2
\$\begingroup\$

my brother has an old mac computer that didnt work, so he asked me to fix it, its one of those macs from the early 90s, a Macintosh Classic II.

So i opened it up and i found electrolyte all over the power supply, some of the secondary caps leaked it, i also discovered that the MOSFET was burned up, the shotky diodes looked "warmed" because even the part number wiped out, but they tested ok, i tested every other component with my multimeter and the rest tested ok (besides mosfet and fuse).

I replaced every electrolytic with low ESR caps, including the main reservoir cap, and i also replaced the MOSFET, the mosfet wasnt exactly the same part IRFBC40A (original) vs IRFBC40APBF (replacement) but i compared both datasheets and they seem to be the same thing on paper.

Anyway, after replacing those parts (electrolytics and MOSFET) the powersupply lid up, the computer turns on, the OS loads and it seems to be working properly, however theres a very loud tone that appears to come from somewhere between the FET and the transformer, i cant exactly tell which one is vibrating since both parts are very close together.

This tone starts at a very very high frequency, im guessing above the human hearing threshold (say 25KHz), and starts sweeping down slowly, first you can barely hear it, but as the frequency decreases, you can hear it very loud, the tone is always sweeping down, i can compare it to the sound of a piezoelectric buzzer on old computers, i havent actually waited for it to end, i've turned it off when the sound is around 1KHz, but the sound is so intimidating that i can imagine the SMPS completely melting down at some point.

Could it be that the transformer's primary is not completely dead, but somehow damaged? or that the electrolyte somehow leaked into the transformer?, could it also be that the MOSFET may not be the correct one?

Any ideas of what might be the culprit? like i said the PSU works but its unbearable to hear that tone.

Thanks!

\$\endgroup\$
5
  • 1
    \$\begingroup\$ I'd guess that the change in switching frequency is driven by the change in temperature due to warmup of PSU components and not wall time since power-on. It implies that there is at least one other damaged component in the PSU that requires replacement. Is the output voltage stable while the tone changes? \$\endgroup\$ – HikeOnPast Aug 15 '12 at 5:16
  • \$\begingroup\$ Thanks for the input DeanB, The voltage seems to be stable, the tone sweep takes a considerable amount of time (minutes), and i havent waited that much, mainly because its unbearable. However, if i turn it off and on again, the tone's frequency will start sweeping down from around the same spot it was before i turned it off, which makes me believe that what you say about temperature may be true, ive tried warming up components, no luck thou, although i havent tried freezing them.Do you think the switching frequency change could be due to a faulty optocoupler? or something on the feedback circ? \$\endgroup\$ – S.s. Aug 15 '12 at 5:20
  • \$\begingroup\$ If it's designed to run at 50 KHz or 100 KHz, then it probably can't work at 2 kHz like what you are hearing. That makes me think the oscillation is imposed onto the existing PS running frequency. Is it load sensitive? Fire up the floppy drive and see if the tone is affected. \$\endgroup\$ – gbarry Aug 15 '12 at 6:01
  • \$\begingroup\$ The tone is always dropping, meaning its not steady in a certain frequency like 2k, which makes me believe theres something either leaking or just plain wrong with it. I will try the floppy trick, thats a good idea. The computer keeps working while all of this is happening so that adds more to the confusion... \$\endgroup\$ – S.s. Aug 15 '12 at 6:12
  • 1
    \$\begingroup\$ Can you add a photo of the PSU? Mark the spot where you think the sound is coming from. \$\endgroup\$ – Szymon Bęczkowski Aug 15 '12 at 9:56
3
\$\begingroup\$

The PBF suffix at the end of IRFBC40APBF means lead free/RoHS-compliant. It should be a one-to-one substitution with the original part.

I found a web site where the same symptom was described. It's evidently common, and is called the 'falling bomb' effect. There are most likely some bad electrolytic capacitors on the main board in the audio section that need replacing. I would just go ahead and replace whatever electrolytic capacitors you're safely able to.

I also found a website with a repair guide: Classic Mac Tech Docs, v2.0 It has power supply schematics, etc. and is a good general reference. (They recommend replacing C1 as a precaution - your mileage may vary)

Exercise extreme caution in the high voltage circuits feeding the CRT. There are lethal voltages present there!

