Back to Back Tantalum Caps

[PaulCsouls]

I was recently looking at a design that had two Tantalum caps reversed
biased to each other to make a bipolar cap. I know tantalums can work
down to -1v and don't need a DC bias the way aluminum electrolytic
caps do but this looked strange. What are the pitfalls to this
method.

Thanks
Paul C

 

[John Larkin]

I was recently looking at a design that had two Tantalum caps reversed
biased to each other to make a bipolar cap. I know tantalums can work
down to -1v and don't need a DC bias the way aluminum electrolytic
caps do but this looked strange. What are the pitfalls to this
method.

Thanks
Paul C

If there's an unknown DC polarity and a small AC signal, it's OK. If
the AC across the combination gets to be larger than a volt or so the
the capacitance will change during each AC cycle (as the
reverse-biased cap goes from being a capacitor to being a resistor)
and that can cause distortion and other weirdness.

The potential nonlinearities are complex and depend on the situation.

John

 

[Pooh Bear]

I was recently looking at a design that had two Tantalum caps reversed
biased to each other to make a bipolar cap. I know tantalums can work
down to -1v and don't need a DC bias the way aluminum electrolytic
caps do but this looked strange. What are the pitfalls to this
method.

The cost of tantalum caps perhaps ?

They *don't* need bias ? I rather thought they did to perform properly. In
fact, under the right condirions, aluminium electrolytics work well for
audio coupling under zero bias conditions. You need a big cap to ensure
that the ac component of the signal is essentially miniscule but you might
be interested to know that almost every single top-end modern audio mixing
console uses unbiased aluminium electrolytics to couple signal - right at
the very top-end too. I'm talking $100,000 + products here.

I do know an audio company that ( many years ago ) used back to back
tantalums for coupling that biased their centre junction to ensure biasing
polarity. Today that would be considered esoteric overkill and financial
suicide for a large product.

I have used that technique with a pair of aluminium electros though, where
the DC offset voltage at the source is subject to tolerances. Works
nicely.


Graham

 

[Tam/WB2TT]

I was recently looking at a design that had two Tantalum caps reversed
biased to each other to make a bipolar cap. I know tantalums can work
down to -1v and don't need a DC bias the way aluminum electrolytic
caps do but this looked strange. What are the pitfalls to this
method.

Thanks
Paul C

You might be better of putting a low Vf diode across each cap to keep the
cap from ever conducting forward current.

Tam

 

[John Larkin]

You might be better of putting a low Vf diode across each cap to keep the
cap from ever conducting forward current.

Tam

Makes it more nonlinear in many situations.

John

 

[PaulCsouls]

The cost of tantalum caps perhaps ?

They *don't* need bias ? I rather thought they did to perform properly. In
fact, under the right condirions, aluminium electrolytics work well for
audio coupling under zero bias conditions. You need a big cap to ensure
that the ac component of the signal is essentially miniscule but you might
be interested to know that almost every single top-end modern audio mixing
console uses unbiased aluminium electrolytics to couple signal - right at
the very top-end too. I'm talking $100,000 + products here.

I do know an audio company that ( many years ago ) used back to back
tantalums for coupling that biased their centre junction to ensure biasing
polarity. Today that would be considered esoteric overkill and financial
suicide for a large product.

I have used that technique with a pair of aluminium electros though, where
the DC offset voltage at the source is subject to tolerances. Works
nicely.


Graham

I'm just following Ian Sinclair's Passive Components for Circuit
Design. He says that Alumunium Electrolytics require a DC polarizing
voltage to maintain the insulating film and for Tantalum Caps he says
"Tantalum electrolytics can be used without any DC bias and can accept
a small reverse voltage, typically less than 1.0v." But he doesn't get
into the frequency characteristics and I would guess biasing would
improve the usable bandwidth.

Paul C

 

[PaulCsouls]

If there's an unknown DC polarity and a small AC signal, it's OK. If
the AC across the combination gets to be larger than a volt or so the
the capacitance will change during each AC cycle (as the
reverse-biased cap goes from being a capacitor to being a resistor)
and that can cause distortion and other weirdness.

The potential nonlinearities are complex and depend on the situation.

John

That's my intuition. I would pay for a bipolar cap with a data sheet
to give me that warm fuzzy feeling of knowing what's going on.

Paul C

 

[PaulCsouls]

You might be better of putting a low Vf diode across each cap to keep the
cap from ever conducting forward current.

Tam

Yes, or clamp the signal to +/- 0.7 volts.

Paul C

 

[Tam/WB2TT]

Makes it more nonlinear in many situations.

John

Wouldn't the cap acting as a diode be as bad or worse? He does not say what
this is for; so, I am assuming he wants to put 60 Hz through it, which I
have done at 24VRMS..

Tam

 

[John Larkin]

Wouldn't the cap acting as a diode be as bad or worse? He does not say what
this is for; so, I am assuming he wants to put 60 Hz through it, which I
have done at 24VRMS..

Tam

Without knowing the DC bias, AC current and frequency, and the
capacitance it's impossible to say how weird or maybe not weird this
would behave. It's complex.

