Low Leakage Capacitors Question

[klauseva]

Now I have recently replaced almost all of the electrolytic caps in my AU-20000, (86 in total as I recall from my last count). In this process I noticed, that Sansui on some boards used polarized ELNA caps of low leakage type as coupling caps (in the direct signal path) instead of bi-polar or non-polar caps.

Now for a question or two:

What is the advantage of "low leakage" caps vs sound quality?
Is anybody using "Low Leakage" caps today for audio?

ELNA still has a low leakage series (RLB) available, but their so called Audio capacitors called Starget (discontinued), Silmic I/II and Cerafine, they are not low leakage types.

BRgds Klaus

 

[Mooly]

All modern polarized electrolytics are "low leakage". It's all relative. If you are in any doubt just bring one to it's rated voltage slowly with a sensitive micro ammeter in series.
Data sheets should also give leakage values usually as either a value say 5 micro amps or more commonly as for example I=0.01CV or 3 microamps maximum.

 

[klauseva]

Thank you for your answer. I am well aware that todays capacitors are much better in most ways than the capacitors in the 1970'ies. I do not have data showing the leakage current values for the old orange ELNA low leakage series RL, but data were probably better (relatively) than for the normal capacitors at that time. However, even their low leakage capacitors today are better regarding leakage current, as these data shows from the data sheets for ELNA's capacitors currently on the market:

ELNA SILMIC II and CERAFINE: Less than 0.01CV or 3 whichever is larger (after 5 minutes),
whereas
ELNA RLB: Less than 0.002CV or 0.3 whichever is larger (after 2 minutes),
which is at least 5 times better.

This just makes me curious, if low leakage current were such an important parameter for the quality of the sound for a coupling caps back then, why is it not so today? Furthermore, I would really like to understand what the influence of leakage current is to the sound quality ie, does it have influence on the noise level, distortion, or .... As I understand, Sansui used to call the "low noise" capacitors.

The ELNA RLB datasheet says about the capacitor: "MINIATURE ALUMINUM ELECTROLYTIC CAPACITORS FOR AUDIO", but is anybody using them for audio? I have not been able to locate a place to buy them either, does anybody know where they are for sale?

BRgds Klaus

 

[Mooly]

I quite like the Panasonic ELA series, miniature electroylitics for audio. I wouldn't worry to much about leakage currents, it definitely won't upset the operation of the circuit. Opinion is divided on how "audible" electroylitics are. My own experience is that a good quality part of adequate "value" which of course depends on the impedances it works around is totally transparent.

 

[linuxguru]

Regarding Elna RLB (blue/green lettering on orange sleeve), the range of voltages is limited to 50v maximum, and the available capacitances are only a limited number of small values starting at 1 uF. These are perfectly adequate for use for input coupling, but not large enough for bootstrap, bypass, PSU filtering, output DC blocking, etc.

I have tried the 2.2uF/50v for input coupling, and it is audibly flat and smooth - perfectly adequate for general use, probably much better than an average gritty-sounding polyester film. I would rate it as sonically better than most General-Purpose caps, including Elna RC2, RE2, RE3 and similar lines.

I have not done comparative testing, but my guess is that Elna ROA, ROS, ROD and RFS (Tonerex, Cerafine, Starget, Silmic I/II, etc.) are all better sounding than RLB.

I can confirm that Nichicon Muse SW (subminiature series) is better sounding than Elna RLB - the difference is a rounded mellowness as well as increased transparency. Elna RLB can be bypassed with a suitable small film cap (say 100 nF) to improve its highs, but still does not match the Muse SW for overall sonics. YMMV, but Muse KG, KW and KZ are probably better than Muse SW.

Edit: Low-leakage caps are also useful for accurate setting of input bias voltages using high-resistance, low-current networks - such as those encountered on the classic JLH or similar single-rail topologies.