[sdiy] Hand-matching capacitors for filter stages

[thx1138]

On 2/9/10 1:17 PM, "David G. Dixon" <dixon at interchange.ubc.ca> wrote:

>> Well then, this is directly my question, what does better imply?
> 
> Precisely!  I can't see how one cap or another of similar capacitance would
> make a filter sound "better" or "worse" than another one.  The only thing I
> can think of is that better value matching among the caps in a filter would
> make the filter cutoff track better, and would cause the circuit to
> oscillate in a more predictable fashion at the critical gain threshold.
> These are certainly desirable properties, but only relate to the
> value-matching of the caps, not their dielectric properties (within
> reasonable limits).
> 
> IMHO, if someone could detect an audible difference between otherwise
> identical filters constructed with similarly well-matched and equal-valued
> caps of polystyrene, polypropylene, or silvered mica, then that person has
> superhuman hearing.
> 
>> Without hearing it or seeing a measurement in parallel how are we to
>> ascertain sonic quality or difference when caps are inserted in circuit?
>> 
>> Can someone describe the sonic characteristic before or after cap
>> changeout?
>> 
>> Or is there some other parameter or parameters we are measuring here
>> besides acoustic properties such as frequency or resonance stability when
>> Q is turned up etc.
> 
I have to admit one thing here.

Since I designed analog circuits back in 1970 to 1980 period, I notice that
Capacitors have changed somewhat as well.

We would never use a 5% cap back then at E-Mu as they were to cost
prohibitive.

It appears after checking around on the web in parallel to this discussion
that I have found that pricing on tighter tolerance capacitors is quite a
bit better than I remembered it.

Subjective listening test might be interesting but I tend to design for
stability over freq. And temp over tracking range.

It might be interesting to see what these newer products would impact my
older Modular designs.

Perhaps it is time to rebuild some of these modules with other materials to
see what works and does not work as well.

DSP response is so predictable that it takes the fun out of analog design in
general. 

Modeling filter characteristics in software is becoming popular in guitar
products I see.

I recently purchased FilterShop from Linear X and it is quite nice for
design work if you check it out.

Regards,

Terry