Eliminating delay chip "portamento"

[Taylor]

I've done some experiments with a sequencer changing the time of a PT2399 delay. It's fun, but there is an inherent glide between times, which creates a glide in the pitch change that happens changing times. This happens even when the resistance that sets the time is changed instantaneously.

I wanted to try a square wave LFO changing the time, which would give something like a continuous pitch shifter with symmetrical up and down shift. The problem is that the shift needs to be a real square in order to be heard as something like constant pitch shift, instead of just deep vibrato. The PT2399 doesn't seem able to do this without the glide.

Is there a chip that doesn't do this, but instead can really change instantaneously? Or is there a way to make the 2399 do my bidding?

[Processaurus]

Linear change in delay time (a ramp) produces a constant pitch shift.  a triangle LFO modulating delay time will have a squarelike sound to the pitch shifting, making an abrupt change between sharp and flat as it changes direction.  A square wave lfo will not produce pitch shifting, because you aren't producing a change in delay time during the flat parts (the majority) of the cycle, it will just go between two delay times with a glitch in between as the LFO turns on and off.

To get a sequenced sounding pitch shift, you'd need some kind of special multi slope ramping LFO.

[cpm]

so it would be easier to do with a chip that uses an external clock?, like analog delay ics, that you would switch  from two or more clock signals at different rate.

[R.G.]

so it would be easier to do with a chip that uses an external clock?, like analog delay ics, that you would switch  from two or more clock signals at different rate.

Very much so.

Google "Shepard function" or "Shepard tone".

[Taylor]

Linear change in delay time (a ramp) produces a constant pitch shift.  a triangle LFO modulating delay time will have a squarelike sound to the pitch shifting, making an abrupt change between sharp and flat as it changes direction.  A square wave lfo will not produce pitch shifting, because you aren't producing a change in delay time during the flat parts (the majority) of the cycle, it will just go between two delay times with a glitch in between as the LFO turns on and off.

To get a sequenced sounding pitch shift, you'd need some kind of special multi slope ramping LFO.

Hmm. With all due respect (I'm sure you know a ton more about electronics than I do), I've already done this in a DSP program with a square wave, and it works as I want it to. Likewise, I've already made a sequenced pitch shifter with real analog components, so I already know that it works with a regular 4017 sequencer.

Perhaps there's some confusion about what I'm talking about. I don't want Sheperd tones ( constantly rising or falling shifts) I want to shift up by a specified amount, then shift back down by that amount over and over. Using a triangle wave just creates vibrato. I want to use a square so that the portamento between the down shifted pitch and the up shifted pitch is as small as possible. I'll put some clips up to show what I mean.

[cloudscapes]

I know what youre talking about, though I wouldnt know how to fix it.

currently have breadboarded a vibrato based on the 2399 and an AVR chip that cycles through different LFO shapes, including a perfect squarewave. I also get the portamento.

[doug deeper]

the portamento is due to the delay catching up with it self, if the lfo is fast enough you wont get it, but i think its unavoidable at slower rates.

[Taylor]

Here's a clip I made with Plogue Bidule:

http://ironether.com/square%20delay2.aiff

It's a square wave LFO controlling delay time, with the delay signal 100%, no dry signal.

I start with low LFO depth. This equates to small pitch shifts. I then turn the depth up, which makes the low pitch shift lower, and the high shift higher. You can hear how this creates a shift that is (momentarily) stable, then it goes down and is (momentarily) stable at that pitch.

Then, at the last 1/4 of the clip, I switch to a triangle LFO. You can hear that this is just a deep vibrato, instead of a shift that stays steady for a split second. I don't want deep vibrato.

Now, I haven't tried doing this with the PT2399, but I'm already pretty sure that it will smooth the square off enough that I'll just get vibrato instead of that glitchy shift. With this new explanation of what I want, would BBDs be the way to go here? I've got some sitting around...

[Taylor]

the portamento is due to the delay catching up with it self, if the lfo is fast enough you wont get it, but i think its unavoidable at slower rates.

But even using a 4017, which presumably changes resistance instantaneously, it does the glide.

[jacobyjd]

hmm...that's interesting--I was planning on doing the same thing with a 4017. I figured the 2399 would react the way you're expecting it to...strange, because my Experimentalists Anonymous Space-Time Continuum delay (PT2395 based) reacts the way you're wanting...just not with a 4017, rather with a longer initial delay and manual turning of the knob. I'll have to play with my rebote to see if I can get it to do the same thing manually. Weird...

[earthtonesaudio]

so it would be easier to do with a chip that uses an external clock?, like analog delay ics, that you would switch  from two or more clock signals at different rate.

+1.  The fastest change in delay time you can get with the PT2399 is limited by the internal capacitances of its VCO.  If you had an external clock you could make the transition much faster, limited only by the rate the external clock changes and parasitic capacitance in the layout.

[Taylor]

Does that mean I'd need 2 clocks? Or is it possible to "bend" clock time with an LFO somehow?

If using 2 clocks, what would be a good way to swap between them at very fast rates (around 10hz)? Would the 4017 be good here?

[earthtonesaudio]

One high frequency clock and one LFO, which could be considered a "clock" is all you'd need.  You could wire up a couple 40106 Schmitt trigger inverters as oscillators, and have one at 10Hz controlling another set up as your delay chip clock.  A simple way, but tricky to adjust, would be to put add a resistor and transistor in parallel all in series with the timing resistor for the HF clock.  Then use the LFO as the control signal for this transistor, which would toggle between fast (transistor on) and slow (transistor off).

Another way (still with a 40106) could be to set up two independent HF clocks, with gating diodes on each one controlled by an LFO.

I can imagine ways of doing what you describe with a 4017, but not as simply as with a 40106.

[Processaurus]

Oh, I get it now, you're working with a long delay rather than a short delay, I can see how you'd get square pitch shift with a square wave LFO changing delay time, as the audio within the delay line got recorded at one speed but played back at a different speed.  1/2 the square wave LFO cycle always has to be shorter than the fastest delay in the delay time range, otherwise if the LFO is too slow the audio can momentarily come out recorded at the same speed it is played back at.  It would step sharp, then in tune, then flat, but the in-tune segment would last for an unrelated amount of time to the lfo, and last different times on the hi and low periods of the LFO.  The LFO also couldn't be so short compared to the delay that the lfo segments would overlap and you were hearing audio recorded during the last hi cycle during the current hi cycle or vice versa.  Might be able to boil it down to a ratio between the two.

Not that that wouldn't sound cool, but unsynced LFO and delay would always take hand adjusting the delay time to get a specifically square vibrato.

To get LFO cycles longer than the delay line one'd have to have a stairstepped waveform controlling the delay time, with the period of the stairs synced to the delay time.

With short delays, a linear ramp in delay time equals pitch shift, though when it changes direction there is a portamento (a segment of audio recorded at the same speed as it gets played back at).  The shorter the delay, the less the portamento happens.  Some choruses with the dry signal cut to hear the pure pitch vibrato sound unnaturally square because of that, the boss vibrato goes to great lengths to make a sinusoidal LFO to make it more natural and bendy.

I like your sound sample, it reminds me of NES music.  I'm lazy and might try to reproduce the effect with a square LFO hooked up to the expression pedal jack on a boss PS-5...