Audio
equipment makes large use of electrolytic capacitors. Commonly, these will
be doing one of three things:
Power supply smoothing (capacitors used for this are
commonly called reservoirs)
DC
coupling (keeping DC bias voltage within a circuit
but letting AC signals such as sound through)
Frequency control (usually audio uses smaller
value capacitors for this but some things such as
loudspeaker crossovers will often use electrolytics)
You can
think of electrolytics as a necessary evil, they are
inferior to other types of capacitor but they can pack a
far higher charge into a smaller case (polypropylene
capacitors in comparison are almost perfect for audio
but are at least 10 times the size of electrolytics for the
same value and far more expensive); this compact size is
what you need when fitting dozens into a piece of audio
equipment.
 |
|
New
Electrolytics in a Rotel RA-820
|
To get
this large increase in value, the plates which carry the
charge inside the can are etched to increase surface
area and are insulated by a thin layer of aluminium
oxide with an electrolyte fluid filling the void between
the two; the layer of oxide means that the capacitor can
only pass voltage in one direction, reverse this voltage
and the oxide layer will quickly fail, effectively
creating a short circuit, and, depending on the current
available in the circuit, heating up the electrolyte -
sometimes with explosive results if the build up of
pressure inside the can is high enough to rupture it.
This
oxide layer also relies on the working voltage the
capacitor usually sees to keep it topped up, if that
voltage is switched off for a prolonged period of time
the oxide gradually dissipates. Another reason why it's
not such a good idea to suddenly apply full power to
that amp which has been sitting in a box in the attic
for the past 15 years.
Testing
ESR
Some time
ago I bought a tester to measure ESR (equivalent series
resistance). This tester measures resistance across the
capacitor at a very high frequency (100kHz), a frequency
that an ideal capacitor should be virtually short
circuit at, and at which a failing capacitor will find
it very difficult to function at and will hence present
a large resistance to the tester - simple 1 / 2π*F*C will
return an XC (capacitive reactance) of virtually zero
for anything in the realms of microfarads, which is what
an ideal capacitor should be showing - anything above
this is down to series resistances inside the capacitor,
making ESR a good health indicator.
An ideal capacitor is just that - a component which will
exhibit a reactance (AC resistance) which is inversely
proportional to frequency, with no
other factors mixed in. In reality, a capacitor will
'look' (in an electronic sense) like a series LCR network -
the actual capacitance (C) will be mixed with a small
amount of series inductance (L) - from connecting leads
etc - and also a series resistance (R) - and it's the
series R which we're interested in when we talk about
ESR.
ESR only really becomes an issue with electrolytics -
one factor is that the liquid which forms the
electrolyte forms a resistance, especially as it ages -
this resistance can behave in a non-linear way with
regards to voltage level and frequency and also has the
effect of heating up the inside of the capacitor,
further accelerating the ageing effect.
The ESR tester doesn't claim to be an exhaustive test of capacitor specifications and
the results won't be of any use to compare to
manufacturer's specifications (different manufacturers
often test at completely different frequencies and
voltages etc and to mimic this would need a signal
generator, power supply, volt meter etc.) but the
handheld tester is still a convenient way to perform
a basic health check on them while you're working. This
tester won't actually tell you that a part
is bad, it'll just dutifully give you the capacitance
value and an ESR reading in ohms and then
it's up to you to decide.
 |
|
Testing
ESR on a set of reservoir capacitors |
The two
things here are to check the actual capacitance value -
this should be near the value stated on the can, perhaps
within +20%/-10% for good quality ones; in my
experience, it's not unknown
for capacitors to go up in value as they age. The second
is the actual ESR value as we've already seen. This isn't quite so clean cut
as it seems at first and depends on a whole slew of factors, mainly the relative
size/voltage rating of the capacitor and the quality of
the part (cheap general purpose will likely have a far
higher ESR than high quality parts).
A general rule when
using a tester similar to the Peak ESR is
less than 0.5Ω for the smallest parts, falling to around
0.01Ω or lower for large power supply capacitors. 0.01Ω
is the lowest the meter can read so most larger
capacitors will bottom out at this value, if in doubt a
check with a known good capacitor of similar
voltage/value helps.
Armed
with this meter, I started testing random parts of
differing ages, including some which I'd removed from
vintage audio equipment over time. Most of the higher
grade Japanese parts tested well within spec, usually
still better than brand new 'general purpose' parts
which would usually be what many people would replace
them with. On the few which had shown signs of stress
(overheating or bulging) it was a different story but
overall around 90% of them have tested well within
reasonable limits so far.
