[sdiy] SVF & phase

Wed Nov 27 18:46:26 CET 2002

Oooohhh  I'll answer that one (right, or wrong)

The loss of high frequencies from tape head
asimuth error 'is' a phase shift effect in itself.
One edge of the head (leading) is phased forward
of the trailing edge... and cancellation occurs as
magnetic fields of both polarities are read by the
coil.

The effect only occurs for frequencies whose wavelength
(on the tape) are close to the headgap spacing. 

In its severest form... it actually will make a comb filter
and selectively notch those frequencies which are closest
to the headgap spacing.

but fooey on you, JH... we were counting on you telling us
how to MAKE that zero phase shift SVF...  With no group delay
I could make a dome filter and shift the guitar frequencies up
into ultrasonics... where I would do F/V conversion and have
a near zero delay guitar synth.

Oh well maybe on some kinder, gentler planet  ;^)

H^) harry

--------Original Message--------  

From: jhaible at debitel.net
To: "patchell" <patchell at silcom.com>, "Grant Richter" <grichter at asapnet.net>
Cc: "mark s" <n0nspaz at mindspring.com>, <synth-diy at dropmix.xs4all.nl>
Sent: Nov 27 2002 17:32
Subject: Re: [sdiy] SVF & phase

>>     We used a crude F->V converter, and when the voltage (frequency) got
>up to a
>> certain point, it would trip an analog switch preventing the signal from
>getting
>> through.  On the Network Analyzer it looked like the perfect filter,
>however, if
>> you were to use this in practice, it would not be a good filter at
>all...still, I
>> cannot say you can't make a filter with zero phase shift.
>A filter in the EE sense (not in the musical sense, where nonlinearities
>are an important part of the sound) is a *linear* circuit. (It's certainly
>a matter of definition, but that's the definition I learned.)
>Your circuit is not only highly nonlinear, it's also unable to separate
>the frequencies of a two-tone signal, letting one pass, but not the other.
>Which is a result of being nonlinear, of course: In linear circuits
>you have the principle of superposition (sp?), i.e. you can look
>at each spectral component independently. You loose this with
>nonlinearity, and the circuit you described is the most excellent
>example to show this. (They didn't show this example in college
>just for that very reason, did they ?!)
>> I seem to recall from
>> the same class that it was mention that certain digital implementations
>could
>> realize filters that you could not realize with standard components.  But,
>we
>> never got into that.  But, if you are using good old poles and zeros, you
>will
>> alsways have phase shift.  Although, like I said before, having no phase
>shift I
>> would think, would make the filter sound uninteresting anyway...
>What you can do in digital is *linear* phase (constant delay).
>Zero delay or zero phase is not possible.
>Now let _me_ play the devil's advocate:
>There is a remarkable filter effect that attenuates high frequencies and
>might not result in a phase shift: The reproduction of an analogue recording
>over a mis-aligned tape head (wrong azimuth):
>Questions:
>(1) Does this really cause no phase shift?
>(2) Of course it does *not* go against the above rules, even
>      if it doesn't cause phase shift. Why ?
>JH.