Aha, a smoking, erm, smoked gun!
Is that R50 with a fuckin' hole blown out its side?
Ceramics are very uncommon to fail; can happen, but not primary suspects. (On leads anyway. Can be more suspicious for SMT chips, can crack due to board flex/vibration. But still not the first thing.) The one there just looks like its paint has peeled off. But if it's actually dead yeah that's definitely something.
Missing aux, that TOP200 is your prime suspect. Small, tall transformer, off to the side, with whatever reg/controller is on it.
Mid 90s I'm surprised they splurged for a regulator like that at all (wait, how old is the TOPSwitch family anyway? I didn't know they were around back then.. Aha, introduced '94, cutting edge technology it was!), mostly they used a single BJT blocking oscillator with zener regulation. Such circuits are most common in AT/X PSUs, and VCRs and other consumer video equipment for some reason (and also, semifamously, Apple ]['s).
Troubleshooting, in general: maybe you realize this now, maybe it bears noting, maybe others will find it useful --
1. 9/10 times, replacing the fuse just begets another blown fuse. Don't be a fool. Something caused that fuse to blow! Ohm it out.
The remaining 1/10 times, maybe it died on its own (thermal cycling/fatigue, just a shoddy part, or failed due to mains surge, who knows?), and replacing the fuse is actually all that's needed. Very suspicious, and a bit spooky to be honest.
2. Switching supplies all have very typical designs; you have to learn a bit, but once you get the idea you can follow along just reading the board. In this case, to wit (referring to
https://www.eevblog.com/forum/repair/90s-delta-power-supply-repair/?action=dlattach;attach=1237205;image ):
Mains input: top right. I see a ground lug with ferrite bead, smaller (5x20mm?) fuse, a couple chokes and caps. Looks to be two common mode chokes (two windings, same direction), and one differential mode (with a yellow/white core I'm guessing?). Two boxy caps are X1 (across-the-line) film type, and the blue discs (sometimes other colors) are Y1 (line-to-ground) ceramic.
Top left: rectifier, main filter, and some startup stuff I guess. I haven't traced the circuit on the little riser board but I'm guessing it's related to either aux startup / bias, or precharge/NTC bypass timing circuit. The main FWB would be on that heatsink (on the inside I guess, can see just an edge of it here?), but there's also a TO-220. I don't see a resistor or NTC so I'm guessing there's one just out of view, and what happens is, some moment after power-on (either by a fixed time, or by the main DC supply voltage coming up to near nominal), the TRIAC is turned on, bypassing the resistor and supplying full mains to the circuit.
So, we have our first potential points of failure. If the DC bus doesn't come up, the TRIAC turns on into a hard short, of at least momentary nature due to the capacitors (which it might grunt at but still be able to handle?), but if it's fully shorted out then the whole thing is just toast, and the mains fuse must blow to clear it. Note that this won't happen if the startup circuit monitors DC bus; it'll never pull in. In that case, the resistor burns out, because it's dropping full mains power for, like, whole seconds or minutes even, however long it takes to melt, when it was only designed to endure that for a hundred milliseconds!
SCRs and TRIACs are pretty robust devices, enough that they can often survive blowing a fuse, though it usually has to be a special very-fast-blow type ("semiconductor" fuse, so named because of what it can protect -- not what it's made of, if you were wondering
). Rectifiers (the FWB) are usually a bit more robust in turn, though not so much to avoid the same requirement. That said, I have seen plenty of rectifiers which survived, not only blowing a cartridge fuse, but a
whole fucking panel breaker -- these do not open quickly, and a residential circuit probably supplies one or two thousand amperes for a cycle or two, before opening. (The breaker has a magnetic loop that opens it much quicker under sudden fault like this, in contrast to the slow thermal (bimetallic strip) mechanism that opens in seconds to minutes.)
Another option for the TRIAC, is automatic mains voltage selection. I don't see a connector on here for a "120/240" switch; it may be just out of view, or jumpered over, but this is another application where a TRIAC can be used. (Which might explain the lack of startup resistor/NTC and so it doesn't actually have inrush limiting.) The function is most likely something like: when total supply voltage is low (under say 200-250V), and after some delay (to allow for initial startup time), turn on the TRIAC, which is connected between neutral and capacitor midpoint. This turns the FWB into a FW doubler, and now you get the full ~340VDC from a 120V input.
[Double checking with the underside view, that's exactly what it is, auto voltage setting.]
[Looking at the output side now, but noticing things about the primary side traces. There are both! That must be an NTC under the FWB heatsink, and it's bypassed by another TRIAC on the inverter heatsink; and powered by an aux winding on the power transformer itself -- so it's only turned on in normal operation. Crazy!]
[And, it looks like the fly leads off the transformer, aren't a thermistor at all. They look to just dead end, actually... weird. Ah, but I haven't gotten to thermistors yet...]
In any case, the big electrolytics get charged, and on to the inverter. The upper-middle heatsink ought to have 2 or 4 transistors on it, probably BJT for this age. Looks like there's some kind of controller board to the right, or control circuitry on board. TL494, SG3524 and KA7500 (and a couple older ones I don't remember) were typical at the time, making this a voltage-mode full-wave forward converter.
Further to the right, there's orange wires for a thermistor on the heatsink, and I think a few others -- looks like there's one in the transformer itself, nice, and one on the bottom left heatsink too. These are probably read by a jellybean comparator (or op-amp being used as same -- LM339 and 324 are common sights), along with other status signals like aux voltage or DC bus, to implement UVLO, OVLO (under/over voltage lockout), overtemp, overcurrent, etc. Simply disabling the main controller as a consequence (either with a delayed auto-retry ("hiccup") or latching behavior).