On the
complete opposite side was a 3 year old computer power
supply. This had been used regularly but to my knowledge
had always kept cool and hadn't overheated, it had
started causing some symptoms with voltage drops and
random computer restarts so was replaced. Most of the
electrolytics showed signs of drying out and bulging, a handful
looked like they had been leaking for some time. The two
main power filter caps were fine but most of the smaller
ones which were unknown Eastern brands had suffered
badly, most tested way below spec with ESR in the tens
of ohms, two had gone totally open circuit - this would
cause major issues on a switched PSU and quite possibly
be dangerous. Again, this is the quality factor - you
get out what you put in. That power supply had two large
fans and probably ran barely warmer than most vintage
receivers, probably never saw anything over 35°C yet the
parts had struggled with 3 years of normal use.
I always
pull a few electrolytics at random and test them, if they are
well within spec I'll leave them in until next time -
this is the case 95% of the time, again it depends on
what temperatures they would have seen during their
life, what voltage level they would have been working at
etc.
Capacitor
quality and lifespan
Look
around online and you'll likely see many horror
stories about old electrolytics, either leaking
corrosive 'alien blood' over everything else around them
or exploding and destroying entire PCBs.
Some people believe that these components are a
ticking time bomb and that you're playing a game of
Russian roulette by not renewing them. There are a few
things to understand here in order to dispel part of the
myth.
First of
all, it's not entirely a myth. Things can go wrong, and
sometimes they do. However, it's not entirely clean cut,
there are other factors which determine how reliable a
part is likely to prove and how long it is likely to
last.
Secondly,
quality is a large factor in all of this. In the good
old days, most high end Japanese Hi-Fi was actually
built in Japan. The build quality was very high, the
same went for the majority of parts these companies
fitted to their products. Compare this to modern
consumer grade (lower end) electronic gear built in
places like China, compare the sheer
difference in quality and reliability/lifespan. The
capacitors you'll find in a good vintage amplifier will
be better than those in a cheap modern audio system or
DVD player by an order of magnitude.
Thirdly,
temperature and voltage. As in the paragraphs above,
these are two things which will work a capacitor harder
and shorten its life. In theory, keep a capacitor cool
and its life expectancy will increase exponentially -
although this rule can only be taken so far, keeping
everything cool will make everything inside the case a
lot happier. All electrolytics are marked with a maximum
working voltage and many with a maximum working
temperature. Run just under these limits and it
certainly won't last forever.
Again,
this isn't trying to make the assumption that people are
wrong to say these parts wear out, they do. Just not to
the extent that some would have you believe. A full
rebuild including new electrolytics is a large undertaking and can get expensive,
especially if you source good quality parts. Sometimes
it does become necessary, but on average with good
quality gear, most of the time it isn't, at least not
necessary to keep the equipment working; replacing parts
on the basis of improving sound quality is another
matter entirely, there are definitely sonic improvements
to be had with new components (especially those in the
signal lines) - but that's outside the scope of this
article.
Ripple
current is yet another factor which you need to be
careful of, although this should only matter where large
amounts of power are involved (power supply reservoir
and loudspeaker decoupling capacitors etc). Ripple
current is the charge entering and leaving the capacitor
and is notoriously difficult to calculate in theory (at
least it is to me, I've got reference books here with
long equations to calculate it but maths really isn't my
strong point).
Capacitor
quality and relative size has a major influence on the
maximum amount of ripple current which that component
can stand (the type of connection is also a major
factor, the reason why some high end audio uses power
supply caps with lug connections - the type I prefer to
use when doing restorations). Exceeding the current rating will, at the very
least, shorten the life of the component, perhaps even
affecting the operation of the circuit. That's another
reason why you need to use good quality parts -
especially where large power levels are involved; cheap
general purpose components might look fine on paper but
probably wouldn't stand up well to being used in a
high stress environment such as the smoothing circuits on a power
amplifier where they had a high peak current flow
demand placed upon them.
 |
|
Inspecting a pair of Quad 303 power amps |
In the image to the left is a pair of Quad 303 power
amps - the one on the left shows Quad's later preferred
method of placing the terminals upwards; I'd imagine to
prevent bad caps venting electrolytic goo over the amp
driver boards. Notice that the blue capacitors have all leaked to some
extent, with crystals growing out of one.
Strangely they all tested OK on my LCR and ESR meters but
I certainly wouldn't want to power them up in that
condition.
Conclusion
So, to sum everything up:
- Never run an electrolytic near its voltage rating, the
more 'headroom' you can allow, the better. A higher
voltage part will be superior in most respects, for
example ESR will be lower.
- Never allow an electrolytic to become reverse biased -
if there's a possibility that this may happen, either
use a different type of capacitor or make a bipolar
electrolytic by connecting two capacitors of twice the
value back to back in series.
- Always make sure electrolytics are kept as cool as
possible.
- If you replace any components, quality is key - use a
good quality part from a known manufacturer.
Back to amplifier
rebuild main page
