Coupling capacitor voltage rating?

[Paul Burridge]

For best long-term reliability, if there is no DC present, you should
use a reversible electrolytic, or two ordinary ones back-to-back, maybe
with some DC bias applied to their junction. But in many cases, just a
single ordinary electrolytic is quite OK. I don't like to use anything
less than 6 V caps, though, in such applications.

ISTR someone on this group stated a while back that if a UK-spec
electolytic is marked at say 16V or 24V or example, that it wouldn't
be much use where there's only a very small DC potential across it.

If your caps are elderly, re-form them by connecting to a DC supply
equal to the marked working voltage and monitor the leakage current.

Again ISTR that the proper way to do this is to slowly turn the
voltage up bit-by-biy from a low level over a period of a few hours...

 

[John Woodgate]

I read in sci.electronics.design that [email protected]

Most caps rated at dc.
Many have only 1/4 (or less) rating at 1kHz.

It should be clearly understood that the whole point of audio *coupling*
capacitors is that their capacitance values are big enough to *prevent*
any significant audio voltage being developed across them, even at the
lowest frequency of interest.

 

[Reg Edwards]

ISTR someone on this group stated a while back that if a UK-spec

electolytic is marked at say 16V or 24V or example, that it wouldn't
be much use where there's only a very small DC potential across it.

===================

Just goes to show that half the information disseminated via this newsgroup
is dangerous rubbish.

 

[John Fields]

===================

Just goes to show that half the information disseminated via this newsgroup
is dangerous rubbish.

--- ^^^^
Which half does that information belong in?^)

 

[Reg Edwards]

Just goes to show that half the information disseminated via this
newsgroup

Which half does that information belong in?^)

Congratulations. You have hit the nail right on the head. From the
unfortunate, uninformed questioner's point of view there is no way of
distinguishing one from the other.

The most elequent response, which has nothing to do with valdity and
correctness, stands the best chance of being accepted.

 

[Terry Pinnell]

I read in sci.electronics.design that Frank Bemelman
anadoo.nl>) about 'Coupling capacitor voltage rating?', on Sat, 14 Feb
2004:


It bootstraps the 100 k resistor, making it look like maybe 10 Mohm at
low frequencies. I'd rather not go into what it does at 20 kHz!

Why? It looks fine here at 20 kHz. Here are a couple of 'scope shots,
at lowish and highish amplitudes respectively:
http://www.terrypin.dial.pipex.com/Images/Follower20k1.gif
http://www.terrypin.dial.pipex.com/Images/Follower20k2.gif

Even at 200 kHz it doesn't perform badly for its intended purpose (as
a simple test instrument to ensure minimal loading):
http://www.terrypin.dial.pipex.com/Images/Follower200k1.gif

Presumably, you're thinking of something like this, at 50 kHz and
highish amplitude?
http://www.terrypin.dial.pipex.com/Images/Follower50k1.gif

(Vcc = 20 V throughout.)

-------

For interest, I also made the following measurements:

f kHz Vout p-p before distortion observed
------ -----------------------------------
1 15.0 (clipping, no distortion)
5 15.0 (clipping, no distortion)
10 11.0
20 5.5
30 3.3
40 2.4
50 2.0

Increasing supply to 30 V made little difference. Reducing it to 12 V
more so; e.g. the 10 kHz figure became 4.3 V p-p.


I'm sure I'd see an improvement if I exchanged it for a more modern
chip, but it does its job well enough. And I still have a fair stock
of 741s...

 

[Terry Pinnell]

I read in sci.electronics.design that Frank Bemelman
anadoo.nl>) about 'Coupling capacitor voltage rating?', on Sat, 14 Feb
2004:


No only that, but eliminating R3, C4 and C5, and making R4 10 kohms to
earth would give more output swing.

But would it retain all other aspects, including the impedance in the
hundreds of megohms range?

A 741 driving 1 kohm is not a very happy bunny.

Well-spotted, that was my error. R3 is actually 10k not a 1k (the
Circuitmaker default for resistor, whch I forgot to change).

 

[Terry Pinnell]

Your op amp +input dc level is 0V, thanks to R4 and R3. Why is the
output at 270mV ? I expect that to be closer to 0V.

Departure of 741 from ideal? Poorly-matched R1/R2?

