>> The PSU is built into the main pcb on the left hand side. The circuitry on
>> the left hand bottom corner does look like it is a switched mode power
>> supply. I can't see the number on the top of the integrated circuit there
>> as
>> the picture isn't close up enough but if you google the part number you
>> can
>> confirm what it's doing from the manufacturer info.
> 
> I'm inclined to disagree, but it would be the short-sighted leading the
> blind. My reasoning being the apparent simplicity of the circuit and the
> presence of 4 discrete rectifier diodes adjacent to the transformer. The
> transformer itself also seems to have 5 output lines, which I imagine to
> be +12V, +5V, 0V, -5V, -12V (even though -5V is not used).

Remember that a switcher will rectify the mains "upstream" of the transformer 
- it's the job of the chopper transistor(s) [1] (under control of the feedback 
circuitry) to then pulse (switch) this HV DC and feed it into the transformer.

[1] An average home machine will only have one. I've seen up to eight in more 
serious systems though!

The two most common culprits in any failure though are dried-out filter caps 
(usually on the mains side) and a blown chopper transistor. However, a shorted 
chopper should take out fuses too unless there's some sort of additional 
failsafe circuitry present (such as fusable resistors). I've seen input 
rectifier diodes go too, and also faults in any thermal protection circuitry 
(a typical 'home' machine may be built to a price and not have that, though!)

If you've got a multimeter, check for shorted caps, shorted chopper 
transistor, and the presence of rectified mains downstream of the diodes as a 
start. As someone else said though, caps often fail high-ESR rather than 
shorted, and you'll need an ESR meter to detect that (or just do some random 
replacement of most likely culprits).

I don't have a photo from the right angle handy, but in the one that I have it 
looks like there's probably a chopper transistor mounted on the heatsink which 
separates the transformer from the bulk of the machine. The round toroidal 
transformer is likely the isolation between the low-voltage pulse generator 
(the IC) and the chopper transistor which works on the HV side.

>> It's quite unusual to have a transformer along with a switcher, normally
>> switchers just rectify AC mains in and use a reservoir capacitor to
>> provide
>> a 400v DC feed which is chopped by the switcher. Possibly Acorn wanted to
>> they drop the mains voltage down first then use the switcher to provide
>> multiple outputs at high currents.

There's still a multi-tap transformer though to drop rectified and then pulsed 
DC down to the various output voltage levels, which then get rectified again 
and smoothed back to steady DC. Typically though this transformer's pretty 
small (compared to a linear supply) - I'm still surprised at how large the one 
is for the A3020.

Check diodes on the output side too - if one of those has died on the +5V line 
  then this could explain your lack of +5V but *something* happening on the 
other rails, as the feedback controller will be spazzing out quite merrily to 
itself. It could well be cycling as it tries to bring the output up to +5V 
before detecting a fault and shutting down.

> On the up side, I've managed to bodge a PC ATX PSU into running it, so the
> the onboard PSU is completely unrepairable, the machine is still usuable
> with the application of a suitable replacement PSU.

Hmm, so you know the machine's OK. Note that switchers will often do odd 
things if there's no load present (even tear themselves to bits in some 
designs :) - so it's worth checking for any dead fuses between the PSU and the 
computer itself (just in case you're feeding the PC's PSU lines in downstream 
of any damaged fuses).

cheers

Jules

-- 
there's a carp in the tub
there's a carp in the tub
so nobody's taken a bath