This is only a preview of the April 2025 issue of Practical Electronics. You can view 0 of the 80 pages in the full issue. Articles in this series:
Articles in this series:
Articles in this series:
Articles in this series:
Articles in this series:
Articles in this series:
Articles in this series:
Articles in this series:
|
AUDIO
OUT
AUDIO OUT
L
R
By Jake Rothman
Using Capacitors for HiFi, Part 2
T
his series started last month by
explaining how suitable different
types of capacitor are for audio
use. In this second and final instalment,
we'll continue that investigation,
concentrating on some of the more
unusual types of capacitors for HiFi,
and give a ranking of all the major types.
Non-polarised solid tantalum capacitors
exist, but don’t be fooled, they are just
two standard tantalum capacitors in one
case (see Fig.24). I sometimes solder a
bias wire to the case.
I have also seen capacitors wired in
inverse parallel. Of course, this does not
work well with voltages above 1V either
way, but it can be useful for blocking very
low offset voltages while maintaining
a high capacitance. The distortion
performance is shown in Fig.26; it's
about half that of a single capacitor of
the same type.
When using back-to-back polarised
capacitors, you might expect the
capacitance of the pair to be half that
of an individual unit, but that is not
necessarily so. Depending on the biasing
and signal level, one capacitor may be
acting like a diode at any time if it is
sufficiently reverse-biased. Switching
in and out of that mode can introduce
extra distortion.
This is mainly a problem with series
back-to-back polarised capacitors in long
time-constant integrators, such as LFOs
in sweep generators. One way to avoid
this is to provide a centre-bias to the
junction of the capacitors. With audio
coupling, where there is little voltage
drop across the pair, the capacitors will
act more like you would expect.
The general consensus appears to be
if the frequency is above 1Hz and the
voltage across the composite bipolar
capacitor is kept below about 10% of
the rated voltage, there is no capacitance
doubling.
Bypassing
It’s often said by HiFi buffs that
bypassing an electrolytic capacitor with
Fig.24: a bipolar tantalum capacitor.
I was shocked to find it was just two
standard polarised units in one case!
INPUT
22mF 22mF
10V
10V
TANT. TANT.
OUTPUT
150W
LOAD
470nF
63V
POLYESTER
Fig.25: bypassing two back-to-back
tantalums with a 470nF polyester cap.
Total Harmonic Distortion (%)
0.5
0.2
0.1
0.05
0.02
0.01
.005
.002
.001
.0005
.0002
.0001
20
50
100
200
500
1k
Frequency (Hz)
5k
2k
R1
47kW
R2
470W
POWER
AMP
C3
2200mF +
47kW
RF CURRENT
FEEDBACK
C2
100mF
8W
R3
0.22W
Fig.27: this power amplifier circuit
applies feedback around the output
electrolytic, linearising it.
48
INPUT 1.5mF
V+
–
20k
Fig.26: the distortion curve of a reverse parallel network (two 47µF 10V caps). It
isn’t as bad as expected, seeing as they were being thrashed at 6V RMS.
V+ (74V)
INPUT C1
10k
33kW
POWER
AMP
8W
4700mF
+
80V
½ V+
+50V
8W
1000mF
+
40V
INPUT
TRANSISTOR POWER
Q1
AMP
C
INPUT 1mF
B
22kW
33kW
10mF
50V
47kW
3.3mF
FILM
–
3.9kW
330W
(SEE AUDIO OUT,
MARCH 2018)
Fig.28: low distortion can be obtained
using a standard 4700µF 80V output
capacitor in the amp’s feedback loop.
47kW
FILT.
E V+
C
Q2
1MW
22kW
ACFB
B
E
–
DCFB
CURRENT
BOOSTER
1mF
FILM
2 20 W
Fig.29: G. Nalty’s solution for linearising
the output capacitor (Q2 allows the DC
feedback resistor value to be high).
Practical Electronics | April | 2025
INPUT 100nF
220kW
NE5532
47mF
TANT. OUTPUT
C2
C1
INPUT
10kW
220kW
330nF
1kW LOAD CAUSES HUMP
IN RESPONSE <at> 4Hz
100nF
PLASTIC
FILM
1k W
NE5532
OUTPUT
10mF
TANT.
