Mark, Peter, Stephen: Time-domain Quiz (?)

[puretube]

Check this out:

imagine sending a signal (1kHz) into a delayline, say 100ms;
2 bilateral analogue switches (4066):
one ("S-in") at the input of the delayline (0ms),
the other ("S-out") at the output (100ms);
these switches are operated alternately on/off;
both switches (when either is  "on") conduct their signal
into a summing point where the signal is made audible.
If S-in is closed, you get the 0ms (dry) signal,
if S-out is closed, you get the 100ms (wet) signal
at the summing point.
Now if you don`t close/open the switches for a constant time each
with DC,
but "chop" the control-voltages well above audio-frequency
(say with 100kHz),
you`ll get the chopped 0ms (dry) signal at the output
(after suitable lowpass-filtering),
when the (PWM) dutycycle is 100% for S-in & 0% for S-out;
also you`ll get the chopped 100ms (wet) signal at the output, when
the dutycycle is 100% for the S-out & 0% for S-in.

Now my question: what do I get at 50% duty cycle?
(Half of the time 0ms, half 100mS): imaginary 50mS or cacophony??

(The above circuit is known to produce vibrato, when the delayline
delays the signal for 360 degrees at a given frequency,
and the pulsewidth is modulated by a sine- or triwave LFO;
US-patent 3260785 by W.H. Krug, filed Oct.`63 - Tubes of course...).
But
what happens with real "long" delay times instead of just a shifted (delayed) phase ?

[Peter Snowberg]

I'm going to guess that with a long time and an ultrasonic switching frequency the result at 50% would be more like a fader set half way between dry and 100ms delayed audio. I'm thinking slapback echo. I don't see how you could effectively get the equiv. of a 50ms delay or cacophony from that.

Am I right? Do I get a doggie biscuit?

If you fed it with a pure wave (from a tone generator, no envelope, just a tone) and modulate the delay time I'm guessing you would have a chorus effect.

That's my guess.

Take care,
-Peter

[puretube]

thanks, Peter, but: sorry, no biscuit; biscuits only go for the 50ms answer!

Fun aside: I wasn`t quite sure myself,
but now you said it: the circuit will very probable turn
out to be like a wet/dry balance pot, when the duty-cycle
is varied - my thoughts got stuck here...

but then again: what`ll it sound like, when I wiggle this
pot between dry and wet 5 to 10 times per second:
time-warble?
(here again: with a simple 360 deg. phaseshifter instead of
real delay, I get the "vibe"-thing...)

The idea behind it is not to have to change the delay-time (=clockfreq.), but only having (multiple!) pwm circuits do the
scanning to get different pitches.

(Similar things can be done (and have been so) with multi-tap
delays and a couple of VCAs scanning the taps to and fro
like in the good old mechanical Hammond scanner-vibrato;
but a single fixed delay sounds too nice of an idea to achieve
"string-ensemble" sounds with...)

[Paul Perry (Frostwave)]

In my opinion, if the delayline is 100ms, then it is impossible for output to be delayed by anything else.
you might get strange aliasing effects, phase strangenesses, but certainly no other delay times.

[puretube]

guess I`ll breadboard something, and check out the phase-strangeness or stretchiness that will sound in the transition between fixed100ms and 0.
"Virtual PhaseVerbTrem ?"

curious! (aliasing from the "chopping" is assumed to be completely filtered out...)

[R.G.]

You can't get anything except the pot-blend answer.

You're chopping two signals one delayed by 0, one delayed by some amount. The higher the chop frequency (i.e. the better the chopper does sampling) the more it approaches a linear mix. For all practical purposes, chopping over Nyquist frequency (2X highest frequency of interest) is good enough to dealias back to a linear mix.

Principal of superposition says that for any linear processes, the sum of two processes is equal to the sum of the results of either alone. And chopping is a linear process under the conditions you have described.

Specific frequencies and sampling rates might make other answers look like they're happening if the sample rate is not high enough. Under those conditions, chopping is not linear.

[puretube]

OK, thanks to Peter, Paul & R.G. for the answers so far!

What it boils down to, is: what kinda "vibrato" will it sound like,
when a 0ms signal is tri-wave amplitude-modulated by an LFO, and "mixed" (added) with a 100ms signal, which itself is amplitude-modulated with the inversed tri-wave of that LFO.

(btw.: the vibrato in a vox AC30 is done similarly by oppositely modulating 2 signals, which are 90 degrees phase-displaced).