I would not expect an output capacitor in this situation, where
you have dual supplies. The same applies to power amplifiers, or
bridge amplifiers, like the ones in car-radios. No output cap.

This *is* a single-supply circuit, with artificial Common rail.

With an amplifier that uses a single supply, the DC level from the
output transistors is 'placed' at about Vsupply/2. In that case
you need a cap between output stage and speaker.

Yes, as here! (Although, as a Voltage Follower, used as a simple piece
of bench test equipemnt, its load is usually a DMM or 'scope probe,
not a speaker.)

 

[Frank Bemelman]

Departure of 741 from ideal? Poorly-matched R1/R2?


This *is* a single-supply circuit, with artificial Common rail.

Yes, but the circuit doesn't know it's articficial

Yes, as here! (Although, as a Voltage Follower, used as a simple piece
of bench test equipemnt, its load is usually a DMM or 'scope probe,
not a speaker.)

You would still not need a capacitor, since the ground of your output
is attached to the artificial ground. It has a dual supply, you converted
your single source into a dual one.

In a single supply situation, without an artificial ground at 1/2 Vsupply,
and the load (or speaker) attached to the real ground, yes, then you need
that cap.

I hope it's not confusing

 

[Fred Bloggs]

Not sure I follow that, Frank. Maybe a specific example would help, so
here's one of the circuits I had in mind:
http://www.terrypin.dial.pipex.com/Images/ACFollowerOutputCap.gif

I applied 1V rms as input. Regardless of supply voltage, DC across C3
is a mere 270 mV.

You can accomplish the same thing with:

Please view in a fixed-width font such as Courier.



*
NON-POLARIZED



*
35V |\ *

---||---+---------|+\ 35V

| | >-------+---||--+------>
Vin | +---|-/ | |
| | |/ | | Vout

-+ / | | |

| 10K | 10K | | +->
--- / +-----/\/\-----+ | |
/// \ | ---
| | ///
+----------------------------+
|
|
/
10K
/
\
|

 

[John Woodgate]

I read in sci.electronics.design that Terry Pinnell <terrypinDELETE@dial

Why? It looks fine here at 20 kHz. Here are a couple of 'scope shots, at
lowish and highish amplitudes respectively:
http://www.terrypin.dial.pipex.com/Images/Follower20k1.gif
http://www.terrypin.dial.pipex.com/Images/Follower20k2.gif

No. I was referring to what the input impedance looks like at 20 kHz. It
won't be 10 Mohm and it won't be a pure resistance.

Even at 200 kHz it doesn't perform badly for its intended purpose (as a
simple test instrument to ensure minimal loading):
http://www.terrypin.dial.pipex.com/Images/Follower200k1.gif

Presumably, you're thinking of something like this, at 50 kHz and
highish amplitude?
http://www.terrypin.dial.pipex.com/Images/Follower50k1.gif

Well, that's slew-rate limiting, which is hardly surprising at 50 kHz.

 

[John Woodgate]

I read in sci.electronics.design that Terry Pinnell <terrypinDELETE@dial

But would it retain all other aspects, including the impedance in the
hundreds of megohms range?

No, obviously not. But when do you really need hundreds of megohms (at
low audio frequencies only)?

With 10 k ohms, the bootstrap isn't causing any loading problems at the
output, but it still has problems at highish audio frequencies.

 

[Paul Burridge]

===================

Just goes to show that half the information disseminated via this newsgroup
is dangerous rubbish.

Hello, Reg.
I've found the original posting which was the "old fart electronics
quiz" thread initiated by Max Hauser who posed and answered the
following question:

Q: "If an aluminum [UK: aluminium] electrolytic capacitor is rated for
a
"working voltage" of 10 volts, in what range of voltages is it
designed to
operate?

A:
Not more, but more importantly not much less, than 10 volts. Five
volts,
probably OK; two volts, it may not form the dielectric layer
reliably."

You're not saying Mr. Hauser is talking bollocks are you, Reg?

 

[Terry Pinnell]

You can accomplish the same thing with:

Please view in a fixed-width font such as Courier.