INPUT
OUTPUT
C1*
* CRITICAL
CAPACITORS
220kW
HIGH-PASS
SALLEN-KEY FILTER
10kW
INPUT C1*
Fig.30: an op amp follower with its
electrolytic output coupling capacitor in
the feedback loop.
Fig.31: an inverting amplifier trick to
bootstrap C1 on a virtual earth to
increase its effective value by 10 times,
a plastic film type can reduce distortion,
as shown in Fig.25. Since capacitor
distortion is a low-frequency effect, it
doesn’t really work. It does work for
high-frequency decoupling, however.
When I made some measurements,
it did reduce some of the high-order
distortion harmonics when gross LF
distortion was occurring by about 30%.
But if that’s happening in your circuit,
it probably isn’t what you’d call HiFi,
anyway.
well with normal power amps, as shown
in Fig.28.
Fig.29 shows a technique developed
by engineer/PE writer Graham Nalty for
single-rail amplifiers with a single input
transistor. He used this in his Class One
Sound amplifier in Everyday Electronics,
January 1989. Op-amp followers (Fig.30)
can also have the output capacitor
included in the loop.
Another trick is to apply NFB to input
capacitors on inverting amplifiers, as
shown in Fig.31. This allows a smaller
capacitor C1 for a given LF roll-off to be
used. Calrec used this in their virtualearth mixer amplifiers. The output
capacitor, C2, can also be included in
the feedback loop to linearise it.
The input capacitor is not linearised,
but it can be reduced in capacitance
enough to be a film capacitor.
Care has to be taken to prevent gain rise
at subsonic frequencies when applying
NFB around capacitors. The circuit
should be tested for LF instability with
tone bursts and the required load.
If the capacitor values are not carefully
selected so the frequencies are spaced
apart, LF bouncing on tone burst and
pulses may occur. Sometimes LF
oscillation or ‘motorboating’ can result
from the phase shift.
Negative feedback
LOW-PASS
SALLEN-KEY FILTER
C2
C2
OUTPUT
Fig.32: minimising the voltage drop
across C2 allows a cheap polyester
capacitor to be used in an active filter.
second bootstrapped capacitor (C2) in
Sallen-Key active filters was much less
sensitive to non-linearity than the first,
C1 (Fig.32). This reduces the part cost
for high-performance active crossovers.
Mechanical construction
The classic axial cylinder construction
tends to give the lowest distortion, and
most high-quality audio capacitors are
of this shape. The late Cyril Bateman,
a capacitor engineer at Erie (later ITT)
found that flattened oval shapes and
stacked-foil forms increased distortion.
I suspect this is because the mechanical
stresses are less evenly distributed, and
there may be regions where movement
could occur, leading to self-modulation.
Foil capacitors are large and expensive.
They can be inductive if each lead is
attached at only one point to the foil.
This was later eliminated with the
extended-foil technique, where the
whole of the foil on the end was soldered
together (see Fig.33).
Most plastic film capacitors today are
metallised film types, which have higher
losses but are smaller and cheaper. They
also have the benefit of possessing selfhealing properties (Fig.34). Foil capacitors
are designated F in the European (DIN)
system, with M for metallised. Some
Wima foil capacitors are shown in Fig.35.
Capacitor distortions produced within
a negative feedback (NFB) loop, such
as driver-stage bootstrap capacitors,
emitter bypass, and the VBE multiplier
bypass in power amps are all reduced by
the feedback. Feedback compensation
capacitors, such as CDOM (the dominant
pole ‘compensation’ capacitor, usually
across the VAS transistor), must be
linear, though.
I once used an X7R ceramic capacitor
instead of an NP0 type for CDOM and the
high-frequency distortion went up by
a factor of four. (By the way, NP0 and
C0G are different terms for the same
specification).
The distortion-reducing powers of
NFB can be applied to output capacitors
on power amplifiers. The distortion
reduction is usually in the order of 10
times (20dB). I discovered this effect
when making constant-current power
amplifiers, as shown in Fig.27. Including
the output electrolytic in the NFB loop
reduced the distortion.
