[cctalk] Re: Repairing an Olivetti M24 PSU

Thank you Brent. Your re-drawn schematic looks amazing! I am going to be away for a few days, I have reassembled the machine and there is a lot to digest, so it may take a while for me to go through it all. Regards Rob From: Brent Hilpert &lt;bhilpert(a)shaw.ca&gt; Sent: 07 September 2025 04:45 To: General Discussion: On-Topic and Off-Topic Posts &lt;cctalk(a)classiccmp.org&gt;rg>; rob(a)jarratt.me.uk Subject: Re: [cctalk] Repairing an Olivetti M24 PSU On 2025Sep 5,, at 11:35 PM, Rob Jarratt via cctalk &lt;cctalk(a)classiccmp.org <mailto:cctalk@classiccmp.org> > wrote: Thanks Scott. I removed a couple of other caps on the 12V output, tested them and then put them back as they seemed OK. I have tested the ESR (in circuit) on all the other electrolytics and they all seemed fine. I tested a few resistors but not particularly rigorously though. The resistive load I applied was calibrated to the specs of the PSU, above the minimum load and below the maximum. I didn’t load the -12V and +15V outputs though, only +5 and +12. The -12V output has no minimum current spec. I can always consider the meanwell supplies but so far I have been able to repair my PSUs and I hope to do the same here. I am going to try installing it back in the machine today and see how it fares. Here is yet another (4th) version of the schematic, redrawn for greater functional clarity: http://madrona.ca/e/misc/OlivettiM24PS.pdf This version follows the component IDs of the marked-up version of the main Olivetti schematic, though I’m not clear where these IDs came and they don’t seem to have much order (perhaps they came from grid locations on the PCB?). These IDs are inconsistent with the main Olivetti schematic and Olivetti ThOfOp, which seem to use a component type designator rather than component instance ID. There are some number of errors or inconsistencies in the Olivetti docs. - - - Some ThOfOp further to the Olivetti manual: This is a 'self-oscillating’ design in that the switching-oscillation is built into the primary drive circuit - there is no independent oscillator driving the primary driver. The primary drive circuit is basically a dual-supply push-pull amplifier driving the main transformer primary, though the drive output has a capacitor C141) and a winding of the control reactor transformer (T365) in series in the circuit. The feedback loop to make it oscillate goes through the base-drive transformer (T366) with its two secondaries feeding the bases of the two drive transistors. The main feedback path is presumably via T748.4,5. I would guess winding T365.5,6 is a 2nd-level regulation control altering the drive level. The T366 base-drive secondaries would be polarity-oriented so they drive the bases in cross-over, that is one transistor ON driving the primary feeds back (with delay) to eventually drive the other transistor ON. So to speak, the oscillator is a high-power magnetically-coupled positive/negative-pulsing astable flip-flop. You see that in the +/- pulses on the input to the output rectifiers (per your scope pics). The Olivetti ThOfOp says the output rectifiers are half-wave, but they’re two half-wave rectifiers operating on opposite polarities from a center tap .. which is a full-wave rectifier. Not sure, but I surmise that C141 may be a/the primary determinant of the oscillation frequency, while the 4 caps around the driver bases (C131,etc.) determine the intra-cycle pulse periods and delays. Or C141 may be there to keep any DC drive imbalance out of the primary (..?). Regulation-control operation in summary: - Q83,85 form a differential pair for the regulation error sense. - Q83.B is the reference input, supplied by U49. - Q85.B is the sense input, supplied by V+5s from the +5V output via divider R80,R87,RV89. The trail of regulation-control influence, by example: - A + increase of V+5s results in increased drive current thru Q85.BE, sending Q85.C more +. - This drives Q86 to greater conduction, thus Q86.C goes lower to GND. - This lowers Q44.B, reducing its conduction and so reducing current through the control reactor input winding (T365.7,9) - The reduced current in the reactor input winding reduces T365’s core saturation. This enables the primary driver pulse currents flowing through the primary-side control winding (T365.3,4) to have more magnetic influence in the T365 core, which is to say those driver currents now see a higher impedance to their current flow. - The current through the main transformer primary winding (T748.1,2) is thus reduced, counteracting the increased V+5, thus completing the control loop. - - - As the oscillation is built into the primary drive circuit, presumably the control system has a lower limit that inhibits shutting the primary drive completely off because if it did, the primary oscillation would stop. This lower limit may be designed into the maximum impedance of the T365 primary-control winding, but it may also involve the (relatively high) minimum load specs per the manual. You didn’t specify the currents for your dummy loads. The min spec for the V+15 is 1A, i.e. 15W. That isn’t trivial so you may want to ensure the total dummy load is accounting for that: total power of dummy load(s) > total power of min output specs. Hypotheses, in the absence of other measurements: the control circuitry is faulty and holding the primary drive level to a minimum, enough to maintain oscillation but not enough to produce significant output. The primary drive is minimised by zero current in the control reactor input winding (T365.7,9) and increases with increasing current in that winding. Observe: Measure the V across the 16*100 resistor bank when the PS is functioning, perhaps at different load levels, and and when it is in fault. There are all sorts of failures in the control circuitry that could send the supply into minimum drive (basically anything that shuts off current in the T365 input winding). The Olivetti ThOfOp suggests fault-detection circuitry (involving Q59,75) can latch-up the supply into minimum drive. I haven’t looked into that circuitry in much depth, but the latch-up mechanism seems to be: - at lesser/minimum drive the T365 core goes out of saturation, allowing the now-varying magnetic field from remaining primary drive pulse current to induce V in T365.1,2. - this is rectified and filtered and supplies drive to Q86 via Q75 (if Q75 conditions are appropriate). - this holds Q44 off to shut off T365 input current, so keeping the primary drive at minimum.