Author Topic: PSU repair - vintage Macintosh G3 Desktop (Read 1979 times)

blackmasked · « **on:** May 05, 2020, 04:17:53 pm »

Hi there
I'm a beginner with an IT background, but little in terms of electronics. I'm trying to repair a PSU on a vintage Macintosh G3 Desktop. It was working fine and one day gave up a ghost. When it did go, there was no smoke, no fireworks, nothing. Now I'm getting no output voltages, not even +5VSB.

I started by replacing all of the electrolytic caps. I assumed this would be a good first step due to the age of the machine (1997-1998). No joy.

I inspected the board and found no burned out/exploded components, no damaged/burned traces.

I also checked my own recapping multiple times. All solder joints are fine and polarities are correct.
Cleaned the board to make sure flux is not shorting anything. Still no joy.

Next I tested the following components in circuit:

Fuse - fine.
Bridge rectifier - fine.

Transistors and MOSFETs - one of the Schottky rectifiers (MBR3045PT) on secondary side tested short in circuit, but after removal tested fine. Would this indicate that perhaps some of the nearby components could be faulty which resulted in dodgy readings?

I'm a bit confused by TOP200YAI MOSFET. All pins are shorting, but instead of constant beep I'm getting something that could be described as a beeping grinding noise. I've got one on order just in case.

Tested some of the diodes in circuit - also fine.

I don't want to continue removing parts until absolutely necessary. After all it's a vintage piece of hardware and sooner or later I may get unlucky and lift a pad or two.
If that happens getting a replacement will be difficult and exspensive.

What would cause complete lack of any output voltages? If I'm not mistaken auxilary supply is seperate from primary and secondary side and even if these two are dead I should be getting +5VSB.

Apologies if some of the terminology used is a bit off. I learn as I go.

I would appreciate any suggestions as to what to look for next.

Manul · « **Reply #1 on:** May 05, 2020, 06:56:44 pm »

Maybe you can post good quality pictures from both sides of the board?

Anyway, please be very carefull with your troubleshooting, and note that capacitors may stay charged for some time.

Some starting ideas:

1. Measure all power rails (outputs) for shorting.
2. Connect to mains, and see if main bulk capacitor has 300V DC.

Manul · « **Reply #2 on:** May 05, 2020, 07:05:28 pm »

If rectifier is shorted in circuit, but ok out of circuit it probably means that rail is shorted.

blackmasked · « **Reply #3 on:** May 06, 2020, 09:10:13 am »

RE charged capacitors - I always use a resistor to bleed of the excess voltage.

Great tip regarding shorts. It turns out that the +3.3V line shorts to ground.

Links to the pictures are as follows:
https://ibb.co/Xx5CX08
https://ibb.co/SPmF5YB
https://ibb.co/tMsCsxL
https://ibb.co/X2JWzJQ
https://ibb.co/rwMhLnC
https://ibb.co/wN0rGTM
https://ibb.co/pfV8FKH
https://ibb.co/4tPPMwD

kripton2035 · « **Reply #4 on:** May 06, 2020, 10:29:09 am »

I would check the resistor that powers the oscillator in the primary side. and the small electrolytic capacitor that is next to the oscillator IC.
or like stated above, the mains big 400v capacitors for ESR (but it should blow the fuses or mosfet if too high)

kripton2035 · « **Reply #5 on:** May 06, 2020, 10:29:57 am »

blackmasked · « **Reply #6 on:** May 07, 2020, 01:04:52 pm »

Unfortunately, I do not have an ESR meter, but all of the electrolytic caps are new and I would expect them to be fine.

kripton2035 you said to check the resistor that powers the oscillator IC. Pin 7 on UC3843AN seems to be the Vcc, but none of the resistors nearby seem to be connected to it.
Apologies if I'm being a bit daft here. ;]

What makes you think that the resistor could be at fault here? From what I understand resistors rarely go short. Not trying to be arrogant here, I'm genuinely interested why.

Not sure if this is a valid method of narrowing things down, but as I already know that 3.3V shorts to ground and the separation between primary and secondary is quite obvious, I did a continuity test between every leg/pin on the primary side and 3.3V on the secondary side. None of the components on the primary side a shorting to 3.3V. Does that mean that the fault must be somewhere on the secondary side? Would this also mean that all ICs that bridge primary and secondary side are most likely fine?

kripton2035 · « **Reply #7 on:** May 07, 2020, 01:16:08 pm »

when a smps is dead, and no fuse is blown, it sometimes means that the oscillator is not working, often not powered on.
to power this oscillator, a resistor, or serie of resistors send a fraction of the mains to this chip. if they are cut the oscillator does not start.
can you (carefully there are high voltages on the primary side) measure the voltage of the power pins of the uc3842 ? see if they are valid.

kripton2035 · « **Reply #8 on:** May 07, 2020, 01:19:26 pm »