NFB around the output capacitor works
While researching his book The
Design of Active Crossovers (2nd ed,
Routledge, 2020), Douglas Self observed
that small polyester capacitors improve
with use. However, the effect was not
permanent. He also showed that the
Most leaded components now have
copper-clad steel leads. This can
Fig.33: the clear epoxy end-fill on Wima
polyester foil capacitors reveals that the
contact is on the extended foil.
Fig.34: self-healing of a metallised film.
Metal burns away from the defect,
isolating it (note how thin it is!).
Fig.35: a Wima metal-foil capacitor
compared to its smaller metallised
version.
Practical Electronics | April | 2025
Subtler stuff
Capacitor leads
49
Fig.36: a Wima MKB3 metal-cased
polycarbonate capacitor. Unfortunately,
mechanical resonance ruined its sound.
Fig.37: dipping silvered-mica capacitors
at Charcroft in epoxy powder. This
protects them from moisture etc.
Fig.38: the final encapsulated mica
capacitors. Other types of capacitor,
like ceramics, are protected similarly.
Fig.42: a sheet of capacitor-grade
Indian ruby mica (Charcroft).
Fig.43: a polyester foil capacitor being
unwound.
Fig.44: an unusual cellulose acetate
(MKL) cap, the first plastic film type.
cause hysteresis-induced distortion
in some high-current situations. It
may also increase distortion from the
contact inside the capacitor. There is
now a market for the older versions
of capacitors that used copper leads.
Generally, high-quality audio capacitors
have pure copper leads; you can check
this using a magnet (it will stick to steel
but not pure copper).
capacitors in their tweeter crossovers
so one would dampen the resonance
of the other. It's always better in audio
to have multiple low-level resonances
than a high-amplitude single one.
These effects are only audible to skilled
listeners with good high-frequency
hearing, though.
has to be done in a vacuum chamber to
remove bubbles (see Fig.40).
Acoustic resonance
Total Harmonic Distortion (%)
A subtle effect in film capacitors is
where the capacitor has its own acoustic
resonances that are driven by and
contaminate the original signal. The effect
is usually in the 10-20kHz region and can
occur in large plastic-film capacitors used
in passive crossover tweeter sections.
I had it once with some Wima MKB3
polycarbonate capacitors (Fig.36). You
could hear an acoustic metallic “ting”
around 7kHz when you tapped the
metal case. Apparently, there were
two versions of this capacitor; one was
epoxy resin filled and one that had free
air inside. The epoxy ones were nonresonant and sounded better.
Mordaunt Short used to parallel two
Encapsulation
Most capacitors are dipped in epoxy
powder and then heated (see Figs.37
& 38). Higher-quality units are epoxy
potted. Fig.39 shows a special potted
80nF mica capacitor made by Charcroft.
All capacitors are susceptible to
moisture ingress, especially paper and
solid tantalum types. It is possible to
obtain glass/metal hermetically sealed
tantalum types at high cost.
However, there are plenty on the
military surplus market. Because they
are hermetic, they are usually perfect,
even if they have been stored since the
Cold War.
Paper and mica capacitors were
traditionally impregnated with silicone
wax to improve stability and moisture
resistance before encapsulation. This
Here's a list of capacitors from best to
worst for audio work.
• Teflon (PTFE), K=2.1. These types are
very rare and expensive. Teflon is still
used for trimmer capacitors.
• Polystyrene foil (FKS). K=2.56. Easily
melted by soldering and difficult to
source, but still the best for audio
filters such as RIAA equalisers and
active crossovers. Typically they have
a 1-2.5% tolerance.
• NP0/C0G class 1 ceramic, K=30.
These are good for HF compensation
capacitors for amplifiers. They
typically have 2–5% tolerance with
a small size, excellent for SMT up to
10nF. Sometimes marked with a black
paint band across the top.
• Polypropylene foil (FKP), K=2.2. They
are good for values up to 47nF.
• Metallised polypropylene (MKP; see
Fig.41), the standard quality plastic
film capacitor type for audio. Note that
there are fakes around. I've unwound a
0.05
0.02
0.01
.005
.002
.001
.0005
.0002
.0001
20
50
100
200
500
1k
Frequency (Hz)
2k
5k
10k
20k
Fig.48: the very low distortion a 100µF 35V bipolar Suntan electrolytic <at> 8V RMS.