\$\endgroup\$
8
  • \$\begingroup\$ Wow, that seems to be the problem, however those users describe it to sound on the mainboard, mine distinctly sounds on the PSU, i've put my ear as close as possible without getting shocked and thats whats telling me, thanks very much for the answer thou, i shall look at the mainboard. And thanks for the schematics! \$\endgroup\$ – S.s. Aug 15 '12 at 22:27
  • \$\begingroup\$ You were right, i checked ESR on the smt capacitors on the computer mainboard and i was amazed that most of them were bad and absolutely out of specs, some of them were not even recognized by the ESR meter, when i desoldered the first one, suddenly that nasty burned electrolyte smell emerged, and i knew those were the ones causing the problems! thanks! \$\endgroup\$ – S.s. Aug 16 '12 at 5:04
  • \$\begingroup\$ +1; Note: Quick check, if possible: Disconnect any load from the SMPS. If "falling bomb" effect goes away - bingomat! Reconnect different subcircuits until weird sound reappers to track down the problem. \$\endgroup\$ – zebonaut Aug 16 '12 at 6:49
  • 1
    \$\begingroup\$ It can be incredibly challenging to isolate a noise source in a unit under test. I generally go with a plastic stethoscope or a rolled-up tube of paper to acoustically 'sweep' the area to try and find where the sound is loudest. I often find that the source is rarely where I originally speculate it is. \$\endgroup\$ – Adam Lawrence Aug 16 '12 at 12:37
  • 1
    \$\begingroup\$ Never an easy task on a multilayer PCB. You may have to use two irons on each leg and walk the capacitor out. Worst case, cut the leads off and solder the new cap to the old leads :) \$\endgroup\$ – Adam Lawrence Aug 16 '12 at 18:50
2
\$\begingroup\$

Mandatory safety note: Don't work on mains-supplied circuits unless trained to do so. Lethal voltages are present! Now, ...

It is possible that your power supply a flyback type with variable switching frequency. You still have a duty cycle, except this duty cycle is not derived from a fixed frequency oscillator, but from a variable frequency that acts around the power path. Such designs often run with a switching frequency inversely proportional to the output power, i.e. the more power you take at the outside, the lower the frequency gets.

Some possible reasons why your power supply would slowly reduce its switching frequency the longer it runs:

  • A faulty component on one of the boards that are supplied, i.e. power supply o.k., but another problem further downstream.

  • A faulty component on the secondary side (Cap inserted with wrong polarity? Cap operating beyond its max. rated voltage? Diode with excessive leakage current? ...)

  • A bad snubbing circuit? Often, the spikes at the onset of the flyback cycle are subbed into a capacitor via a diode (something like 1 A, 1 kV, ultrafast). If the diode fails open, no snubbing takes place, and if the main transistor survives, it will get hot (GOTO "A bad driver for the main transistor"). The snubbing capacitor needs to be continuously discharged and therefore has a paralleled resistor (some 10...100 kΩ). If this resistor fails open, the power transistor is also in danger.

  • A bad driver for the main transistor: If the main switching transistor is not driven as it should be, i.e. not fully off or fully on with really fast transitions, it will become hot. The high temperature will change its behavior. However, in most cases, this will not be a slow process like you mention, but a rather surprising event, often including a smallish explosion. Check all components in the driver. Often, when the main transistor blows, it takes parts in its driver circuit with it.

  • Also, electrolyte is conductive. This is a safety hazard, because spilled electrolyte across the barrier between primary and secondary may allow high voltage to be present on the output side. It is also a possible cause for your problem: Electrolyte, in places where it shouldn't be, may alter the circuit's behavior (This would even explain why the oscillator of a fixed-frequency design varies its frequency). Also, electrolyte is not really healthy (weird salts, all kinds of nasty aliphatic or water-based chemicals, you name it...), and it is corrosive. Before proceeding any further, wash your circuit board using alcohol or water with a mild detergent. Do not apply power unless the board is completely dry again.

\$\endgroup\$
1
  • \$\begingroup\$ Thank you very much for taking the time to answer like that, it does make sense to me. My thoughts are what you are describing, mostly electrolyte, i already cleaned the board but it may have fallen in all the wrong places. Also im betting for some leaky component since i already test them all out of circuit, perhaps i should add a bigger reverse voltage and see what happens, or perhaps i overlooked something. Either way, i never had a SMPS do this before, they either work, or they shut down, or they ring, but they dont sweep!. Thanks again ill keep you posted \$\endgroup\$ – S.s. Aug 15 '12 at 15:49

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.