Recently built amp with a 10000uf 35vdc Nichicon GU smoothing cap on the power rails. Actually fed regulated 24v from an SMPS so not a lot of ripple to deal with and a small class a design running pretty small bias.
Amp ran for about a couple hours and then what sounded like a leaking radiator.... something under pressure escaping. The room was dark but soon realised the cap was venting 'steam' like a boiling kettle. Quite impressive/alarming. Cool how music still played but guess the system would work without the cap. The singed resistor looks like it does due to being in the firing line of the base of the cap splitting...i think.It does measure out of spec though having removed it. 1.5ohm instead of 0.33 it should have been. This is an RC filter.
So considering new capacitors (eBay!) and we'll within limits; would you consider fake cap, unlucky with a rogue, or other reasons.
And what was the actual mode of failure, can that be ascertained?
Amp ran for about a couple hours and then what sounded like a leaking radiator.... something under pressure escaping. The room was dark but soon realised the cap was venting 'steam' like a boiling kettle. Quite impressive/alarming. Cool how music still played but guess the system would work without the cap. The singed resistor looks like it does due to being in the firing line of the base of the cap splitting...i think.It does measure out of spec though having removed it. 1.5ohm instead of 0.33 it should have been. This is an RC filter.
So considering new capacitors (eBay!) and we'll within limits; would you consider fake cap, unlucky with a rogue, or other reasons.
And what was the actual mode of failure, can that be ascertained?
Rule 1 never buy anything like this from ebay.
Now where did you get these caps from? I'd suspect you had a fake cap which died.
And did you measure the output from the SMPS to ensure it really was 24v?
Now where did you get these caps from? I'd suspect you had a fake cap which died.
And did you measure the output from the SMPS to ensure it really was 24v?
Yes was 24v as needed that to be known for setting the amp up.
Is this a sign of over voltage so maybe the cap isn't actually 35v rated.?
Have bought plenty successfully off eBay but to be fair looking back this was a one off purchase from a seller I hadn't used before and looking at the sellers items now they don't appear to be electronic component specialists!
Is this a sign of over voltage so maybe the cap isn't actually 35v rated.?
Have bought plenty successfully off eBay but to be fair looking back this was a one off purchase from a seller I hadn't used before and looking at the sellers items now they don't appear to be electronic component specialists!
Phloodpants, looks like the resistor was a secondary casualty that got toasted from the exploding capacitor.
Jim, I would agree with previous comments…FAKE cap.
Unfortunately, this scenario might get worse with the way prices are heading of genuine parts from major suppliers like Mouser, DigiKey, Newark, etc….
Jim, I would agree with previous comments…FAKE cap.
Unfortunately, this scenario might get worse with the way prices are heading of genuine parts from major suppliers like Mouser, DigiKey, Newark, etc….
Enlarging the picture shows the resistor was incredibly overheated "from inside", the paint layer cracked and separated, to boot it radiated so much heat that it burnt a hole in capacitor shrink wrap exactly by it side.
No external heat source, not even a blowtorch would have burnt it that way, even less an overheating cap.
I suspect that rail was WAY overvolted for some reason.
Guessing is not enough, MEASURE.
Description is not clear, we need precision:
A schematic would be great.
No external heat source, not even a blowtorch would have burnt it that way, even less an overheating cap.
I suspect that rail was WAY overvolted for some reason.
Guessing is not enough, MEASURE.
Description is not clear, we need precision:
Please answer this.
A schematic would be great.
What are those resistors for, anyway, are they 13k?
Looks like they might be in parallel with the caps.
Looks like they might be in parallel with the caps.
Had that happen on start up, within a minute, reversed capacitor. Not after hours.
Check voltages again, and buy 50 or even 63V rated caps.
And check circuit before power up.
Use dim bulb method at first power up.
Check voltages again, and buy 50 or even 63V rated caps.
And check circuit before power up.
Use dim bulb method at first power up.
Fake cap for sure but still wonder why the other fakes didn’t blow too. Real caps have that x on top so if they do blow they predictably split. The question is why did the resistor nuke too? Definitely need a schematic to see what’s going on.
Ground is in the centre , caps are oriented correctly . Pads are at 45 degrees.
That isn't a bleed resistor. It is part of the rail RC filter. Smps at 24v.
Single rails. Stereo amp so the PCB has a seperate RC for each channel.
See comments in first post....0.33ohm in an RC filter .
Schematic attached. Rx10 and Cx3. I used 10mf as my PSU is fairly ripple free to begin with.
Shall remove offending cap and report back.
The cap does have the '+' embossed in the top, albeit behind the black plastic disc .
Why the other RC on the other channel is ok I don't know. Need to do some more exploratory work .
The amp was still happily playing music nicely through all of this. So I think it was solely the power rail filtering that was affected. I guess as an amp, unless Rx10 was open it will still work, and I am only filtering the rail after a pretty good Tracopower SMPS. Ultimately I could short Rx10 and omit Cx3 and it would still work.
The amp was on the bench for 2 hours to set bias etc the day before, and then 2 hours the day after with music playing.
Last edited:
Reminds me when I was studying City and Guilds electronics on a course.
We all had to build a small amplifier on strip board.
Out of the 10 of us about half had exploding electrolytics.
Each time one blew up with a loud crack a cheer would go up.
The course was at Durham, UK. It is quite close to Geordie Newcastle.
One of the lads could never remember which x/y axis was which.
So I suggested he remember "Why aye" which is a Geordie saying or Y high on the graph.
Great anecdote Nigel and I liked the picture you attached. An interesting capacitor indeed.!
Why aye man !
Why aye man !
Cap removed and cleaned. ...and measured. It appears to be 'ok'! Now it's all cleaned up it doesn't appear to have fractured.OK it's far from OK and won't be used again but it is still measuring as a capacitor . So maybe all the excitement was down to the resistor afterall as mentioned by some here. Could the heat from the R have overheated the cap to make it swell ?it measures out of spec but then the other one does . Good one was about 6000uf @120hz. The 'bad' one was a bit over 4kuf with higher DCR and ESR but not massively so.
So maybe the caps are old stock and have lost some capacitance or are maybe fakes but I do think the failure mode was the resistor now. Further tests to continue.
So maybe the caps are old stock and have lost some capacitance or are maybe fakes but I do think the failure mode was the resistor now. Further tests to continue.