Capacitor ESR, how much is too much?

[olddawg]

I think the capacitor plague was responsible for a lot of the younger techs now wanting to do total recaps on everything as routine maintenance. We never used to assume that a 10 or 15 year old unit needed all new caps, now it seems to be the accepted idea.

I only did that on modern equipment with smd caps. But I see your point!

[Mick Bailey]

This doesn't always hold true, but generally if there are one or two defective caps (high ESR and/or yellow, brown or grey crystals) then they'll all go the same way. If none are bad I leave them alone. Maybe batches and manufacturer come into it, even if one board looks like the next. I sometimes get boards where every cap is bad and this makes me wonder at what point did the owner notice a fault and how long it got used for. Some of the equipment is recent manufacture (I had a run of amps from between 2011 and 2015 with faulty boards), so post-dates the original cap plague.

Cap replacement in low-voltage and SS equipment used to be restricted to the odd one, but now I get equipment where there are serious problems with caps - almost always SMD. Three year-old mixer amps that need upwards of 60 caps. Takemine preamps that need every one replacing. That isn't right. Older equipment with miniature leaded electrolytics can be forgiven - some of these are now 30 or more years old.

One thing I have got pretty good at is SMD cap replacement.

[R.G.]

My personal take on ESR meters is that they ought to be called "ESR Indicators", partly because the method of measuring ESR is tricky, and partly because of the nature of ESR in general. The word "meter" implies that there is some kind-of-accurate reading to be had, maybe in the range of a few percent in error from the known-only-to-the-deity real number. This is really hard with ESR because the nature of the measurement requires making an estimate of the real number based on a fleeting pulse of current. Timing, sampling and estimation are needed.

Then there's the issue of what ESR does for/to you. The idea of ESR is that the distributed resistance messes up the capacitances that you paid for. For electrolytic caps, you mostly want one of three things as regards ESR: (1) low resistance for a gain-control resistor, as in a cathode or emitter bypass, (2) low phase shift if you're incorrectly using an electro for audio filtering, or (3) low "feedthrough" of current pulses in rectifier/filter power setups. The first two uses may not make any difference at all, as there is often another resistor involved that is much larger than the ESR unless the cap is about to die. In these cases, the exact value of ESR is almost inconsequential. You only need to know if the ESR is skyrocketing or not as an indicator that the cap is about to die. For the third, The only good number for ESR is zero. ESR in power filtering situations causes a spike of voltage across the cap that the cap's value cannot reduce, so there is a spiky ripple in the DC voltage the cap makes. Worse, power filter caps have ***BIG*** ripple currents running through them, The ESR causes internal self heating that can make the cap actually cook itself to death. We're not used to thinking of power filter caps as needing cooling, but in high power situations, they do. Fortunately, this is also a situation where you need to know whether the ESR is stable, but an annoyance, or skyrocketing and an indication that the cap is about to go the great capacitor chassis in the sky.

So I reconciled myself to using an ESR meter as an indicator, either "hmmm, I gusess that's OK" or "YIKES!! She's about to blow!!"