PS: if you want to repair (other) smps, you should really build yourself :
an esr meter (easy like this one : http://kripton2035.free.fr/analog%20esr/esr-dmm-adapter-.html)
a short circuit meter (also easy like this one : http://kripton2035.free.fr/Continuity%20Meters/continuity-short.html)

blackmasked · « **Reply #9 on:** May 07, 2020, 02:03:02 pm »

Can I plug it to the mains without all three rectifiers from the secondary side? I desoldered them all in order to test them.

kripton2035 · « **Reply #10 on:** May 07, 2020, 04:06:43 pm »

if they test ok you should put them back in place.

blackmasked · « **Reply #11 on:** May 12, 2020, 11:14:53 am »

I tested the voltages on UC3843AN and the 200V caps.
I'm getting 680mV on Vcc pin and 170V on both caps, which then drops rather quickly to 20V after unplugging the power.

If I'm reading the datasheet correctly nominal input voltage for UC3843AN is 11V. Does that mean that resistor(s) have gone bad, increased resistance and now limit input voltage for the oscilator IC?

It also turns out that the TOP200YAI was faulty after all. I replaced it with TOP222YN. Not like for like, but close enough I hope ;]

3.3V rail still shorts to ground.

kripton2035 · « **Reply #12 on:** May 12, 2020, 05:43:31 pm »

it seems the uc3843 is not powered correctly. so yes search for a cut resistor coming from the 170V line, high values in the megaohm, or some 100k in series that goes to the vcc pin of uc3843, generally they have a small capacitor and a zener diode.

blackmasked · « **Reply #13 on:** May 14, 2020, 01:40:45 pm »

Good call kripton2035.

The only high value resistor in close proximity to UC3843AN failed short.
After pulling that rotten tooth out I also removed a bunch of other high value resistors from the secondary side. These also failed short for a grand total of 6.
After getting them all out +3.3V rail no longer shorts to GND.

Looks like I'm slowly getting there and you help has been invaluable.

There is only one thing I'm wondering about. Are these failed resistors the cause of all problems or the end result? If I simply replace them are they likely to go again?

kripton2035 · « **Reply #14 on:** May 14, 2020, 03:07:13 pm »

strange that they are shorted instead of cut ? what are their values ?
if they are shorted, this means that the uc3843 got some 170V instead of 11V ... not sure it survived !

blackmasked · « **Reply #15 on:** May 15, 2020, 04:18:05 pm »

After checking their values I was even more confused than before.

First of all I made a wrong assumption that resistors are a bit like caps and bigger resistor means higher resistance. It turns out that the majority of the ones I removed are between 1-22 $\$\Omega\$$ .

What I also didn't realise is that continuity test is not a good method of determining if resistor is short or not. I did not know that beeper turns off at values higher than 250 $\$\Omega\$$ . ;]

Therefore out of the ones I have removed so far only 1 seems to be out of spec.
0,18 $\$\Omega\$$ rated one that measures between 0.4 and 0.5, depending on the meter used. I'm not too sure if I got the colour of the first band right. Sometimes red and brown are very similar.
It does not help that I'm slightly colour blind ;]
https://ibb.co/BgCc7QB

There is also a 1 $\$\Omega\$$ rated resistor that I missed before.
https://ibb.co/9T47mWH
Cracked side was facing the PCB and not visible until it was removed.
This resistor goes from 200V caps towards a zener diode that checks out fine and then to an auxilary supply. I guess that would explain lack of +5VSB.
Despite how it looks it still measures 1 $\$\Omega\$$ . However, I'm not sure how accurate that reading is as the lowest scale I have is 200 $\$\Omega\$$ .

I've also traced 2 resistors to UC3843.
[grey, red, green, orange, brown -> 825k +-1% (816.75k - 833.25k)]
Both measure 833k so I guess these are not the culprits.

What I also described as +3.3V shorting to GND is not really a short after all. I tested a bunch of other ATX PSUs I have and all of them have continuity between +3.3 and GND, so I guess it's perfectly normal.

blackmasked · « **Reply #16 on:** May 26, 2020, 07:10:03 pm »

It's alive!

I ended up desoldering way too many parts in the process of troubleshooting, but I've learned a lot along the way.

Not bad for someone who 3 weeks ago didn't know the difference between bridge rectifier and a filter capacitor ;]

Thanks everyone for suggestions.

kripton2035 · « **Reply #17 on:** May 26, 2020, 08:46:16 pm »

what component did you finally change ?

blackmasked · « **Reply #18 on:** May 26, 2020, 09:31:27 pm »

TOP200YAI AC/DC converter and a single 1 $\$\Omega\$$ resistor. ;]

Is it necessary to replace silicone that I had to remove when recapping?

From what I've read neutral curing RTV silicone would be safest for electronics.


EEVblog Main Site	EEVblog on Youtube	EEVblog on Twitter	EEVblog on Facebook	EEVblog on Odysee

Author Topic: PSU repair - vintage Macintosh G3 Desktop (Read 1979 times)

Share me