Another definitive advice on capacitor distortion.

[JohnRoberts]

I won't rehash the semantics but my last serious investigation into failure/reliability was while working at Peavey last century when we decided to extend our warranty (from something like 3 years to 5 years). We did an extensive review of warranty claims and repairs to look for trends. Surprisingly we found no sudden increase in repair claims in the 3-5 years time period.

I did find one SKU with unusually high failures proportionate to its sales volume. Looking into that one SKU more closely revealed a design flaw that we then corrected. It was below everyone's radar because it was a modest seller.

The result of our investigation was we extended our warranty another two years with no negative consequences or regrets.

===

Over the years dealing with literally truckloads of capacitors I saw some variability between vendors with better and worse. It worth avoiding the cheapest parts and most established brands are serviceable.

JR

 

[vintageampfixer]

Over the years dealing with literally truckloads of capacitors I saw some variability between vendors with better and worse. It worth avoiding the cheapest parts and most established brands are serviceable.

My experience is similar: electrolytics rated so they operate within specified temperature and working voltage parameters (and occasionally ripple current), tend to last far longer than MTBF stats would lead us to believe ... with the possible exception of tantalum bead which seem less reliable although nothing like as bad as people suggest. I've regularly seen kit that's 25, even 40 years old, still quite operating quite happily with original components. Without naming names, I don't think any of those manufacturers used particularly expensive or "high quality" capacitors.

On the other hand, I'm now seeing modern kit, mainly SMPS modules and Class D amplifiers with failed capacitors. It feels as though designers are (perhaps intentionally) running components far closer to their limits - or even beyond tolerance? Or possibly, the demands placed on those circuits aren't properly considered ... even ignored? Clearly not everywhere ... there's a lot of phones out there with Class D amps in them that work fine and a similar number of laptop PSUs which don't seem to fall victim...

 

[JohnRoberts]

My experience is similar: electrolytics rated so they operate within specified temperature and working voltage parameters (and occasionally ripple current), tend to last far longer than MTBF stats would lead us to believe ... with the possible exception of tantalum bead which seem less reliable although nothing like as bad as people suggest. I've regularly seen kit that's 25, even 40 years old, still quite operating quite happily with original components. Without naming names, I don't think any of those manufacturers used particularly expensive or "high quality" capacitors.

On the other hand, I'm now seeing modern kit, mainly SMPS modules and Class D amplifiers with failed capacitors. It feels as though designers are (perhaps intentionally) running components far closer to their limits - or even beyond tolerance? Or possibly, the demands placed on those circuits aren't properly considered ... even ignored? Clearly not everywhere ... there's a lot of phones out there with Class D amps in them that work fine and a similar number of laptop PSUs which don't seem to fall victim...

I've shared this story before, but back at my old day job I knew an amp designer who played fast and loose with capacitor breakdown voltage ratings. In a big dog analog audio power amp sizing the reservoir capacitors conservatively for high line mains voltage could bump BOM costs a few dollars. I knew one amp design engineer who ASSumed that capacitor manufacturers specified their max voltage ratings conservatively, so he could cheat a few volts here and there. Likewise there may be a capacitor manufacturer out there who ASSumed that the design engineers would be conservative and give him a few volts of safety margin (I would). Of course this engineer screwed up and encountered much higher than expected field failures (or why else would I share this story).

JR

 

[Matt Syson]

Of course the 'failure' rate of electrolytic capacitors does not say WHAT the failure is, be it below a certain percentage of the capacitance value marked on it OR failure such as leakage, smoke etc. Saying a piece of gear 'still works' after 30 years covers a range of possibilities and it is unlikely to be working to the same sonic performance as a freshly manufactured unit. A nominal 100 uF cap in an audio circuit will start to reduce it's capacitance over a number of years and it is position(in the circuit) (the impedance it is coupling into) and signal level that will define how faulty it is.

 

[vintageampfixer]

Of course the 'failure' rate of electrolytic capacitors does not say WHAT the failure is, be it below a certain percentage of the capacitance value marked on it OR failure such as leakage, smoke etc. Saying a piece of gear 'still works' after 30 years covers a range of possibilities and it is unlikely to be working to the same sonic performance as a freshly manufactured unit. A nominal 100 uF cap in an audio circuit will start to reduce it's capacitance over a number of years and it is position(in the circuit) (the impedance it is coupling into) and signal level that will define how faulty it is.

Electrolytic capacitors are usually around 20% tolerance (at best) when new; most of the new ones I've checked tend to show measured values towards the lower end of that 20% then, with time, the electrolytes dry, further reducing reduce their capacitance. As you say, it's a trade off between what's still clinging-on-and-just-about-working versus the performance you expect from the circuit. We humans are pretty fickle and we'll continue to believe a favoured bit of kit still works well after n years even though it may need "recapping" (to use the modern phrase) which is the aural equivalent of rose tinted glasses!

 

[Gold]

I’ve said this before but in the Neumann VG66 cutting amplifier rack I’ve never found a bad big PSU cap. These were new from the mid 1960’s through the early 1970’s.

The psu modules are simple series pass transistor designs with a huge amount capacitance at the output. Caps mounted in the frame not in the module. I am just finishing up a VG66 now. All of the caps measured above the rated value with unmeasurable ESR and leakage.