Looking for advice on replacing DC blocking input caps on audio preamp

[nez]

After searching through a variety of audio forums with mixed results (and suffering through a lot of mysticism about capacitors in audio circuits), I thought I'd just ask this here.


I'm in the planning and ordering process to re-cap and repair the power supply for an old computer speaker set I have as a learning project, and I thought I'd consider the preamp module as well.

On the preamp, the only electrolytics are a pair of back-to-back 10uF 25V polar caps on each of the 4 input channels to block DC content.  They likely aren't really bad yet (I haven't removed them from the circuit to check on my DE-5000), but they are 21 years old and not exactly a top brand (G-Luxon) so I figure I may as well throw in some replacements with my Mouser order.

I have little experience with these types of things so I'm wondering what audio aspects I'm ignorant of that may be important.

For polar caps, the originals are about 4 x 7mm in size, with max space for 5 x 8mm (plus a tiny bit extra for lead bending beneath).

I can select various polar caps that fit fine and match the specs fairly well, however I'm wondering about bi-polar caps instead.  I found some 4.7uF bi-polar electrolytics that are 5 x 11 that should fit sideways in the space of each existing pair of polars.

Is there any benefit to switching to bi-polars in this case, or is it really just equivalent to back-to-back polars?

A second uncertainty that I have regards a post I read somewhere else saying that connecting the shared negative node of the two polar caps to the negative rail (through a high-value resistor) to put a charge across them can (1) help the electrolytic caps last longer, and (2) puts them in a more linear range...

It's easy to do, but is this true?  Where can I learn about how much (or little) this actually matters?  Are there negatives to doing it?

I hope my explanation isn't confusing.
Thanks!

[Gyro]

After searching through a variety of audio forums with mixed results (and suffering through a lot of mysticism about capacitors in audio circuits).

Yeah, you get that. 

As you rightly say, the input caps are very unlikely to be bad, they don't have to handle any ripple current. At the same time, they will be small can types and could suffer a natural drying out of electrolyte over a 21 year period, depending on the quality of their seals.

The fact that they put two caps back to back implies that the designer wasn't terribly sure what direction any input offset voltage would be (as you would, for instance in a transistor common emitter input stage).

Is there any benefit to switching to bi-polars in this case, or is it really just equivalent to back-to-back polars?

Bipolar caps are effectively two caps back to back (in operation, even if not in physical construction - both electrodes have an oxide layer). Their only real advantage is space-saving.

A second uncertainty that I have regards a post I read somewhere else saying that connecting the shared negative node of the two polar caps to the negative rail (through a high-value resistor) to put a charge across them can (1) help the electrolytic caps last longer, and (2) puts them in a more linear range...

Having a definite bias voltage and no ripple current to deal with is pretty much nirvana for an electrolytic, pampering on a grand scale. Whether it makes any difference to linearity etc is highly debatable (and will be, at great length, on audio forums), but it will tend to make them last much longer. Operation at zero or slightly reversed bias is akin to storage life conditions. There is absolutely no harm in biassing the junction between the capacitors if the implementation is easy. The worst you might experience is a switch-on thump if this isn't muted by the amp.

[Doctorandus_P]

If you do a bit of research on storage conditions of electrolytic capacitors, then you find that the oxide layers do degrade while lying on a shelf, and this results in leakage currents when a voltage is applied.
This voltage both "repairs" the oxide layer through oxidation, and the leakage current heats up the capacitor.

Reforming electrolytic capacitors is a real thing. Especially with high capacity and high voltage capacitors current is limited, and voltage gradually increased to re-build the oxide layer, while at the same time keeping dissipation low and making sure the leakage currents can not heat the capacitor too much.

----------------
As to your application.
Electrolytics have no purpose in the signal path of "above average" low level audio.
If you're concerned about distortion and other effects, then just replace your two anti-series 10uF capacitors, with a single 4.7uF film capacitor, or a few smaller capacitors parallel.

[David Hess]

Is there any benefit to switching to bi-polars in this case, or is it really just equivalent to back-to-back polars?

The result is the same.  I am not aware of any advantage or disadvantage to either method except for minimizing the number of different parts kept in inventory.

A second uncertainty that I have regards a post I read somewhere else saying that connecting the shared negative node of the two polar caps to the negative rail (through a high-value resistor) to put a charge across them can (1) help the electrolytic caps last longer, and (2) puts them in a more linear range...

There is a difference because electrolytic capacitors have a significant voltage coefficient of capacitance leading to high distortion in audio coupling and filtering applications.  In a filtering application where the value of capacitance is important, biasing a pair of back-to-back capacitors is done.  In coupling applications, this is not necessary because a much larger value can be used so the voltage variation is minimized.

[nez]

Thanks for all the responses!  It makes much more sense to me now.

I'll stick with similar polars so I have the option to connect the mid-point to the negative rail if I feel like gilding the lily (Edit: or for this device, more like gilding the dandelion).

There is a difference because electrolytic capacitors have a significant voltage coefficient of capacitance leading to high distortion in audio coupling and filtering applications.  In a filtering application where the value of capacitance is important, biasing a pair of back-to-back capacitors is done.  In coupling applications, this is not necessary because a much larger value can be used so the voltage variation is minimized.

Is there a way to roughly find out the voltage coefficients of capacitors? I don't recall seeing anything like that in the datasheets.


Edit: (almost forgot this part)

As to your application.
Electrolytics have no purpose in the signal path of "above average" low level audio.
If you're concerned about distortion and other effects, then just replace your two anti-series 10uF capacitors, with a single 4.7uF film capacitor, or a few smaller capacitors parallel.

I had briefly looked at film capacitors a few days ago, but the ones I saw seemed too large to fit in the tight space of the preamp enclosure and surrounding components. I do like the idea so I'll double-check the size options and what clearance I really have.

[David Hess]

There is a difference because electrolytic capacitors have a significant voltage coefficient of capacitance leading to high distortion in audio coupling and filtering applications.  In a filtering application where the value of capacitance is important, biasing a pair of back-to-back capacitors is done.  In coupling applications, this is not necessary because a much larger value can be used so the voltage variation is minimized.

Is there a way to roughly find out the voltage coefficients of capacitors? I don't recall seeing anything like that in the datasheets.

I do not think I have ever seen it specified in datasheets except recently for high dielectric constant ceramic capacitors which might have a graph showing nominal capacitance versus voltage.

In audio applications instead of measuring the voltage coefficient of capacitance, distortion is measured because it is more relevant.

Operational amplifiers have the same problem with variation of input capacitance as their common mode voltage changes, so low distortion circuits take this into account by equalizing the source impedance at the inputs of the operational amplifier, using bootstrapping, or using the inverting configuration.