Simple Way To Reduce Distortion From Electrolytic Capacitors

[jdiamantis]

Samuel:

Just for grins, try a pair of caps in anti-series, + ends connected to one another, and a 1M resistor connected from the junction between them to V+. I'd be interested to know how that compared with anti-parallel.

Peace,
Paul

There were several examples of this in pro gear back in the last century. I personally used this topology, with a parallel film across the electrolytics in modifying some broadcast consoles back then. The thought was to keep the caps biased, to prevent reverse voltage. Seemed to work well. I do recall measuring the distortion back then, but don't remember any specifics....

jD

 

[Dan Kennedy]

I am working on a new design where this idea fits the circuit's implementation.

It's the old standard two opamp with the pot in the middle as gain control
for both stages, a la Trident 80, if I remember right.

The first stage needs to be AC coupled, so I will report back to class on
how it holds up.

 

[JohnRoberts]

If we are talking about canceling capacitor nonlinearity, how about putting similar caps in series with the input and feedback resistors of a simple inverter? Of course we'd also have to provide a DC feedback path but that could be suitably below the audio bandpass.  A high value resistor across the feedback cap only is not a huge cost burden. 

Alternately for a similar circuit complexity (as the added dc path) putting a single cap between the opamp's inverting node and junction of the input and feedback resistors would also reduce distortion by a different mechanism than cancellation.


JR

 

[Jim Zuehsow]

SSL uses two capacitors, positive to positive with a 100K or so resistor to the + rail to bias them for the sole purpose of making the capacitors more "transparent". But wait a minute.... aren't all these different distortions part of what gives certain preamps their "color"? We've all heard stories about someone "improving" an old Neve module by replacing all those crappy old caps they used, only to totally destroy the thick Neve sound. Most of the time it seems, if you give someone a really clean preamp, they complain that it sounded "unexciting" and had no character. Granted, you need both clean and not so clean preamps to cover all bases these days, and to be honest, I don't really think I could hear some of the distortions that are claimed to exist, even if compared side by side to something  different, unless it was really obvious.

 

[Morning_Star]

I am working on a new design where this idea fits the circuit's implementation.

It's the old standard two opamp with the pot in the middle as gain control
for both stages, a la Trident 80, if I remember right.

The first stage needs to be AC coupled, so I will report back to class on
how it holds up.

I'm interested in this.  For some reason I love the Trident S20 mic pre.  Will it be the same sound or just similar design?

 

[Dan Kennedy]

Similar topology, much different implementation.

 

[Samuel Groner]

It is not entirely clear to me how your setup is. Are the capacitors directly across the output of the generator? If yes this would severely load the output and perhaps invalidate the measurement results. You should add say a 300 ohm resistor in series with each output pin.

I've measured 10 capacitors with every possible pairing and found quite consisten improvement. One capacitor gave less improvement with all combinations; it also measured differently alone. Also I've correlated the distortion improvement with capacitance (mis)match, but there was no significant correlation.

Samuel

 

[abbey road d enfer]

If we are talking about canceling capacitor nonlinearity, how about putting similar caps in series with the input and feedback resistors of a simple inverter?

The distortion mechanism that Samuel describes is directly dependant on the AC voltage across the cap. As a consequence, this form of distortion is particularly measurable on outputs, where the load is quite often variable.
Depending on the distortion mechanism, a cap in the negative leg may cancel distortion.

Samuel, have you qualified the distortion? Is it 2nd or 3rd harm?

 

[Samuel Groner]

You have the RC as a LPF, correct?

Yes. As far as I know a HPF would give identical results.

Is it 2nd or 3rd harmonic?

The parallel arrangement gives a mixture of odd- and even-order harmonic distortion. The anti-parallel combination naturally cancels the even-order harmonics, with no change for the odd-order harmonics.

Have you looked at the residual of your measurements? Just to make sure the increase in THD+N is due to distortion and not noise. What is the residual of the analyzer? Theoretically there might be some distortion cancellation with the analyzer residual, and a reduction in distortion from the capacitors might lead to less cancellation and hence higher total distortion. Also I'd check with a smaller resistor to move the onset of distortion upwards in frequency.

Samuel