50
Audio ranking
Fig.49: Nichicon make excellent bipolar
electrolytics. This one is suitable for
speaker coupling.
Practical Electronics | April | 2025
Fig.40: top-quality capacitors are
wax impregnated in a heated vacuum
chamber to exclude air bubbles.
Fig.41: standard audio grade
polypropylene capacitors, as used in
speaker passive crossovers.
Fig.45: a wax-covered paper foil
capacitor popular in 1950s valve gear.
Fig.46: old metallised paper capacitors.
These black Hunts types are unreliable.
Fig.47: Philips ceramic plate capacitors
are colour-coded by dielectric type.
few “audio polypropylene” capacitors
and found a standard polyester type
inside. You can tell as polypropylene
is soft and polyester is hard.
• Mica (Fig.42), K=6. A proven natural
material for capacitors over 100 years.
Very stable, with 0.1% tolerance
available, but expensive.
• Polycarbonate (MKC). Used in original
BBC LS3/5A crossovers. A more stable
version of polyester with a higher
dielectric strength. K=3, and lower
losses above 20kHz. They are no
longer made.
• Polyester (FKT/Mylar/PET), K=3.2,
used for soft drink bottles and the first
practical plastic film dielectric. It was
developed by German chemists during
the Second World War to deal with
the capacitor paper shortage. They
are smaller than paper foil capacitors
by 15%. Fig.43 shows a polyester foil
capacitor being unwound.
• Metallised polyester (MKT). The
general low-cost audio capacitor. 50%
smaller than foil types.
• Glass, K=5.6. The original dielectric
used for Leyden jars. They are
radiation-resistant and have a very
wide operating temperature range.
• Lacquer film/cellulose acetate (MKL),
K=4. Like paper, but very small and
radiation resistant. Were only made
by Siemens (see Fig.45).
• Paper in oil. Used by valve enthusiasts.
The viscously damped soft paper gives
a very low mechanical resonance.
• Impregnated paper-foil K=4. Often
only dipped in wax, as shown in
Fig.45. They absorb moisture, so they
can be leaky.
• Metallised paper. Avoid the Hunts
(Fig.46) and Rifa types like the plague.
• Mixed dielectric, made by Dubilier.
They look like a blue-and-white chewy
sweet. They have the low leakage of
polyester with the sound quality of
paper. Polyester is the main dielectric
between the foils. They were popular
in old TVs.
• Medium-K ceramic. Unstable, with a
large negative temperature coefficient.
Philips used to make a whole range
of square ceramic plate types with
differently coloured tops (see Fig.47).
Orange was -150ppm/°C and violet
was -750ppm/°C. I used them for
temperature compensation in tuned
circuits with ferrite inductors.
Esoteric electrolytics
Fig.50: a pair of Cerafine low-distortion
electrolytic capacitors.
Practical Electronics | April | 2025
Total Harmonic Distortion (%)
Fig.39: or this big ‘special’ 80nF
silvered-mica capacitor was potted in
epoxy resin.
• Wet aluminium bipolar types are the
cheapest option for a low-distortion
electrolytic (Fig.48). Nichicon's UEP
series (Fig.49) capacitors are excellent.
• Cerafine by Elna, with a paper separator
impregnated with fine ceramic
particles. The distortion specification
is ﹣120dB at 10kHz & 0.1A. Distortion
does increase at low frequencies in the
normal way (see Figs.50 & 51).
• Slit-foil electrolytics (Fig.72). These
capacitors are made by Kemet and
designed by Denis Morecroft of DNM.
They are the best speaker coupling
and main reservoir caps. The foils
are slit to reduce circulating eddy
currents, lowering distortion (in
theory). They were used in Mission
Cyrus II amplifiers. There is even a
four-terminal version, the T-network
capacitor, with internal connections
at optimum points on the foil for
minimum rectifier spike induction,
giving 10dB less HF noise.
• Silmic branded audio electrolytics
with special silk paper made by Elna.
