Harald Bode Barberpole Phaser Questions

[Sean Costello]

Hi everybody:

Yesterday, I wrote:

>   2) Is it necessary to have a fully functioning, lab-quality frequency
>   shifter for this effect, or is it possible to cut corners if all you want is
>   a barberpole phasing effect?  Do you need such a complicated setup for the
>   dome filter, or could you construct a dome filter with less stages and less
>   precise components? 

Don Tillman replied:

>The accuracy of the quadrature filter is more important with
>audio-frequency frequency shifting, because having a little shift-down
>mixed in with your shift-up can be a problem.
>
>Wait a second...  What if it was a *completely* inaccurate quadrature
>filter?  Well, that means you'd have both up-shift and down-shift
>mixed together.  But that's just a simple ring modulator.  If you
>replaced the frequency shifter with a simple ring modulator you'd
>effectively have a rising barberpole mixed with a falling barberpole.
>That might be nice effect.  Sure it's cooler to have rising and falling
>barberpoles as a stereo pair, but if you're on a budget this might not
>be a bad thing.  Hey Juergen, you're set up for this experiment right
>now...

Excitement prevails over the simplicity of the idea!  However, fellow
Shallow Alto resident Harvey Devoe Thornburg replies with:

>Wouldn't this lead to amplitude modulation, because you are
>essentially mixing some of the upshift with downshift, making 
>part of the circuit act like a ring modulator? If you 
>encountered even slight variations in the perceived signal
>level this could be problematic (or useful depending on 
>your goals).

And gloom settles over all.  But then, Sean Costello, the dim-witted "idea
man" (read: I don't know enough to build the gear to test these ideas
myself), comes up with an idea:

In a frequency shifter, the signal passes through a "dome filter" (two
parallel allpass filters, which maintain a constant 90 degree phase
difference).  The output of each signal is multiplied by a waveform from a
quadrature oscillator (one allpass output is multiplied by the sine output,
the other by the cosine output).  The outputs from the multipliers are
summed by two parallel amplifiers: one adds the outputs, the other subtracts
the outputs (I think - I'm doing this from memory here).  Now, and here's
the tricky part (and quite possibly the stupid part):

Leave out the dome filter.  One of the multipliers is used to multiply the
straight, unfiltered signal with one of the quadrature oscillator outputs
(let's say sine). The other multiplier is used to multiply the output of the
static allpass filter (the big multistage one used in Bode's patent to
emphasize the phase shift sound) with the other quad osc. output (the
cosine).  The signals from the two multipliers are summed as above.

What would this do?  I would doubt that the result would be simple amplitude
modulation; the profound phase differences between the straight signal and
the output of the static allpass filter would prevent this, I think.  But
would it be a barberpole effect?  Would it sound like phasing at all?

I have no idea if this will work.  Maybe the result will sound more like
conventional phasing.  However, even if this is the case, it would be a very
economical way of creating multistage phasers - just add static op-amp
stages in the allpass filter to increase the degree of phase shift.  Use a
cheap quadrature oscillator (perhaps a state-variable filter, with diode
limiting of the feedback and high Q, made from three stages of a 4069
inverter chip as described by Don Lancaster in Electronotes #102), the 1496
multipliers that Juergen uses, and see what happens.

If anyone can create a patch with their frequency shifter components (or
just a quadrature oscillator, two multipliers, two summers, and a static
allpass filter), please let me know what it sounds like.

Thanks,

Sean Costello (who will be able to start building in February, when he moves
into a house with a nice workbench.  My apartment is too damn small.)