John

 

[PaulCsouls]

Wouldn't the cap acting as a diode be as bad or worse? He does not say what
this is for; so, I am assuming he wants to put 60 Hz through it, which I
have done at 24VRMS..

Tam

It was in an AGC circuit. Basically a low pass filter with the caps in
series to ground two 47uF caps and 1k resistor so a 4Hz 3dB. Not very
demanding with respect to distortion or tolerances. I was just curious
about how the limitations of the configuration.

Paul C

 

[Tam/WB2TT]

It was in an AGC circuit. Basically a low pass filter with the caps in
series to ground two 47uF caps and 1k resistor so a 4Hz 3dB. Not very
demanding with respect to distortion or tolerances. I was just curious
about how the limitations of the configuration.

Paul C

What I had done was put 2 caps back/back with diodes to get a NP 200uF
capacitor for use with a reversible 2 phase electric motor in an antenna
rotator. Worked fine. I know of people who have don this with speaker
crossover networks, but as John points out, probably not a good idea.

Tam

 

[Rich Grise]

What I had done was put 2 caps back/back with diodes to get a NP 200uF
capacitor for use with a reversible 2 phase electric motor in an antenna
rotator. Worked fine. I know of people who have don this with speaker
crossover networks, but as John points out, probably not a good idea.

Aluminums can do this all day - that's what a "non-polar" cap is -
essentially two aluminum electrolytics back-to-back in one package.

The OP was asking about doing that with tantalum.

Nobody seems to know. ?:-|

I'd say, look for followups on the performance of the circuit design the
OP saw with that configuration, or put a couple on the bench in the
breadboard with some kind of test setup, and see what happens.

Cheers!
Rich

 

[Pooh Bear]

I'm just following Ian Sinclair's Passive Components for Circuit
Design.

Any link ?

He says that Alumunium Electrolytics require a DC polarizing
voltage to maintain the insulating film

Oh well. That means almost every piece of current pro and consumer audio is
built wrong then !

and for Tantalum Caps he says
"Tantalum electrolytics can be used without any DC bias and can accept
a small reverse voltage, typically less than 1.0v."

Alumiums are good for a few 100mV reverse volts typically. Perfect for split
supply audio apps.

But he doesn't get
into the frequency characteristics and I would guess biasing would
improve the usable bandwidth.

What 'frequency characterisitcs' ? Bias affects bandwidth ? Some very strange
ideas here.

Maybe this Mr Sinclair is one of those audiophool gurus ?


Graham

 

[Nico Coesel]

Aluminums can do this all day - that's what a "non-polar" cap is -
essentially two aluminum electrolytics back-to-back in one package.

The OP was asking about doing that with tantalum.

Nobody seems to know. ?:-|

I don't even want to know. If possible, I avoid tantalum caps because
they short out when they fail. They can also burn a hole right through
the PCB if you are unlucky.

When it comes to distortion, I think the best idea is to read the
datasheet and determine how the capacitance changes with the voltage.

 

[Nico Coesel]

I was recently looking at a design that had two Tantalum caps reversed
biased to each other to make a bipolar cap. I know tantalums can work
down to -1v and don't need a DC bias the way aluminum electrolytic
caps do but this looked strange. What are the pitfalls to this
method.

This may be interesting reading material:

http://www.epcos.com/inf/20/40/db/tec_03/01070134.pdf

 

[Clarence_A]

PaulCsouls wrote

This may be interesting reading material:
http://www.epcos.com/inf/20/40/db/tec_03/01070134.pdf

This document IS interesting, and it seems to be good technically.
It will be useful if someone can get past the poor phrasing and
reverse orders of many statements.

Perhaps the inserted reverse logic was intended as humor?

 

[Ian]

It was in an AGC circuit. Basically a low pass filter with the caps in
series to ground two 47uF caps and 1k resistor so a 4Hz 3dB. Not very
demanding with respect to distortion or tolerances. I was just curious
about how the limitations of the configuration.

Paul C

A better method would be to put a bias resistor to a sufficiently large bias
voltage
connected to the junction of the 2 capacitors. That way, both are always
operating
correctly biased.

Regards
Ian

 

[PaulCsouls]

This may be interesting reading material:

http://www.epcos.com/inf/20/40/db/tec_03/01070134.pdf

Interesting. They call for the diode protection across each cap.

Paul C

 

[PaulCsouls]

Any link ?


Oh well. That means almost every piece of current pro and consumer audio is
built wrong then !


Alumiums are good for a few 100mV reverse volts typically. Perfect for split
supply audio apps.


What 'frequency characterisitcs' ? Bias affects bandwidth ? Some very strange
ideas here.

Maybe this Mr Sinclair is one of those audiophool gurus ?


Graham

The bias / bandwidth idea is one of my fool ideas. Ian Sinclair's book
is recommended by Bob Pease in his Troubleshooting book, so it has
been proofread by someone with some experience in the field.

Paul C

I'd be really smart, if I wasn't so stupid.