They have lower distortion than
Cerafine types.
0.05
0.02
0.01
.005
.002
.001
.0005
Suntan
.0002 Cerafine
.0001
20
50
100
200
500
1k
Frequency (Hz)
2k
5k
10k
20k
Fig.51: the distortion of Cerafine versus normal (Suntan) electrolytic caps.
51
Total Harmonic Distortion (%)
0.05
0.02
0.01
.005
.002
.001
.0005
F95 tantalum
STC tantalum
.0002
F95 back-to-back F95 back-to-back, biased
.0001
20
50
100
200
500
1k
Frequency (Hz)
Fig.52: a Black Gate electrolytic cut
open (from Jimmy’s Junkyard). Note
the black carbonised winding.
Fig.53: distortion plots for F95 (single & back-to-back) vs normal STC tantalums.
Fig.57: solid tantalum bead types. They
can suffer from moisture ingress. The
type on the right seems prone to failure.
Fig.58: Oscon solid-aluminium polymer
capacitors from the SG audio series.
These have an epoxy-sealed base.
Fig.59: Panasonic SEPF, Nichicon FP
and Würth Elektronik solid polymer
electros. These all have rubber bungs.
• Black Gate by Rubycon, an audio
electrolytic with graphite-impregnated
paper separator. The paper winding
looks black if you open them up
(Fig.52). They had the best distortion,
down to -150dB at 10kHz as per the
data sheet. They were discontinued in
2006. Audio Note Kaiseis are the new
replacement, but their cost is beyond
engineering comprehension.
• Audio F95 tantalum by AVX (Fig.54).
Sadly they are only available in SMD
packages, with a quarter the distortion
of conventional solid tantalums, as
shown in Fig.53. Apparently, the
lower distortion is possible due to
a thicker dielectric and frameless
design. When biased, distortion drops
further to 0.001%. These are the
electrolytic caps to use for top-quality,
high-reliability audio. At £1.92 a pop,
they are expensive, but very small.
• Metal-cased solid tantalum (Fig.55).
These seem to last forever. They are
made mainly by Kemet today, eg, the
T110, CSR13, Vishay CTS13 and STC
TAA series. They are very expensive at
over £5 new, but cheap on the surplus
market. These were used in Nagra tape
recorders and avionics.
• Wet tantalums are very rare, and the
most expensive, an eye-watering £40
plus. Good for high voltages. Distortion
should have been very low, but when
I measured the few military surplus
samples I had, such as the bipolar
15µF 100V Stantalum in Fig.56, it was
worse than a normal solid tantalum.
I ran it at 8V RMS into a 600Ω load,
whch gave 0.15% distortion at 20Hz.
• Tantalum beads (Fig.57). Kemet's and
Union Carbide's are excellent. The
old ITT/STC red and blue types have
poor reliability; their later yellow ones
were better.
• Polymer tantalums have lower ESR and
don't ignite if abused, like manganese
oxide tantalums. Their distortion may
be lower, but I haven't checked yet.
• Oscons, the first solid polymer
electrolytics, developed by Sanyo but
now made by Panasonic. Distortion
is still high, but the ESR and hence
decoupling ability was in a new league.
They did an audio version shown in
Fig.58, the SG, which had copper leads,
but the distortion was still slightly
higher than solid tantalum. However,
their CV product went up to a higher
level than tantalum types, allowing
more wet electrolytics to be eliminated
from high-reliability designs.
• Solid aluminium manganese dioxide
electrolytic, a Philips invention to
get around the high cost of tantalum.
They have too much leakage for audio
coupling use, and high ESR, although
this is beneficial for regulator output
damping. They were the most reliable
electrolytics, with high reverse voltage
ratings, but are now obsolete.
• Generic solid polymer electrolytic
(Fig.59). Their very low ESR makes
them fantastic for decoupling, but
can cause regulators to oscillate. To
stop this, I've had to insert a resistor
in series, but this does allow the ESR
to be optimised and defined. Their
leakage is four times higher than wet
aluminium, so it says “not suitable for
coupling” on the datasheet. Also, the
distortion is twice as high as tantalums;
back-to-back connection and biasing
does not seem to help. They easily
go short circuit if reverse biased. I've
found them to be excellent emitter
bypass and bootstrap capacitors in
audio amplifiers, though.
