Switching PSU problems

[Rattanee]

Okay folks, I have a rather mysterious problem here with an arcade power supply. Basically, this is a switching power supply that runs from 100VAC, and supplies 3.3, 5, and 12VDC.
The voltages are all fine, however I'm having the weirdest problem ever. Whenever I use this power supply with an arcade PCB, I'm having graphics issues that develop over time, or the game connected won't even boot up.

When a game won't boot up with this power supply, if I leave the power supply running, and just disconnect and reconnect the power connector on the game board, the game springs to life, but as with almost all boards, will develop graphics glitches over time. The game boards all work fine on other machines.

I'm thinking the PSU probably has capacitor issues?

Does anyone know what I should be looking for on the outputs with an Oscope? Or a guide on how to check power supplies for unfilteredness?

I could just swap all the lytics in it and see if that helps, but I'd like to actually pinpoint the issue.

Thanks guys

 

[KrisBlueNZ]

Yes, from your description I would suspect noisy supply rails too.

You should be able to see the noise if you connect an oscilloscope across each rail in turn. The switching frequency will probably be in the range 30 kHz (for older power supplies) up to perhaps 100 kHz, so look for that in the scope trace. You'll be able to see the actual frequency if you wave the scope probe in the direction of the chopper, or connect it to a secondary of the transformer before the rectifier.

When you're looking for small signals on switching power supply rails with an oscilloscope, be sure what you're seeing is actually differential-mode noise - i.e. noise across the output, between the positive rail and the 0V rail. If you touch the probe tip to the ground clip and still see some noise, it's common-mode noise, and when you're measuring across the rail, the scope will show that common-mode noise as well as the differential-mode noise, so you'll think there's more noise across the rail than there really is. Try using a battery-powered scope to reduce this problem.

Personally I would just replace all of the output electrolytics first and see whether that fixes the problem. Find the specifications for the original parts and make sure the replacements are rated for at least as much ripple current and the same or somewhat lower ESR than the originals. Also suspect them if they're cheap Chinese brands rather than one of the reputable Japanese ones - UCC (United Chemi-con), Rubycon, Nichicon or Panasonic.

The problem could be something else - have you actually measured the rail voltages? Do you have a specification for the minimum and maximum allowable voltages of each rail? Can you measure them over time to see whether they drift?

 

[Rattanee]

Well I was impatient. All the caps were labeled KME and rated 105°C which from what I've seen are actually UCC caps. This is a good quality Japanese made PSU after all, but not exactly young. So I went ahead and pulled all of the lytics in the powersupply, and checked them each with my ESR meter... guess what? Literally -all- the lytics were out of spec both capacitance and ESR wise...

I'm not sure what the concensus on Jamicon caps is, but they're pretty much the only brand name caps I have readily available around here, so I replaced all of the lytics with Jamicon caps of equivalent rating (higher voltage here and there though... kind of tough to locate 10V rated capacitors). It was the weird ratings that made me want to measure things in the first place... like 6800uF 10V... 820uF 200v.... not exactly easy to get. I actually couldn't get a replacement for the 820uF 200v main filter cap but luckily that one's more or less still okay.

The specs for 12VDC and -5VDC are pretty broad and tolerant as they're mostly only used for audio purposes (except on some very old boards where some Sram uses them as well), it's only the +5VDC that should be between 4.8-5.2 VDC and it's adjustable (always precision adjust it to 4.9V with MM connected). I actually did check that and the voltage was stable and at the preset voltage even when things started acting up.

Anyways, after reassembling, the PSU is -much- better. I actually managed to have a board run stable with 4.7VDC from it after the new caps, which is somewhat of a miracle in itself. Arcade boards generally don't wake up below 4.85 VDC (at least they're not stable).

Cheers!

 

[KrisBlueNZ]

OK, good!

Jamicon seems to be Taiwanese. I have no idea whether they're any good or not. Personally I would replace them with the real thing, but you may want to wait until they start to fail before you do that, I suppose.

Did you check the ESR and ripple current ratings on the replacements?

The main filter cap probably won't be a low ESR one anyway, and they don't normally fail as early as the output capacitors do. I wouldn't bother replacing it unless it measures significantly low on capacitance.

Nice work!

 

[Rattanee]

I didn't bother checking the ratings as I needed this fixed asap. Plus looking up capacitor datasheets always irks me... too much confusion around them most of the time. What matters is that it's working fine finally!