Coupling Capacitors - Bipolar or Polarised?

[Narks]

I need to replace the audio coupling capacitors between the various stages of a Luxman Pre-amp.

Should I be using Bi-polar Electrolytic caps, or Polarised?

The original caps are "Elna for Audio", but plenty of caps are starting to fail in this amp, so I want to replace them.

[indianajo]

If the caps are in sizes greater than 1 microfarad "mf" (uf in post 1980 typography) and have a plus on one end or a stripe with a minus near one terminal, then buy polarized electrolytics. However, in 1,2,3,5, 10 mf, film or ceramic caps would never have to be replaced again. Film and ceramic are non-polar inherently and will work fine unipolar too. For capacitors greater than 1 mf and no plus, buy non-polar electrolytic if greater than 10 mf. For capacitors smaller than 1 mf, usually a line means the outer wrap is attached to that terminal, which is important in radio and teevee but not audio, particularly. These capacitors should not be electrolytic, and unless paper dielectric (before 1967?) should not need replacing unless you want to tweak your sound slightly.
John Curl doesn't like ceramic caps for coupling because of distortion, but I have gotten away with it buy buying 50v rated caps for 1 v signal applications, which I did because organs produce static and need the high voltage rating sometimes. I find 10 uf 50V aerovox ceramic caps sound fine on 1v signals. They are really tiny. Ceramic manufacturers run in quality down to just above dirt, since computer boards use so many of them, so brand is probably important in ceramic. John Curl like polyprophylene dielectric, and electrolytic, and he probably enjoys changing the electrolytic every 20 years. If you buy surplus or auction electrolytics, you can enjoy changing them out again next month. Buy recently manufactured 2000 hour rated or up new electrolytics if you buy electrolytic at all. The hour life rating is in the selector tables of upscale distributors like newark/farnell. Higher rated life electrolytic caps have better rubber seals.

[jcx]

most often today uF is used for micro, "u" standing in for the greek letter "mu"

SI prefix m is milli: 1 mF= 1000 uF may be confusing given the older practice of "mfd" being used for micro Farad, you often see values like "10,000 mfd" on PS caps - you really should know the value to within 3 orders of magnitude from circuit function and case size

where you need more C than can be easily fit in film, non-polar Al electros are the next choice based on measureable distortion for audio signal functions

Power supply caps are usually polar electrolytics and there's no obvious engineering reason that non-polar parts would work any better in PS

[indianajo]

SI units are interesting, if you are french. I never saw a french amp for sale. Many cap manufacturers marked the same way until the academics or cost cutters got a hold of it about 1980, now nobody knows what "m", "k" or "n" mean anymore. I just bought some 0.1 microfarad 630 V capacitors 2009 manufacture, marked "1 K 630H". What idiocy. At least newark put a lable on the package indicating in complete english language what they were selling. u is not a greek micro, but people born after 1980 think it is.

[thaumaturge]

Rule of thumb in coupling caps is that polarity gets pointed most likely direction of source. e.g. coming off a plate leading to next stage in tube amp plus side of cap would get pointed towards the plate. Same basic idea in a SS amp. Point the cap in direction of likely voltage offsets. Doc

[DF96]

Originally Posted by indianajo

SI units are interesting, if you are french.

SI units are essential. "m" means milli (10^-3), "k" means kilo (10^3), "n" means nano (10^-9). "K" is sometimes used when people mean "k", but it ought to only used for absolute temperature. "F" means farad. "f" means femto (10^-15?), but people sometimes mistakenly use it when they mean F.

"u" is not a greek micro, but as jcx said it often stands in for it as it saves using a greek character set. This is a useful convention, but should be avoided in formal printed matter.

Component marking can be a minefield, but component values are straightforward if we all use SI.