Negative Feedback Caps - Bipolar OK?

[M1K35R]

Just a quick question. Is there anything wrong with using bi-polar capacitors (ie... 100uf 50v) in the negative feedback circuit of an amplifier? I have a Spectro Acoustics 202 (older version) that needs them, but all I have are bi-polar. Any comments on whether this is not advisable and/or maybe why would be greatly appreciated.

 

[ChristianThomas]

Nothing wrong in principle, and at that value you have little choice. And if you use bipolar, which you don't necessarily have to, you will be doing better than using polarised electrolytics.

I assume this is for a subwoofer or something similar, in which case you'd be very hard pushed to hear any difference at all.

 

[Zero D]

If the biploars are for eg, speaker xover types, then they will most likely be 2 x electrolytics in series, inside one case !

Eg, a 100uf 50V biploar = 2 x 200uf 25V electrolytics in series. They are either connected + to + or - to - inside.

You could make your own this way

 

[AndrewT]

The DC blocking capacitor in the negative feedback loop should have virtually no voltage across it.
No DC voltage and no AC voltage.

Based on this any polarised, or non polarised capacitor can be used here.

Just make it big enough such that the little bit of VLF (bass) that gets past the high pass input filter does not develop a significant AC voltage across the DC blocking capacitor.

 

[Fast Eddie D]

The DC blocking capacitor in the negative feedback loop should have virtually no voltage across it.
No DC voltage and no AC voltage.

If the circuit is properly designed, this is indeed the case. This is why the cap will typically be so large (like 220 uF when the math says 10 uF will work fine). This design logic is industry standard. Furthermore, I have seen them still working fine after 30 years. But I have seen a couple amps where this cap was intermittent and the complaint was "gain goes up and down" and "sounds gritty" (maybe from the oscillation?).

Bipolar electrolytics typically have two cathodes and no anode. They would certainly work in this application but I would be inclined to oversize them as well.

You can use "hi-fi" poly caps etc in this application but unless you set the cutoff very low they might introduce more distortion than a big oversized polar electrolytic!

In general, the feedback circuit is not the place to have the dominant pole of the amplifier. The input is the place for that.

 

[ChristianThomas]

The DC blocking capacitor in the negative feedback loop should have virtually no voltage across it.
No DC voltage and no AC voltage.

Based on this any polarised, or non polarised capacitor can be used here.

Just make it big enough such that the little bit of VLF (bass) that gets past the high pass input filter does not develop a significant AC voltage across the DC blocking capacitor.

This is the question. Are we talking about a capacitor actually in the loop of the LF unity gain capacitor? I suppose it does seem a more likely size for it to be the second.

The NP capacitors do create significantly less distortion - see Cordell's book. I don't know how this stacks up against a properly biased electrolytic though.

 

[jcx]

bi/nonpolar caps constructed with full thickness oxide grown on both foils have lower distortion than back-to-back biased polar Al C according to Bateman's "Capacitor Sound" series measurements

 

[Fast Eddie D]

bi/nonpolar caps constructed with full thickness oxide grown on both foils have lower distortion than back-to-back biased polar Al C according to Bateman's "Capacitor Sound" series measurements

Indeed. We just hashed this out in a couple of threads.

 

[Fast Eddie D]

This is the question. Are we talking about a capacitor actually in the loop of the LF unity gain capacitor? I suppose it does seem a more likely size for it to be the second.

If you design your amplifier so that there is virtually no voltage across the cap, no AC and no DC, it will not introduce distortion.

I will use some numbers for illustrative purposes. You design your amplifier so that the input cut off is 1.5 Hz. Then you select your feedback capacitor to provide a cut off of 0.25 Hz. There will be virtually no distortion introduced by the cap, because there will be virtually no voltage across the cap, ever.

Ideally poles should be separated by a decade, but in practice this proves to be cumbersome or impossible. 3 to 5 times is a good comprimise.

 

[M1K35R]

Thank you all for your excellent replies! It definatly is the neg-feedback cap, it does have about 35vdc across it. I have the schematic of a later model 202 that is virtually identical with the exception of in input op amp that isn't in mine. It sounds great so far and dc offset is stable at .15mv (give or take a .0mv or so).

 

[Fast Eddie D]

Thank you all for your excellent replies! It definatly is the neg-feedback cap, it does have asobout 35vdc across it.

In a typical amplifier design, this cap has essentially zero volts across it. So if the amp is working properly, it is either a single supply amp (probably not) or else the cap is returned to the negative rail via a bias/decoupler circuit.

Did you measure 35 volts? If so you have no choice other than a polar cap. Did you have the VOM "grounded" to the negative rail?

 

[Fast Eddie D]

Just make it big enough such that the little bit of VLF (bass) that gets past the high pass input filter does not develop a significant AC voltage across the DC blocking capacitor.

This is where enthusiastic DIYers go wrong. They see a 220 uF electrolytic and think that it must sound terrible. Then they do some math (because they're smart ) and they realise that if they put those 4.7 uF "poly" caps in there that the low frequency cutoff will be 6 Hz which of course is "perfectly acceptable."

What have they really done? They have re-engineered the circuit so that it produces more distortion than the original circuit! Why? Because now there is significant voltage across the cap, and any distortion produced by that voltage will now be introduced at the non-inverting input as distortion!

Engineers do not choose their poles randomly. They carefully juggle them so that they work properly for the intended application. You can shuffle them around a little bit if you know what you're doing, but you have to understand the consequences of it.

 

[M1K35R]

DVM grounded to the neg rail, I show +33.6vdc at one end of the cap. It is bipolar, 100uf 50v, and so far it sounds great. The cap is not heating or showing any sign of stress. Do you think I should get a polarized cap anyway?

 

[M1K35R]

P.S. : ACROSS the cap, it measures +35vdc...