On 29/08/2025 07:35, Wayne S via cctalk wrote: Wayne, Try my copy... https://1drv.ms/i/c/277a0739f125010e/EU8NYVbyy-tIoCBgfquJi5cBfht4wVlyBjAkdg… Dave

I'm confused -- maybe I misread-- the original post says D134/D136 and the most recent post says D15/D16 and D18/D19. Which diodes do the oscilloscope traces go with? Scott On Fri, Aug 29, 2025 at 10:36 AM Robert Jarratt via cctalk < cctalk(a)classiccmp.org&gt; wrote:

Hookup your scope across the 5v and gnd on J254 and see what the trace looks like. Also do +12 v. And -12v on j100 That will tell you if one rail is bad. You should have not much ripple. If there is a lot, trace that voltage rail back. Sent from my iPhone

Rob, replace c21 because of the bulge. If you want to check an electrolytic cap with a ohmmeter ,(kinda hard with a digital one), connect the meter across the ends. You should get either infinite resistance or a momentary reading of zero going back down to infinite resistance as the cap charges. Then reverse the leads and it should be infinite from the start. On an old analog meter with a dial, you can quickly tell by see the needle deflect full scale in one direction then settle back down to infinite. I keep an old meter around just for this purpose. Sent from my iPhone

One other thing… With it powered on and measuring voltage at the outputs but not at the plugs, wiggle the plugs a bit and see if there wierdness with them. plugs and connectors go bad often and it doesn’t seem obvious that they do but saves a lot of time just by checking them now. Sent from my iPhone

Where is the scope connected? Can you measure across the 5v connection and spread out the trace to see the ripple? When connected there you should a fairly square wave with no negative going pulse. The top part of the trace should hsve the ripple. Sent from my iPhone

On 9/3/25 02:24, Rob Jarratt via cctalk wrote: Those are not real.  They are conducted interference from the switching supply getting into the scope preamp via the ground lead. I have seen this MANY times, ignore it. Jon

On Wed, Sep 03, 2025 at 05:18:44PM +0100, Rob Jarratt wrote: One test I do is put the probe and ground clip both on ground. If you still get a good amount of noise you know its not real. Works best when ground and where you are probing are close together so you aren't moving probe much to do the test.

On 9/3/25 11:18, Rob Jarratt wrote: Switching power supplies generally radiate a ton of electrical fields at their switching frequency.  If you see insanely high frequencies in these measurements, you can usually assume they are radiated interference.  You can also turn on the scope's bandwidth filter.  I did see REAL ripple in one of the traces, there were long straight lines with slight tilt between the noise pulses, those are the real ripple. Improving the ground connection at the scope probe also helps. The power supply injects currents into the ground terminal due to capacitance between output transformer windings, and these current flowing in the probe's ground braid contaminates the measurement. Possibly running a HEAVY copper braid between the scope's ground terminal and the power supply ground will reduce the effect. Jon

On 9/5/25 18:04, Rob Jarratt wrote: Many switching supplies have VERY tricky startup circuits that tickle the controller to start and then power the controller from an auxiliary winding on the transformer.  Some of these circuits might require some capacitor to fully drain before they can start properly again.  Some of these circuits may have a really high resistance value that charges the controller's supply cap, and then a transistor turns off that path when the controller is running to avoid overheating that resistor.  These circuits will be at mains potential so be careful when testing them. Jon

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. Thanks Rob From: Scott Baker &lt;smbaker(a)gmail.com&gt; Sent: 06 September 2025 07:11 To: General Discussion: On-Topic and Off-Topic Posts &lt;cctalk(a)classiccmp.org&gt; Cc: rob(a)jarratt.me.uk; Jon Elson &lt;elson(a)pico-systems.com&gt; Subject: Re: [cctalk] Re: Repairing an Olivetti M24 PSU When a power supply gives me trouble, I often pull all the electrolytic capacitors, test them, and then install new ones if I have them (because if you're going to solder a capacitor in, you might as well solder a new one in). A couple times, it hasn't been the big filter capacitors that were the problem. There's some risk in "shotgunning" component replacement like I'm suggesting in that you might make something worse by replacing a perfectly good capacitor with a modern replacement that isn't quite the same. But it has a good track record for me... Also, old resistors can drift or open, so you might want to scrutinize those too. What value is your resistive load? Some switchers specify a minimum load. Another option is just to see if you can find a meanwell open-frame power supply or similar at mouser or digikey that's about the same size and has the right specs and replace it. Had to do this with an Epson QX-10 switching supply. Even though I had the Sam's Notes for it, I gave up trying to find the obsolete semiconductors that it needed. Smaller, more efficient, cooler than the vintage power supply. Scott On Fri, Sep 5, 2025 at 6:56 PM Jon Elson via cctalk &lt;cctalk(a)classiccmp.org <mailto:cctalk@classiccmp.org> > wrote: On 9/5/25 18:04, Rob Jarratt wrote: Many switching supplies have VERY tricky startup circuits that tickle the controller to start and then power the controller from an auxiliary winding on the transformer. Some of these circuits might require some capacitor to fully drain before they can start properly again. Some of these circuits may have a really high resistance value that charges the controller's supply cap, and then a transistor turns off that path when the controller is running to avoid overheating that resistor. These circuits will be at mains potential so be careful when testing them. Jon

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.

On 9/6/25 01:26, Rob Jarratt wrote: Well, maybe.  If it is a steady ~20 KHz on the transformer, the startup circuit is OK.  If it is an intermittent pulsing a few times a second, then the startup is "trying" but the converter is failing to power itself.  (Or the crowbar is killing it.) Jon

On Sat, 6 Sep 2025, Rob Jarratt wrote: Did you test/replace the startup voltage capacitor on the primary side? It is usually a small one (e.g. 10µF/50V) around the switching controller, and often located near a heat sink and thus prone to fail. BTW check all capacitors near heat sinks or power components. Christian