• Hybrid polymers contain liquid
electrolyte but it's impregnated with
polymer particles, giving the low ESR
of polymer but with the self-healing
characteristic of normal electrolytics.
They give the best of both types, but
can still dry up. I bought a couple of
samples of a Kemet hybrid polymer
(Fig.60). These are built into the almost
hermetically sealed Rifa-style axial
case design with a metal end disc.
They show great promise as a speaker
coupling capacitor, being similar to
Black Gate types but at an industrial
price, rather than an “audiofool” price.
• Niobium oxide; although advertised
as “Oxicap audio caps”, these have
higher distortion than tantalums. I
think it is because the oxide's linear
bandgap region is smaller.
• X7R multilayer strontium titanate
ceramic, K=250. Use for decoupling
only. Their distortion is ten times that
of tantalum types.
• Z5U/Y5V barium titanate ceramic,
K=600-1200. Use for decoupling only.
Even worse distortion than X7R!
• Barrier-layer ceramic class 3, K
up to 50,000. Introduced by Erie
(Fig.61) in the 1960s for transistor
decoupling. They often have very
low voltage ratings, eg, 12V. They use
a semiconductor effect to give high
capacitance with a single layer but
can't be made with multiple layers.
The distortion is so bad that they are
useful for audible demonstrations of
capacitor distortion!
52
2k
5k
10k
20k
Practical Electronics | April | 2025
Fig.54: the F95 audio tantalum from
AVX. The spike indicates the positive
terminal.
Fig.55: metal-cased hermetic tantalums
from the ITT TAA series and Sprague
150D series; these last for decades.
Fig.56: a 15µF 100V Stantalum bipolar
wet tantalum by STC. I’m not sure
what is so special about this type.
Fig.60: a Kemet hybrid polymer
electrolytic, which has a very long life
for wet technology.
Fig.61: surface-barrier ultra-high-K
ceramics with distortion in the order of
2%. Maybe useful for fuzz boxes!
Fig.62: Mullard Mustard capacitors,
officially called type C296. A valve amp
maker’s dream for the last 65 years!
Classic audio capacitors
These axial polyester-foil capacitors
never go wrong, apart from the leads
falling off. They were used in the
Philips Electronic Engineer kits and
they were the first UK plastic film types,
introduced in 1958. They are no longer
made, highly collectable and I've seen a
fight at a radio rally over them.
guitar people. Retro Vibe have reissued
the Bumble Bee for £33 (see Fig.63)!
The Fender Stratocaster forum showed
a picture of a fake Bumble Bee cap
(Fig.64), they are meant to be paper foil
types. With a $5000 guitar, you want to
maintain authenticity. Maybe I should
start dipping my Chinese polyesters in
paint to make “tropical fish” caps. They
would be better than the real thing!
Wima
The plastic film capacitor was a
German invention pioneered by the
Wima company, which is still going
strong today.
Cambridge Audio uses them
and they advertise that fact. The
MKS2B051001N00JS 10µF 50V type is
only 11mm wide and 16mm high with a
5mm lead pitch. However, it costs £3.50.
Mullard/Philips
This company had a huge following
in capacitors. They later reduced
production, handing it over to BC
Components and Vishay. Their
transparent shrink-wrapped blue
electrolytics have always looked good,
but go leaky if not polarised.
“Mustard caps”
These are Mullard's most famous
capacitors (Fig.62), more like Dijon
mustard rather than English. If you see
these in an old amplifier or radio, do not
replace them.
Fig.63: a Sprague “Bumble Bee”
cap. This is a Thomann TAD Jupiter
reproduction model for guitar restorers.
Practical Electronics | April | 2025
Tropical Fish, Mullard/Philips C280
capacitors
These are loved by guitar pedal builders
and are so pretty they have been used as
ear-rings. They are quite poor capacitors
with fragile lead attachment, 10-20%
tolerance and high distortion. I had an
old Solina organ full of these and I got a
thrill just looking at the PCBs.
