Overheating Caps

[g1]

I'm not in the habit of feeling the caps, I do look for swelling or I investigate caps when faults lead me to them.
I happened to notice a bunch of hot electrolytic caps in an amp, I think my nose led me to them. They are doing their job, and measure the same as new good caps on my cap meter for value, ESR, etc. Just quite hot to touch.

I'm going to replace them, but wonder what is going on? Maybe it's not that uncommon and nobody bothers to check or comment on it? Or I'm missing something? Close inspection shows the plastic has shrunk a little. These are lower voltage caps, for +/-15V rails and +/-3V DC heater supply. (mesa boogie)

[nosaj]

Cut one open and see if the electrolyte is almost gone. IME dryed out caps get hot.
Are they High uf caps or low ones

nosaj

[g1]

6800uF 10V on the heaters and 1000uF 25V on the +/-15V lines. I'm still running other tests on the amp so won't be able to autopsy til the new ones arrive.

[nosaj]

Well up where your at, any extra heat is a good thing right?
nosaj

[The Dude]

Is it possible there is some high frequency oscillation somehow making its way into the PS? I might hang a scope across a hot filter cap and see.

[Helmholtz]

Either HF oscillation or leakage (doesn't show in ESR).

[g1]

leakage (doesn't show in ESR).

I thought/hoped leakage might at least have an impact on the capacitance value or dissipation readings.

[Mick Bailey]

I'm thinking that a bench meter test may not truly reflect how the cap is operating in circuit. Most meters are operating on very low voltage and high frequency to overcome reactance skewing the results. Leakage can change with increased voltage, as can ESR.

[Helmholtz]

I thought/hoped leakage might at least have an impact on the capacitance value...

Not necessarily.

Capacitance drops with extended use, mostly caused by the electrolyte drying out. Typicall accompanied by increased ESR.

Leakage can be found with relatively new caps, still having full capacitance.
Sometimes re-forming can heal it. Leakage mainly develops when the cap has not been charged over longer periods, as a charged cap permanently re-forms itself.

...or dissipation readings.

No clear what you mean.

ESR is a (low) series resistance, causing dissipation with AC currents.

Leakage corresponds to parallel resistance. Here the dissipation is (mainly) produced by DC current.

[g1]

No clear what you mean.

ESR is a (low) series resistance, causing dissipation with AC currents.

Leakage corresponds to parallel resistance. Here the dissipation is (mainly) produced by DC current.

'D' is one of the measurements on the DER-5000

[Helmholtz]

'D' is one of the measurements on the DER-5000

I see, but the dissipation factor D is an AC parameter. Not sure how sensitive it is to DC leakage.

[Helmholtz]

'D' is one of the measurements on the DER-5000

What D values did you measure at1kHz? Should be 0.1 or lower for quality ecaps.
In my company all design and QC EEs used the Q factor, which is the reciprocal of D for a quick evaluation of AC quality.

[g1]

What D values did you measure at1kHz? Should be 0.1 or lower for quality ecaps.
In my company all design and QC EEs used the Q factor, which is the reciprocal of D for a quick evaluation of AC quality.

The meter does D or Q. I'll run through the measurements again vs the new ones when I get them.

[nevetslab]

Late last year, when I was servicing the Demeter VTMP-2a Mic Preamp, and finding there was a fair amount of hum from the rear panel-mounted power transformer, I measured the AC current feeding the bridge rectifier which was charging the 1000uF/35V cap thru a 2A bridge, all feeding an LM317K regulator set for 12.6VDC to run the heaters of the five tubes in the unit.

There was substantial 120Hz ripple on the heater supply. So, I had replaced both the 1000uF/35V cap with a 4700uF/35V cap and a 4A bridge. The bridge AND the cap was heating up substantially, As the heater current was 5 x 150mA for the five tubes (3-12AX7's and 2-12AU7's), running at 12.6VDC, that excess current had to be the larger filter cap.

Now, that original cap was sandwiched between the heat sink fins of the LM317K Regulator IC and the vertically mounted 2A bridge...both running hot. After finding that, I had installed a 25A Bridge on the chassis wall, so that heat source wasn't in physical contact with the larger 4700uF cap. Primary AC into the bridge was showing 3A Peak currents in both directions, and bridge output feeding the filter cap was showing 3A peak current. I had selected UPW 105 deg C rated Nichicon 4700uF/35V, which at 100kHz, ripple current rating was 3800mA, and at 120Hz, using their 0.85 multiplication factor, that rating is reduced to 3230mA. When I checked the IR temp readings, the cap was measuring 42 deg C, so that did seem to solve that problem. And the hum that I had been measuring was way down, no longer a problem as it had been.

[g1]

One thing I'm suspecting a bit for the +/-15V rails is they are sourcing off the 60V winding and using half-wave rectification. The peaks may be over the 25V rating of the caps. I'm glad I ordered 35V replacements.