bypassing electrolytic caps

[John Washburn]

I'm about to recap my board (Soundcraft 200B) which is getting hissy on some
channels, and in researching options I found some discussion of bypassing
electrolytic capacitors with film caps in parallel.

The textbooks I've been using to teach myself basic electronics don't
discuss this practice, so before I go ordering a bunch of components I may
not even need, I have a few questions:

1. The benefit of bypassing electrolytic caps is typically in improved high
frequency response because the lower value film cap will pass higher
frequencies better than the higher value one, and vice versa? There's a
counter argument about phase shift--how much is that worth worrying about?

2. Are there places where this isn't useful/helpful/worth dealing with? I'm
guessing that the 47uf phantom blocking caps would benefit, but what about
interstage coupling, etc?

3. The value of the film cap should be 10% of the electrolytic. How critical
is that value?

4. In an existing design, you just connect both caps at the same point,
soldering the film cap to the point where the electrolytic attaches to the
circuit board? In this case there's room on the other side of the circuit
board. Since all the points on each side of a parallel network are
electrically the same point, this would work, right?

Thanks!

-jw

[Scott Dorsey]

John Washburn > wrote:
>
>1. The benefit of bypassing electrolytic caps is typically in improved high
>frequency response because the lower value film cap will pass higher
>frequencies better than the higher value one, and vice versa? There's a
>counter argument about phase shift--how much is that worth worrying about?

The electrolytic can be looked at as a capacitor with a resistor and an
inductor in series with it. The parasitic resistance and inductance are
a problem at high frequencies. The smaller value film cap will provide
a lower-impedance path to shunt those high frequencies around.

I don't know what the phase shift counter-argument is. It will indeed
change the phase characteristics of the circuit, but it will improve them.
This could be a problem in some high-feedback circuits that are designed
around particular electrolytics, but it should not be an issue in well-designed
gear.

>2. Are there places where this isn't useful/helpful/worth dealing with? I'm
>guessing that the 47uf phantom blocking caps would benefit, but what about
>interstage coupling, etc?

Certainly it's a good idea for interstage coupling. On a single-ended circuit
it's even a good idea for supply decoupling caps.

>3. The value of the film cap should be 10% of the electrolytic. How critical
>is that value?

Not very. It's a reasonable ballpark, though, but use whatever you have
on the sjunkbox.

>4. In an existing design, you just connect both caps at the same point,
>soldering the film cap to the point where the electrolytic attaches to the
>circuit board? In this case there's room on the other side of the circuit
>board. Since all the points on each side of a parallel network are
>electrically the same point, this would work, right?

Yes. Or (with an axial electrolytic) you can piggyback them on the same
side.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[Paul Stamler]

"John Washburn" > wrote in message
...
> I'm about to recap my board (Soundcraft 200B) which is getting hissy on
some
> channels, and in researching options I found some discussion of bypassing
> electrolytic capacitors with film caps in parallel.
>
> The textbooks I've been using to teach myself basic electronics don't
> discuss this practice, so before I go ordering a bunch of components I may
> not even need, I have a few questions:
>
> 1. The benefit of bypassing electrolytic caps is typically in improved
high
> frequency response because the lower value film cap will pass higher
> frequencies better than the higher value one, and vice versa?

Not really. The bypassed electrolytic has lower impedance at high
frequencies than the electrolytic by itself. In power supply caps that
translates to better filtering of high-frequency components, including RFI.
When the electrolytic is a coupling cap, the impedance at the highest audio
frequencies, even unbypassed, is low enough not to matter. The putative
advantage of bypassing audio coupling caps is that the bypass cap, being
film, suffers less from dielectric absorption, particularly at high
frequencies. My own experience is that bypassing electrolytic coupling caps
indeed reduces the amount of audible high-frequency "hash" on things like
sibilants, but not as well as replacing the electrolytics with film caps. Of
course, this often isn't possible due to space limitations.

>There's a
> counter argument about phase shift--how much is that worth worrying about?

Not at all; my tests show *less* phase shift in a bypassed cap assembly than
in an unbypassed electrolytic.

> 2. Are there places where this isn't useful/helpful/worth dealing with?
I'm
> guessing that the 47uf phantom blocking caps would benefit, but what about
> interstage coupling, etc?

Interstage coupling with electrolytics is already problematic, as typically
the (polarized) electrolytic is used without a DC voltage across it, a
perfect example of bad design. Bypassing is a good idea; connecting up a
*pair* of electrolytics with a resistor to the power supply may be better:
+ +
-----||---------||------
|
1 Meg
|
V+

You then bypass the nose-to-nose electrolytics with a film cap. Note that
the effective capacitance is half of the capacitance of each electrolytic,
so that if you used 100uF caps the combined capacitance would be 500uF.

All this is awkward and space-consuming, yes.

> 3. The value of the film cap should be 10% of the electrolytic. How
critical
> is that value?

That's a good value, but there will be measurable benefits to high-frequency
impedance even bypassing a 1000uF electrolytic with a 0.1uF-1uF film cap.
Typical electrolytics have their resonant point (the frequency where they
stop being capacitors and turn into inductors) anywhere from around 6kHz
(3300uF caps) to 30kHz (100uF caps). Small film caps remain capacitative up
to the 200kHz region (0.47uF), 600kHz region (0.1uF) or well into the
megahertz region (.01uF caps, ceramic discs).

More specific data coming up in an article for audioXpress, probably about a
year from now.

> 4. In an existing design, you just connect both caps at the same point,
> soldering the film cap to the point where the electrolytic attaches to the
> circuit board? In this case there's room on the other side of the circuit
> board. Since all the points on each side of a parallel network are
> electrically the same point, this would work, right?

Right.