They were introduced in 1962 and
discontinued in 1984. They became
the Vishay MKT368, which have a
nice orange epoxy coating, a European
version of the US “Orange Drop” made
by Sprague.
“Black Beauties” and “Bumble Bees”
These are more nostalgia fest capacitors
from US company Sprague, loved by
Mullard/Philips polystyrene
One of the most stable capacitors ever
made, usually with a 1% tolerance. They
are now banned by environmental laws
because they used lead foil. This was,
of course, their great advantage because
the soft lead supporting the hard brittle
polystyrene avoided the cracking that
occurs when aluminium foil is used.
The lead also ensured an excellent
soldered connection on the end of the
winding in comparison to the difficulties
in soldering the wire contact to an
aluminium electrode.
The original yellow “Mini Poco”
version (Fig.65) used copper leads and
is highly sought after, along with its
Fig.64: a fake “Bumble Bee” capacitor.
Once opened, it reveals there was a
cheap polyester capacitor inside it.
Fig.65: “Mini Poco” polystyrene
capacitors. They are top-notch audio
filter capacitors if rather large.
53
Fig.66: Philips 444-series polystyrene
capacitors. Similar to the “Mini Poco”
but with tougher epoxy encapsulation.
Fig.67: the final version of Philips’
polystyrene capacitor. The blue cap is
their polypropylene replacement.
Fig.68: the biggest polystyrene I have
seen, Suflex 432nF. The double leads
give lower parasitic inductance.
Fig.69: current polystyrene capacitors
by LCR. Some versions now use tin foil.
Fig.70: a TCC paper-in-oil capacitor
from the Cold War.
Fig.71: a “Vitamin Q” military paper-inoil capacitor (from an old eBay photo).
replacement, the 444 series, shown in
Fig.66.
The newer green version (Fig.67) was
smaller and, because it had iron leads
was okay with flow soldering. There is
now a blue polypropylene replacement,
which is almost as good.
oil type by TCC Telegraph Condenser
Company is shown in Fig.70. The most
desirable type is the Sprague Vitamin Q
(Fig.71) which, being hermetically sealed,
doesn't suffer from electrical leakage.
mechanical resonance, around 15dB
down compared to normal similar
capacitors.
The end spray is zinc with a copper
sponge matrix soldered on with thick
copper leads. KEF use 5.7µF Clarity
capacitors in their LX50 reference
speakers.
One trick top HiFi companies use with
specialised audio capacitors is to get
their own branding printed on. Fig.73
shows a pair of branded slit-foil reservoir
capacitors in a Cyrus amplifier.
So there you have it, just part of the
mountain of tacit knowledge on the
audio aspects of capacitors. I think I
PE
should write a book!
Suflex HS polystyrene
These are very stable, low in cost
and popular in RIAA networks in HiFi
amplifiers. Their weakness is melting
while soldering.
Suflex were the go-to choice if you
wanted an odd custom value. I had an
old lady neighbour who used to work at
their Welsh plant. A sideline to her main
assembly job was to inform MI5 where
weird custom capacitors were going!
She gave me a few boxes before she died.
There are some very odd values in there;
432nF ±1%, anyone (Fig.68)? I'll have to
design a circuit especially for them.
The world's last polystyrene capacitors
(Fig.69) are still made by LCR in Wales,
who bought up Suflex's stock of film.
They still make the HS series.
Charcroft (ICW) Clarity Caps
I have to confess a vested interest here
since I did some work for the owning
company, Charcroft Electronics, who
make these, and Salford University, who
performed the research.
Clarity Caps are axial series-wound
metallised polypropylene types with
a mixture of polyurethane resin for
damping and epoxy resin end-fill
(Fig.72). This gives them the lowest
Sprague Vitamin Q
Paper-in-oil capacitors are highly
favoured by valve amplifier makers.
Like all wet capacitors, they have very
low resonance. An old British paper-in-
Fig.72: Clarity Caps – highly developed
audio polypropylene (Charcroft).
54
Fig.73: special slit-foil electrolytics in a Cyrus amplifier. Note the company’s
branding in gold. If you are paying for it, you should show it off!
Practical Electronics | April | 2025
|