*
NON-POLARIZED



*
35V |\ *
| | >-------+---||--+------>
Vin | +---|-/ | |
| | |/ | | Vout
| 10K | 10K | | +->
--- / +-----/\/\-----+ | |
/// \ | ---
| | ///
+----------------------------+
|
|
/
10K
/
\
|

Thanks, but that standard non-inverting circuit won't have anything
approaching the very high input impedance of my circuit. Nor does it
appear to use a single, positive supply (although you've not
explicitly showed the supply pin connections). And I'm not clear why
it's a follower rather than an attenuator?

 

[John Woodgate]

I read in sci.electronics.design that Terry Pinnell <terrypinDELETE@dial

Thanks, but that standard non-inverting circuit won't have anything
approaching the very high input impedance of my circuit.

Yes, it does. The 10 k resistor connected to the + input is
bootstrapped.

Nor does it
appear to use a single, positive supply (although you've not explicitly
showed the supply pin connections).

Left as an exercise for the student. (;-)

And I'm not clear why it's a
follower rather than an attenuator?

The - input is connected to the output via a 10 k to reduce offset. Gain
= 1 very closely.

 

[Paul Burridge]

What is the definition of "virtually no DC"? 1mV, 10mV, 100mV ?

The general idea is correct though, isn't it, Jim? If there's no
appreciable DC across the 'gap' then I would imagine one can safely
assume the interstage bias condtions are compatible so no DC-DC
isolation is necessary. The overall design saves on one capacitor and
a little less low frequency loss is a bonus.

 

[Terry Pinnell]

That circuit is crap. The guy doesn't know anything about opamps. Here is a
more appropriate one:

+------+------------o
| | 9-30V
.-. |
220k| | | 100n
| | | ||
'-' +--||--GND
| | ||
|# | |\|
o-+---|#---+----|+\ #|
| |# | | >--+----#|---+--o
| 10u | +-|-/ | #| |
| | | |/| | 100u |
| | | | | |
.-. .-. +---)---+ .-.
| | 220k| | | | |
1M| | | | | 1/2 TL072 | |1M
'-' '-' | '-'
| | | |
=== === === ===
GND GND GND GND
created by Andy´s ASCII-Circuit v1.24.140803 Beta www.tech-chat.de

Did you actually read the thread? Your circuit has an input impedance
(at low f) of a mere 100k (1M in parallel with 2 x 220k), in
comparison with the hundreds of megohms of mine. So 'appropriate' is
hardly the adjective I'd choose to describe it.

I'm sure that the source of my circuit would be amused to hear it
described as 'crap'. And with scores of widely-praised books to his
credit, read worldwide, many covering a range of practical opamp
circuit designs, I think your description of Ray Marston as a 'guy'
who 'doesn't know anything about opamps' equally risible.

 

[John Woodgate]

I read in sci.electronics.design that Terry Pinnell <terrypinDELETE@dial

I think your description of Ray Marston as a 'guy' who 'doesn't
know anything about opamps' equally risible.

Marston is generally sound, but, like everyone, occasionally errs. But
his designs do tend to 'throw components at the problem'. The same
applies to some designs published by component suppliers, who obviously
have little incentive to minimise component count.

 

[Terry Pinnell]

I read in sci.electronics.design that Terry Pinnell <terrypinDELETE@dial


Yes, it does. The 10 k resistor connected to the + input is
bootstrapped.


Left as an exercise for the student. (;-)


The - input is connected to the output via a 10 k to reduce offset. Gain
= 1 very closely.

OK, thanks John. And my apologies to Fred. I thought it was
uncharacteristic - I should have known better! I'll breadboard his
suggestion and try to understand it better.

 

[Terry Pinnell]

You can accomplish the same thing with:

<snip>

OK, thanks, I've now breadboarded that, using a dual supply (a
disadvantage from my viewpoint compared to my circuit).

I was surprised to find its impedance lower than expected, at about
5.5M. I then measured the input impedance of my circuit - and that was
very similar. Certainly not the 'hundreds of megohms' I've been
claiming.

My circuit, the one posted earlier at
http://www.terrypin.dial.pipex.com/Images/ACFollowerOutputCap.gif
was actually adapted from this one by Ray Marston:
http://www.terrypin.dial.pipex.com/Images/ACFollowerMarston1.gif

I had assumed the value of C3 was not critical. But as you see from my
tests, it appears to be very much so. Is that predictable? Marston's
text contains no insights into that aspect.

I'll probably now remove the 8 uF I'm using in my adapted circuit, and
by trial and error arrive at a presumably much lower value.