Old electrolytic capacitors, test higher than rated value?

[lucym202]

Hi,

I'm currently working on a Sony TC-102, it's a reel-to-reel tape machine, circa 1964 with valves/tubes in the power amplifier. Currently it works but sometimes the audio goes quiet after a few minutes of operation. I've ordered a new set of valves and am awaiting their arrival.

In the mean time I decided to check the electrolytic capacitors. I desoldered each of them, checked them with a T7 LCR tester, and replaced the ones that were well out of their rated capacitance, reinstalled the rest.

What seems weird to me is that the ones that differ from their rated value are somewhat higher. A 50uF read as 79.16uF, a 25uF read as 46.30uF, and a 1uF reads as 2056nF (2uF)

I thought old capacitors generally lost capacitance, but these appeared to have gained some? Or is this within original tolerance for caps this age?

Most of the capacitors on the board tested in the order of their related value as did the ones I used for replacements, so I think the LCR tester is probably accurate enough.

Caps are all ELNA brand. Tape machine now seems to be working about as well as it did before the recap.

[colorburst]

ESR and/or leakage went up, so the caps take longer to charge and the meter interprets that as higher capacitance.

[blue_lateral]

Two things, I am not familiar with a T7 LCR tester, but if you were using anything less than an oldschool capacitance bridge like a Sprague Tel-Ohmike or similar, leakage reads as extra capacitance just like Colorburst said. The other thing is that electrolytics back then at their absolute best were +-20% for tolerance. More common was +80% -20%, and +100% -20% wasn't unheard of.

Tubes (valves) are unlikely to cause the problem you describe, and indeed are hardly ever bad compared to capacitors. Have you replaced the coupling caps or just the electrolytics? Any capacitor from a tube plate to a tube grid needs very close scrutiny. The gray plastic clad (paper?) capacitors often used in that spot in Japanese equipment were truly terrible, and even worse than the American ones. Good luck with the project.

[Kleinstein]

It is normal that after longer time of non usage the capacitance of electrolytic capacitors goes up. This is from the oxide layer slowly degrading. When used for some time the layer reforms and the capacitance will go down. This is not just the higher leakage that some LCR meter may interprete as higher capacitance. The higher leakage is part of the reforming of the oxide layer.

[srb1954]

What seems weird to me is that the ones that differ from their rated value are somewhat higher. A 50uF read as 79.16uF, a 25uF read as 46.30uF, and a 1uF reads as 2056nF (2uF)

I thought old capacitors generally lost capacitance, but these appeared to have gained some? Or is this within original tolerance for caps this age?

Electrolytic capacitors of this vintage typically had a tolerance of -20% +80% as the manufacturing tolerances were pretty loose back in those days. The 50uF cap could still be within its original tolerance at 79.16uf but the other two caps are outside an acceptable tolerance.

The excess capacitance could be due to the capacitors not being biased for an extended period. The aluminium oxide dielectric layer tends to erode away a little with time and, with a thinner dielectric layer, the capacitance goes up. The oxide layer will be replenished if the electros are biased at their rated working voltage for a suitable period of time. The oxide layer will thicken up a little and the capacitance will drop again.

[mawyatt]

@ OP

If you have some time and want to investigate more, consider measuring the Caps with a proper LCR Bridge meter with and without DC bias. Believe (with DC Bias) this is how they were evaluated back then with a low frequency AC waveform that didn't become negative (DC Bias). Maybe see if you can find details on how these old Caps were characterized and try and duplicate that setup.

Then bias the Cap up with a Current Limited DC Supply to ~ rated voltage for a few days and reevaluate.

Would be interesting to "see" how the "rejuvenation" effects the results 

Anyway, good luck with your restoration project

Best

[Geoff-AU]

I thought old capacitors generally lost capacitance, but these appeared to have gained some? Or is this within original tolerance for caps this age?

Kleinstein's answer is spot on.  Capacitance dropping is from electrolyte loss/degradation, capacitance increase is from oxide layer degradation (which, incidentally, reduces the voltage rating of the capacitor).  As Kleinstein also said you can "reform" the oxide layer by applying voltage, at manufacturing time the forming voltage is higher than the rated voltage (current-limited, no doubt).  Figure 4 (plus the surrounding text) in the attached TDK tech note has more information