• Please review our updated Terms and Rules here

What generally smokes in an old power supply.

U

Unknown_K

Veteran Member
Joined
Sep 11, 2003
Messages
9,094
Location
Ohio/USA
I was digging for something on a shelf the other day and noticed I had an old boxed Powmac LC-6100xs supply on the shelf I forgot about. While testing it with my newer PS tester that shows voltage I noticed the 5v and 12v were a little lower than they should be and then came the magic smoke.

Usually when a power chip goes you hear a snap or pop and the thing turns off which didn't happen.

I took the supply apart and noticed some bulging caps (a bunch of 2200uf 10v). Can't seem to see what would have smoked, the supply looks new inside and out. Do venting caps actually smoke? Their values were WAY off with something like 40% voltage drop on my tester. Could it be a large power resister?

This is an old Athlon era supply from not the best maker and the only reason I would have kept it is because it is 5V heavy unlike today's supplies.
 
I had a core memory system cap go one time. It scared me how loud it was. And you could see it burning down inside. So Yes, they can smoke.
 
It varies. Most would first assume the filter capacitors but that isn't always true. I've had 2 of the old line filters smoke. I'd say it is often enough that one might just replace them before they smoke up the house. I've had one large filter capacitor fail on a linear supply, blow. On linears, I'd say the most common failure I've seen was rectifier diodes.
Switchers are all over the place. I've seen a pulse transformer fail and the optical couplers used on many are a common failure point. High side filters tend to suffer from high internal resistance, on switchers. There have been a number of other single point failures but as a hobbyist, I don't see common problems too often. I always replace filter capacitors that leak before they can cause other damage.
I don't replace things that don't need to be replaced, like I see many doing. It can often cause more damage than good. I trouble shoot the failure rather than shot gun the failure.
Dwight
 
Capacitors, Transistors, sometimes the transformer smokes as the coils short and/or break. I've seen fuses smoke when they blow. Lots of stuff. I smoked 2 230K Ohm high amperage Resistors in my VersaDock a month ago, here's plenty of things that can do that.
 
Yes a cap can smoke... and even explode. Oil filled caps are REALLY bad actors when they go. Generally, electrolytic caps are the first non-silicon item to go, and for several reasons: 1) voltage rating is too low (must be twice or greater the peak operating voltage), and/or the 2) temperature environment at or exceeding the cap's rating, and/or 3) the P/S caps haven't been used for a long time (the electrolyte dries up/goo's up/separates, etc so in order to power up an OLD electronic device with one or more power supplies, the P/S caps MUST BE reformed before use. THAT, requires a variac and **staged voltage increases over time** in order to not have the caps 'go' on a raw power up) . Any electrolytic cap that is stained, discolored, bulging (even SLIGHTLY! on the top or sides), etc must be replaced. Unfortunately when filter or other electrolytic caps fail the result can be higher DC or even "ac" waveforms hitting silicon based items - which can weaken/damage/or destroy them. And then sometimes you luck out and nothing else gets hurt.
 
Well I finally got around to replacing the bad capacitors. The PS seems to provide the correct voltages (12V is slightly under, 5V is slightly over, 3.x is exact) with no smoke to be seen.

Odd thing is one of the cooling fans is not spinning. There are 2 fans with one having a lead to the motherboard to be temperature controlled from the looks of it (3 wire).
 
Most modern power supplies have "X" and "Y" capacitors to limit the amount of noise from the power supply going out through the power cord. These capacitors are usually connected line-to-neutral and line-to-chassis-ground. These capacitors get stressed because of the high AC voltage across them and are prone to fail when the get old. The most common manufacturer is RIFA. When a RIFA capacitor fails it lets out an enormous amount of really stinky smoke.
 
Yea, I have seen those square capacitors before, but the ones I replaced were standard caps you would find on a motherboard. I haven't tested the supply on a motherboard yet (winter lazyness).
 
The two biggies I run into are "dry" capacitors where due to extreme age the electrolyte has almost entirely evaporated or has leaked out. The other is the electrolyte has internally corroded out the plate contacts and its physically disconnected from the circuit.

The results are, in no particular order:

-The dry capacitor doesn't have any characteristics of a capacitor and the PSU will shutdown from no inrush load compensation/signal filtering
-The dry capacitor will short internally and become a resistor. This will cause weird phantom DC rail loading/no signal filtering until either removed or the capacitor thermally burns up (possibly exploding)
-The leaked electrolyte will cause weird resistance bridges between components, carbon tracking and corrosion (United Nippon Chemi-con and Nichicon are bad for this. 20 years ago they used to be fairly well regarded brands too....)
-A capacitor that is internally corroded itself open will likewise not act like a capacitor at all

That last one's a fun one because I have seen 'lytic capacitors of all types exhibit this. The cap visually looks fine. It might not even be 10 years old. Yet if you test it with most capacitor checkers it fails right away. Had a few jobs where I was told "I looked it over and everything seems fine" and found something like an open 1mfd 50v cap on a MOSFET controller's feedback pin.

...and then of course you still have the *classic* bulged late 90's, early 2000's capacitors where it can be any of the above, plus weird filtering characteristics that cause uneven noise filtering, component whine and induction burn-outs. At least those are obvious. :p

The AC input caps are kinda the reddit meme right now
Something not working? Probably caps
PSU smoking? Gotta be the RIFA's
 
Last edited:
I want to add to that, working from 20 years of recapping motherboards, power supplies, monitors and other general purpose and professional electronics (technically I'm at that $JOB right now ), if you are looking to service something to put back into reliable/full-time service or you are being *PAID* especially to refurbish as a service, do not assume the capacitors are fine regardless of visual condition or age. Never tell the customer that the capacitors might be fine. That is up to you to evaluate. You are not Dave Jones. You are not Shango066. The first thing cheapskates will skimp on is the cost of a full recap because caps are expensive (and technically the labor charge is more but hey, you just paid $500 for a PVM and you also just balked at the $60 DigiKey cap order because "eh, well guess it's not that bad. I think I'll pass on the recap for now.")
 
Last edited:
There are people who replace just what is broken and those who replace all capacitors even when that is not the issue.

Since this is a hobby I tend to replace what is broken and anything else I see as fishy or likely to cause me issues down the road. On motherboards if a couple caps are leaky I replace all of the ones of that type in the area (usually the ones around the CPU heatsink). On 68K macs all aluminum SMT caps get replaced since they all leak anyway but not the axial ones since they seem to test fine.
 
You must log in or register to reply here.
Back
Top