Peace,
Paul

[Eric K. Weber]

You then bypass the nose-to-nose electrolytics with a film cap. Note that
> the effective capacitance is half of the capacitance of each electrolytic,
> so that if you used 100uF caps the combined capacitance would be 50uF.

500 was a typo.....

[Paul Stamler]

"Eric K. Weber" > wrote in message
...
> You then bypass the nose-to-nose electrolytics with a film cap. Note that
> > the effective capacitance is half of the capacitance of each
electrolytic,
> > so that if you used 100uF caps the combined capacitance would be 50uF.
>
> 500 was a typo.....

Quite right; thank you!

Peace,
PPPPPPaul

[Arny Krueger]

"John Washburn" > wrote in message


> I'm about to recap my board (Soundcraft 200B) which is getting hissy
> on some channels, and in researching options I found some discussion
> of bypassing electrolytic capacitors with film caps in parallel.

Recapping can be a darn good idea when you have problems like these. Did I
ever tell you about the time I thought my stereo receiver was losing its
bass, and I found that virtually every electrolytic coupling cap had lost
90% of its nameplate capacitance. Boy, did it sound better after I did THAT.

> The textbooks I've been using to teach myself basic electronics don't
> discuss this practice, so before I go ordering a bunch of components
> I may not even need, I have a few questions:

> 1. The benefit of bypassing electrolytic caps is typically in
> improved high frequency response because the lower value film cap
> will pass higher frequencies better than the higher value one, and
> vice versa? There's a counter argument about phase shift--how much is
> that worth worrying about?

I don't know of a counter argument based on phase shift.

There is a counter argument based on common sense - why do something that
has no tangible benefits? The answer to that challenge is that there may
sometimes be some tangible benefits. But not always.

The stated purposes for adding these caps relates to two well-known
effects - series inductance and dielectric absorption.

Every electrolytic capacitor, no every capacitor effectively has a little
inductor that is effectively in series with it. One historical way to
quantify this inductance is to give the frequency where this inductance
resonates the capacitance of the capacitor. Above this frequency, the
capacitor has lost its capacitive effects due to the series inductance. I
was surprised to learn that the self-resonant frequency of a common
wire-leaded 0.1 uF ceramic capacitor might be like 150 MHz. The major
source of inductance in that case was simply the lead wires.

At times the series inductance of a part may include more than just the lead
inductance, as the part itself may have appreciable inductance inside.
Aluminum electrolytic and larger film capacitors are commonly made up of
sheets of insulation and foil wound up in a roll. This is an obvious
opportunity to unintentionally introduce inductance. OTOH, a skillful
designer can work with the geometry and largely avoid excess inductance.
There are such things as low ESR and low ESL electrolytic capacitors. Most
of them end up in non-audio applications or in power supplies and power
amps.

Dielectric absorption is a more recent issue for audio, maybe 30 years old.
Dielectric absorption refers to the tendency of dielectric materials to
experience longer-term changes, and produce a capacitor that acts like it
has a larger capacitor in series with a large resistor, connected across the
capacitor's terminals.

In such cases quantification is off the essence. It's easy to say that "The
benefit of bypassing electrolytic caps is typically in
improved high frequency response", but what constitutes high frequencies,
and what constitutes a useful improvement?

These days it is relatively easy to make incredibly accurate frequency
response measurements, within a hundredth of a dB or better. Yet, AFAIK
nobody has ever produced an measurement of say a good commercial mic preamp
(to eliminate contrived situations), and showed that its frequency response
in the audio range was significantly improved by paralleling all of the
electrolytics with film caps.

If you run the numbers, you find out the reason why - series inductance in
most elecrolytics isn't a problem, and those few cases where it is, it gets
dealt with in the original design. Why? Because most ESL creates impedance
in the audio range that are a few ohms or less, and most coupling capacitors
drive resistances that are a thousand ohms or more - typically ten thousand
ohms or more. One ohm series reactance versus a ten thousand ohm load leads
to a loss of less than a thousandth of a dB.

If you understand dielectric absorption well, you just might hurt yourself
laughing when someone suggests that you address DA in a large electrolytic
by paralleling it with a film cap. Remember, we earlier found that DA was
due to a phantom capacitance that is in parallel with the cap. It's darn
hard to reduce parallel capacitance due to DA by *adding* parallel
capacitors, is it not?

> 2. Are there places where this isn't useful/helpful/worth dealing
> with? I'm guessing that the 47uf phantom blocking caps would benefit,
> but what about interstage coupling, etc?

Educate yourself. Learn how to make accurate frequency response of audio
gear, which is pretty cheap and easy to do these days. Then tack in some
parallel caps, and see what happens. What does it take to make accurate FR
measurements of audio gear? A PC with even a mediocre sound card and a
freebie piece of software called "Audio Rightmark".

What about all those deadly earnest listening tests where capacitor upgrades
saved some precious piece of legacy gear from the scrap heap? Well, either
the caps that were there to start with, had dried out and stopped being the
caps their nameplate described, or well unhh have you ever heard of
expectation effects, or placebo effects?

Let me tell you about all the DBTs that conclusively showed an audible
difference due to paralleling an electrolytic in a well-designed,
well-maintained piece of electronic equipment. () That was quick, now wasn't
it?

> 3. The value of the film cap should be 10% of the electrolytic. How
> critical is that value?

Depends on the application, but in the current context the value is so non
critical that you often find that adding it or removing it has no effect.

> 4. In an existing design, you just connect both caps at the same
> point, soldering the film cap to the point where the electrolytic
> attaches to the circuit board? In this case there's room on the other
> side of the circuit board. Since all the points on each side of a
> parallel network are electrically the same point, this would work,
> right?

Right, and that is why paralleling caps with film caps does nothing useful
for DA.