[sdiy]Do you need to add more functions to your favorite LP4P VCF ?

[m.bareille at free.fr]

Bonjour ,

Long email , answers in the text ...

> I was curious about the performance of this circuit and ran spice on a
> few of the configurations.  AC analysis shows the response is very
> sensitive to component tolerance,

Like on most of this kind of circuits where phase and antiphase signals
are summed. The parts tolerance introduce errors who affect the quality
of the filter shape...


> and using 1% resistors can cause
> variation up to 30 dB at low frequencies in the 4P BP configuration (I
> set the RC value in the cells to 1K and 0.5u, and stepped the ABCD
> resistors, for example 100K simulated from 99K to 101K in steps of
> 500).  If I get a chance I'll spice all the configurations.

For sure if resistors are matched to 0.1% or 0.01% the result will be better and
closer to the awaited perfect filter shape, at least if the filter itself is
homogen...Matching 99K + 101K is also 2% of error introduced in the circuit ...
>
> I'd be interested in seeing the measured response of the real world
> circuit in various configurations.  I would think that the variation
> seen must cause an audible difference.  I guess it depends on what you
> want, but does this circuit need even tighter tolerance parts, or trimpots?

You are right: this depend a lot of what the user want to do
1) to add more functions to a VCF to play music ...
2) to make a laboratory filter with the required accuracy for measurement
purpose or like...

I think that 1% tolerance is a minimum to ensure a correct result at least for
musical purposes. This is like this into the Matrix1/Xpander schematic and from
what i know nobody has complained about this filter :)

The builder who want more accuracy  will probably choose 0.1% or 0.01% resistors
to avoid matching them or use multi turn cermet trimmers . I believe that at
this level of accuracy it is also very important to compensate all matching
errors of the VCF stages themselves...Music LP4P VCF able to receive the
expansion  board are generaly builded with 5% or 10% tolerance parts ( R& C)...

Ok, i am going to rewire my sound card, record samples with theVCF as this, take
pictures of each of the 16 curves with the spectrum analyser and add them to my
page. May be i will blemish seeing curves shapes, who knows ?

But as i often say : my best and most accurate measurement tool are my ears and
from what i have heard from this multimode VCF, i wont go back to the simple
LP4P never mind if the BPs or notches are not absolutly perfect! Errors are
war's nerves of analog sound :)

Cheers,

Marc B.