Maker Pro
Maker Pro

Capacitor Choice

M

Mike Warren

Jan 1, 1970
0
Please excuse the off-topic post in sci.global-warming. ;-)

I have 2 audio circuits that I need to be capacitively coupled at
around 5uF to 10uF.

The problem is that one of the stages has power removed from it
for about 2/3 of the devices operating time. Obviously, the audio
functionality is not needed at these times.

The main supply is 5V and the switched supply is 12V.

The DC offset for the continuously powered stage is about 2.4 Volts,
and the switched stage is about 5.8V.

In general, I like to design circuits with a small range of components
as all our projects are very low volume. This is especially true with
SM components, which is what I need to use in this case.

I will often use 2 (and sometimes 3) components to avoid buying in a
new value.

These are the possible solutions as I see it.

1/ SM Bipolar Electros.
Expensive, and are not a normally stocked part.

2/SM Ceramic.
1uF is the largest value we keep. I could probably get away with 3uF,
although that would not be ideal, but that is still 3 capacitors,
and the main problem with ceramics is microphonics.

3/ SM Tantalum.
BANG! Well, at least, I don't expect them to last very long before
becoming leaky.

4/ 2 series 10uF Tant. ---||+--+||---
I've done this with normal electros in the past, but don't know how
tantalums would handle it.

5/ SM Tantalum with circuit modification.
By adding a couple of diodes I could power the second stage from the
5V rail while the 12V supply is missing. this would still mean a slight
reverse voltage of about 0.3V when the 12V supply is missing.

This is the one I'm leaning toward. Does anyone know if tantalums can
normally handle that for long periods?

Any other thoughts?
These are the only in-stock parts that come close:
10uF/16V SM Tant.
22uF/16V SM Tant.
1uF/50V SM Ceramic.
 
P

pimpom

Jan 1, 1970
0
Mike said:
Please excuse the off-topic post in sci.global-warming. ;-)

I have 2 audio circuits that I need to be capacitively coupled at
around 5uF to 10uF.

The problem is that one of the stages has power removed from it
for about 2/3 of the devices operating time. Obviously, the audio
functionality is not needed at these times.

The main supply is 5V and the switched supply is 12V.

The DC offset for the continuously powered stage is about 2.4 Volts,
and the switched stage is about 5.8V.

In general, I like to design circuits with a small range of components
as all our projects are very low volume. This is especially true with
SM components, which is what I need to use in this case.

I will often use 2 (and sometimes 3) components to avoid buying in a
new value.

These are the possible solutions as I see it.

1/ SM Bipolar Electros.
Expensive, and are not a normally stocked part.

2/SM Ceramic.
1uF is the largest value we keep. I could probably get away with 3uF,
although that would not be ideal, but that is still 3 capacitors,
and the main problem with ceramics is microphonics.

3/ SM Tantalum.
BANG! Well, at least, I don't expect them to last very long before
becoming leaky.

4/ 2 series 10uF Tant. ---||+--+||---
I've done this with normal electros in the past, but don't know how
tantalums would handle it.

5/ SM Tantalum with circuit modification.
By adding a couple of diodes I could power the second stage from the
5V rail while the 12V supply is missing. this would still mean a
slight reverse voltage of about 0.3V when the 12V supply is missing.

This is the one I'm leaning toward. Does anyone know if tantalums can
normally handle that for long periods?

Any other thoughts?
These are the only in-stock parts that come close:
10uF/16V SM Tant.
22uF/16V SM Tant.
1uF/50V SM Ceramic.

Use two bipolar caps - Al or tantalum - back to back, and place a resistor
between their common terminal and ground.

+2.4V + + +5.8V
--------||-----||--------
|
/
\
/ R
\
/
|
|
GND
 
M

Mike Warren

Jan 1, 1970
0
Mark said:
as long as the audio voltage is a small fraction of the DC voltage,
selection 4 will be fine..

Thanks for your input.

using 2 capacitors is still not the preferred way. I just realised I
could increase the reference voltage by a volt on the second stage by
changing one resistor and by using the diodes so the second supply goes
to 5V instead of 0V I would still have a positive voltage on the caps.

I'd lose a bit of headroom, but not enough to worry about, I think.
 
M

Mike Warren

Jan 1, 1970
0
pimpom said:
Use two bipolar caps - Al or tantalum - back to back, and place a resistor
between their common terminal and ground.

+2.4V + + +5.8V
--------||-----||--------
|
/
\
/ R
\
/
|
|
GND

Thanks.
 
M

Mike Warren

Jan 1, 1970
0
Mike said:
Thanks for your input.

using 2 capacitors is still not the preferred way. I just realised I
could increase the reference voltage by a volt on the second stage by
changing one resistor and by using the diodes so the second supply goes
to 5V instead of 0V I would still have a positive voltage on the caps.

I'd lose a bit of headroom, but not enough to worry about, I think.

I should probably explain why. There are 8 channels of audio involved
here, so 2 diodes is better than 8 capacitors.
 
E

Eeyore

Jan 1, 1970
0
Mike said:
Please excuse the off-topic post in sci.global-warming. ;-)

I have 2 audio circuits that I need to be capacitively coupled at
around 5uF to 10uF.

The problem is that one of the stages has power removed from it
for about 2/3 of the devices operating time. Obviously, the audio
functionality is not needed at these times.

The main supply is 5V and the switched supply is 12V.

The DC offset for the continuously powered stage is about 2.4 Volts,
and the switched stage is about 5.8V.

In general, I like to design circuits with a small range of components
as all our projects are very low volume. This is especially true with
SM components, which is what I need to use in this case.

I will often use 2 (and sometimes 3) components to avoid buying in a
new value.

These are the possible solutions as I see it.

1/ SM Bipolar Electros.
Expensive, and are not a normally stocked part.

Bipolar electros are just 2 'back-to-back normal electros.

My advice ? Couple with 2 normal aluminium electros in series with the +
terminals joined together and bias that centre tap via the highest value
R you have from a non-switched supply voltage. I've done this myself
when film cap values would have otherwise been required but were too
expensive and or too hungry on board space.

Graham
 
E

Eeyore

Jan 1, 1970
0
pimpom said:
Use two bipolar caps

You mean polarised caps surely ?

- Al or tantalum - back to back, and place a resistor
between their common terminal and ground.

+2.4V + + +5.8V
--------||-----||--------
|
/
\
/ R
\
/
|
|
GND

Similar to my suggestion too but simpler in a way since GND is convenient.

The reason I went for another idea is that I normally work with bipolar
supplies and unpredictable offsets would prevent your method working for
me. In this case the positive only supplies makes your idea the better one.


Graham
 
P

pimpom

Jan 1, 1970
0
Eeyore said:
You mean polarised caps surely ?
Oops! Yes, I do mean polarised caps.
Similar to my suggestion too but simpler in a way since GND is
convenient.
The reason I went for another idea is that I normally work with
bipolar supplies and unpredictable offsets would prevent your method
working for me. In this case the positive only supplies makes your
idea the better one.
I too face that dilemma at times, as when coupling two points at near-zero
dc potentials, but with unpredictable polarities and a significant ac
component.

Ceramics of more than 0.1uF are not commonly available where I live, plastic
film caps are too expensive and/or bulky unless signal fidelity demands
their use. Sometimes I try to adjust impedances so that the low-freq cutoff
and phase shift will be acceptable with 0.1uF, at other times I use two
polarised caps back to back - a rather clunky solution.
 
Top