[sdiy] SVF High Pass Mode not Cutting Off

[Magnus Danielson]

On 10/30/2011 05:17 AM, David G Dixon wrote:
>>> Thanks for the help, I tried lower integrator cap values (220pF
>>> instead of 470pF) but without any change. I have plenty of
>> Freq range.
>>> on the pot. A sawtooth waveform is "squashed" to a flat
>> line as Freq
>>> is turned up but a very sharp spike is left at the trailing edge of
>>> the saw no matter what the VCO freq is. Any thoughts?
>>
>> Isn't that supposed to happen?
>
> I don't think so, at least not to an annoying extent.  As I said before, I'm
> pretty sure that the problem here is imperfect cancellation of the LP and
> input signals at high cutoff.  The cure is definitely to tighten up the
> resistors on the input summer, and even perhaps to make one a tad smaller
> than the other and add a trimmer in series which can adjusted to perfect
> cancellation.  The problem is that even a few mV of bleedthrough will be
> quite audible.  I had this problem with the Korgo, and we fixed it (mostly)
> with 0.1% resistors.

The slope of signal overtones "goes the wrong way" typically, the -6 dB 
or -12 dB slope combines well with the -12 dB slope of the filter to 
"close" quickly, a -6 dB slope with a -12 dB slope becomes -18 dB slope. 
However, a -6 dB slope with a +12 dB slope (high-pass slope increases 12 
dB per octave) creates a +6 dB slope... which closes slowly, so you will 
have to raise the filter frequency much higher to loose the same amount 
of total signal power.

A -12 dB slope of overtones cancels with a +12 dB slope, but the level 
is being controlled by separation of filter frequency from fundamental 
frequency.

So... for equivalent effect the high-pass filter needs to be steeper 
than the low-pass filter.

Cheers,
Magnus