Beaten by VT100 PSU.

Brent Hilpert hilpert at cs.ubc.ca
Fri Aug 19 02:01:05 CDT 2016


On 2016-Aug-17, at 1:48 AM, Mattis Lind wrote:
> Normally I manage to repair the SMPSU that I dive into. But this time I
> must admit that I am defeated.
> 
> It is a VT100 PSU (H7831). I tested it with dummy loads and it worked fine.
> But when used in the terminal with the Basic Video board and monitor board
> it gives a jumpy picture. Both horizontally and vertically. First I thought
> that it was related to the monitor board but soon recognised that the +12 V
> had a most peculiar waveform on it:
> 
> http://i.imgur.com/d0z0NQS.jpg
> http://i.imgur.com/gQqmSN5.jpg
> http://i.imgur.com/P0dt5y1.jpg
> 
> This waveform is only present on the +12V, not the +5V, not the -12V or
> -23V.
> 
> So I connected just the Basic video board and a variable dummy load instead
> of the monitor board. I used a Variac on the input. It turned out that
> there were no problems now with the +12V. Until I pulled out and reinserted
> the keyboard. Then it was there. If I lowered the input voltage it was
> impossible to provoke this problem and also if I increased the +12V load.
> Further testing also gave that putting a few amps extra load on the +5V
> also made it resistant to this type of failure mode.
> 
> The amplitude and frequency of this waveform is shifting by +12V loading
> and AC input voltage.
> 
> The VT100 SMPSU is a primary switcher regulating the +5V. Then the +12V is
> handled by a secondary switcher which is synchronised with the primary
> switcher. The other voltages have linear regulators.
> 
> It looks to me that something in the regulation circuitry is not behaving,
> thus oscillating. But what component has failed (or is out of spec)? I
> checked transistors. I checked the waveform from the 555 chip and ramp
> voltage input to the 555. But I cannot figure out what the problem is.
> 
> I checked the 560uF output capacitor but my LCR meter said it was in good
> shape. Around 700 uF and very low ESR.
> 
> I was thinking of breaking up the feedback loop and see what happens, using
> an external +12V as an input to the regulator rather than the generated
> +12V. But hasn't got there yet.
> 
> Anyone seen this type of behaviour? Tony, do you have some piece of good
> advice?


A line of thought for consideration, with the usual disclaimer: I've seen these (VT100) supplies before but I haven't had to repair one,
the following is based only on a cursory purview of the schematic.

Let's look at some of the design nature of the supply.

All the outputs extract their energy from the main switching transformer (T1 here).
The drive for T1 (the primary driver) however, is controlled only by the +5 regulation, that is, T1 is inside only the +5 regulation feedback loop.
This is typical of many multi-output computer switching supplies.

An increased load on the +12 output must extract more energy from T1, but the +12 output has no means to directly increase the primary drive into T1
as it doesn't have a control line into that regulation loop.
Instead, through the interaction of the magnetic fields in the T1 windings the additional extracted energy for +12 lowers what the +5 output can get out of T1,
so the +5 regulation loop kicks in and tells the primary drivers to increase the power into T1 till the +5 is happy again, the +12 now being happy too.

In consequence, there may be a minimum load requirements on the output of the major regulation loop (+5 here) or a relationship (ratio) between the output
load demands to allow the outputs to function properly.

The preceding accounts for why an increased load on the +5 output can affect the +12, and cause the problem to go away.

Let's hypothesize that something in the +12 switcher is 'weak', say the gain of the +12 switching transistor (Q8) has fallen below design requirements.
When the +5 is lightly loaded, for some given demand on the +12 output, Q8 can't conduct enough energy into the +12 storage inductor (L1) to
maintain the +12 in the time it has to do so, because the low +5 demand has the T1 drive down to a narrow switching pulse.
(Note that the reason the +12 switcher is synchronised to the +5 switcher is so that Q8 'knows' when to extract energy from T1 to push into L1.)

As the load is increased on the +5 output, the +5 regulation widens the T1 drive pulse to supply it and now Q8, though weak, has more time to pass sufficient
energy into L1 for the +12 output.

The 'ripple' you are seeing is not 'capacitive' ripple as would result from inadequate filter caps - it's the wrong waveshape (aside from the wrong frequency).
I haven't thought out a full explanation but I suspect it's an oscillation set up by the load demand and an (inadequate) corrective response attempt by the +12 regulation.

Note I'm not saying Q8 is bad, go replace it; I'm just trying to describe how these functional parts interact, to account for (some of) the behaviour you're seeing,
there are other parts of the +12 switcher that could produce the scenario, like limited base drive into Q8, or pulse width or phase issues around the +12 555 monostable.
Something may have slipped out of tolerance and upset the design limits for the output relationship mentioned above.



More information about the cctalk mailing list