On the far right, there looks to be three optos (6 pin DIPs). One should be feedback for regulation, the other two may be for startup trigger or secondary side fault trigger (which would be where the secondary side thermistor tells the primary side to shut down).
Going back a bit to the left, the small green toroid with fine wire on it, is likely a current sense transformer. I don't think anyone used current mode controls at this time (for shame), so this is probably just for overcurrent detect. It may also have windings on it for drive power -- when using BJTs in inverters, it's handy to loop back some of the load current into the base drive circuits, so they hold themselves on; it's a latching circuit, and you actually drive it by turning off the transistors (shorting mode commutation, at the drive transformer).
Usually, such a transformer looks similar to the aux supply's tall, thin transformer; unless it's just out of view, it may well be a MOSFET based circuit. Although I'm still not sure how they'd get high side drive in there. Maybe a much smaller [gate drive] transformer. Or it could be low-side drive, but that only works with push-pull, which requires very high voltages (double the DC bus, so, 800V+ devices would be mandatory!).
I don't think they had bootstrap gate drivers at the time? Not sure when they were introduced. That would be the modern solution though, just a drive IC and a pair of MOSFETs, easy peasy. Oh, or, heh, I suppose there's an outside chance they made their own, a discrete bootstrap driver. That'd be interesting.
[Okay, having looked at the layout a bit more, it's not full-wave anything at all(!) -- it actually looks to be one-switch forward converter, and I think there's an aux winding on there, but I don't think it's a freewheeling winding, or not just; this thing may very well be what's called an active-clamp forward converter. I had thought those were a somewhat newer idea! So then, I think there's a TO-247 or TO-3P main switch in the middle of the heatsink, and a TO-220 active-clamp switch nearer the transformer. Also, this explains the, 0.022uF? film cap right there.]
Anyway, onward down a bit, the output transformer obviously. And on the left, the aux supply board -- this will be a very typical (application circuit) flyback supply, with an air-gapped ferrite transformer, and it appears they used sinterglass (round blob) diodes with it. Not sure what the other heatsink is, maybe a 7805? The aux supply might be just +12V (or anything 8-20V would likely work for the main controllers) and regulated down for the motherboard's +5V standby (or whatever it uses). Hm, at least, I would assume aux powers the main controller; I don't actually see a path for it. It looks more like the TRIAC is triggered by it..! Well, I don't see the diode, but I'm pretty sure there's one tucked away in there, maybe beside the TOP200 itself, and powering it as well as the main controls.
Continuing down from the main transformer: on the right we have a heatsink with rectifiers, probably for the nearby (yellow/white) choke, maybe on the order of say +12V at 10 or 20A?
Note that there are multiple outputs, so we expect multiple rectifiers and chokes, and many windings on the transformer.
Which, with only the one control, they can't all be regulated, independently; normally a weighted combination is taken, and that works good for uniform loads, but a heavier load on one channel will drag it down, while pushing the others up (cross regulation, "sausage effect"). There are solutions for this (e.g. mag amp used in ATX supply +3.3V) but I don't think they used any here, and probably they're just doing 5V and 12V or something like that and it's fine.
Anyway, there's a small heatsink in the middle, maybe a smaller supply, maybe negative, not sure. Heatsink on the right is likely +5V output, maybe another +12 (for the second yellow/white choke)
BTW, yellow/white is the color code for Micrometals mix #26, a very lossy powdered iron, mu_r = 75. It's cheap, and that's it. It almost makes better resistors than inductors... but, just as electrolytic capacitors are rather lossy, if we use enough of 'em we kinda don't care. The consequence here is, the inductance is relatively large, to keep ripple current low, and thus core losses.
The large inductance puts the filter roll-off at a few kHz. Which means the control can only respond to load changes in some ~ms, and so relatively large capacitors are needed to keep the output impedance low. Electrolytics of the day, also had somewhat more ESR than today, which is both a curse and a blessing -- it amounts to AC losses and lack of filtering (which is improved in turn by the little rod chokes, I think you'll find those are in series with the respective outputs, with caps to ground on either side), but the ESR puts a zero in the feedback loop (i.e., the sense voltage changes immediately, in proportion to a change in inverter or load current), which helps with stability. Which is why, when the caps inevitably dry out (ESR rises, C falls), loop compensation can go south, and maybe just ripple increases, maybe it starts whining or hissing (oscillation). And why, when replacing them, they should be replaced with comparable types -- using too low ESR is also likely to cause oscillation!
Anyway, on the far left there's a bigger choke with multiple windings, and I'm willing to bet there's an aux winding on there, rectified (and probably regulated, on the small heatsink to the left?) for whatever -5 / -12V the thing needs (usually not much?). Not sure why red and gold, guessing they're just wired in parallel. And that'll be for like, 5V 40A or something.
Oh, also with two fans and a few transistors around, the thermistor down there might be for controlling one or both, or for overall fault as mentioned earlier. Could be both, staged (fan first, then fault).
One thing I have no idea about: the stranded (coarse litz?) chokes by the transformer. Maybe, like, dI/dt snubbing or something? Haven't seen that in anything else though. Shouldn't be any power transistors nearby, nor any need for additional conversion or something. Oh, or I wonder if they might be mag amp cores, for reverse recovery snubbing? The 12V supplies might be using high speed rectifiers, they didn't have much schottky in higher voltages back then I think. The 5V rectifiers (under the bigger heatsink) ought to be TO-3P or TO-247 schottky though.
Also, as an overall half-wave forward converter type, dual rectifiers are still likely in use -- one from GND to the choke (catch diode), one from secondary to choke (rectifier).
Tim