frequency shifting

[samhay]

I have heard rumours (but seen no schematics) of oldish frequency shifting ideas using BBDs to shift the timebase and cause pitch shifting. This sounds like nice idea that is likely to lead to a big can of worms.
If I am not misrembering, you can get momentary frequency shifting on a PT2399 delay if you twist the delay knob while a note is playing. I guess you could building a vibrato by constantly varying the delay resistance with an LFO. I can't remember if anyone has done that yet - seems like a reasonably simple idea.

[slacker]

There's quite a few designs that modulate a PT2399 in some way, if you want to see some examples  search for echo base, little angel or clarinot.
Someone on here, earthsoundaudio I think, suggested if you modulated a PT2399 at the right speeds with a saw tooth you could get something approaching pitch shifting.

[WaveshapeIllusions]

Cool. I think it was Taylor who mentioned it. He said something about constantly doubling the delay time to get an octave down with the caveat that it would get too long real fast. I think he also suggested settling on a min/max delay time and using a square wave LFO to switch between them.

On the sigma delta modulation, there's no way to use separate clocks for conversion, shifting, and reconstruction? From what I've read from the wiki link so far, it doesn't seem like something that is anywhere near simple to do. Looking at it like PWM (which is something I have a better grasp on) it doesn't seem like something a DIYer could do without doing some coding. Shifting the pulse frequency isn't too simple.

Wait... what if you encode the signal with sigma delta or PWM and then run that into a multiplier or divider? Since you're dealing with a more consistent wave (no amplitude vaiation, just pulse width), would the standard pitch shifting methods react better? Say drive a flip flop with it, or use a PLL.

Samhay, I think the Boss VB2 does that, but with the MN3XXX chips instead of a PT2399. A few members have designed choruses with the PT2399, lifting the dry signal would yield a vibrato. I think the Pitch Pirate and Clari(not) do something like that too.

[samhay]

Unless I'm mistaken, to do frequency doubling (e.g. octance down) with a delay line, you would need to be able to set the min/max delay times to be frequency dependent. This is non-trivial, but not impossible.

A quick trawl shows that the Echo Base, Clarinot and Pitch Pirate all modulate the delay resitance of a PT2399. The Little Angle doesn't directly (i.e. pin 6 is not modulated).

[~arph]

There's quite a few designs that modulate a PT2399 in some way, if you want to see some examples  search for echo base, little angel or clarinot.
Someone on here, earthsoundaudio I think, suggested if you modulated a PT2399 at the right speeds with a saw tooth you could get something approaching pitch shifting.

Don't think that'll work..

Say you increase the clock with a sawtooth.
What happens when you switch form max clock speed to min clock speed?  there are stil samples in the register that are recorded near max clock speed, yet we are playing them back at min %^&* speed. You get a massive downtuned signal albeit rising in pitch.. Only until all samples recorded before the clockspeed dropped are out of the register the pitch will be back at normal and rising. Lasers man! lasers! 

The only way I see this happening is when you pan back and forth between two delay lines both modulated with an out of phase sawtooth. So you will have a perceived constant raising delay time clock resulting in a continuous raised pitch ( as long as the sawtooth it's perfectly linear and the response is perfectly linear  )

[~arph]

Wait... what if you encode the signal with sigma delta or PWM and then run that into a multiplier or divider? Since you're dealing with a more consistent wave (no amplitude vaiation, just pulse width), would the standard pitch shifting methods react better? Say drive a flip flop with it, or use a PLL.

Also, I would be very interested to see a frequency multiplier (that multiplies by a factor > 1)

[slacker]

Cool. I think it was Taylor who mentioned it.

You could be right sounds like his sort of thing.

Wait... what if you encode the signal with sigma delta or PWM and then run that into a multiplier or divider? Since you're dealing with a more consistent wave (no amplitude vaiation, just pulse width), would the standard pitch shifting methods react better? Say drive a flip flop with it, or use a PLL.

I don't think PWM will work, most if not all dividers will destroy the pulse width information and just give you a 50% pulse width, so you will just end up with square waves at some division of the PWM frequency, I'm having trouble figuring it out in my head but I don't think the frequency of the signal will be changed.

If you want an old school way of doing it look at the E&MM Harmony Generator, I think the schematic is on there somewhere. This uses a PLL and a bunch of CMOS dividers to give a variety of pitch shifts and uses a clever trick to superimpose the original envelope on the output giving a relatively clean effect.

[tca]

If you want an old school way of doing it look at the E&MM Harmony Generator, I think the schematic is on there somewhere.

Yes it is, just been reading the PDF: http://hammer.ampage.org/files/EMMHarmoGen.PDF

Cheers.

[WaveshapeIllusions]

I

Unless I'm mistaken, to do frequency doubling (e.g. octance down) with a delay line, you would need to be able to set the min/max delay times to be frequency dependent. This is non-trivial, but not impossible.

I think just doubling the delay time would give the same shift in frequency. You'd be halving the clock speed which would double the amount of time something gets reconstructed. I think that would double the wavelength, regardless of frequency.

I don't think PWM will work, most if not all dividers will destroy the pulse width information and just give you a 50% pulse width, so you will just end up with square waves at some division of the PWM frequency, I'm having trouble figuring it out in my head but I don't think the frequency of the signal will be changed.

I think you're right. PWM doesn't modify frequency so the positive going pulse would start at the same time. Perhaps FM would be a more effective method? Frequency would be variable in this case so I think a flip flop would still preserve the initial wave.

Whether or not it would modify the frequency of the encoded wave is something I'm not sure of. 

Yes it is, just been reading the PDF: http://hammer.ampage.org/files/EMMHarmoGen.PDF

Cheers.

Thanks for the link. That's a cool circuit. I wasn't aware of how PLLs were actually used to shift frequency, just that they were used. It makes a lot of sense, like opamp feedback loops.

[earthtonesaudio]

I don't think PWM will work, most if not all dividers will destroy the pulse width information and just give you a 50% pulse width, so you will just end up with square waves at some division of the PWM frequency, I'm having trouble figuring it out in my head but I don't think the frequency of the signal will be changed.

I think you're right. PWM doesn't modify frequency so the positive going pulse would start at the same time. Perhaps FM would be a more effective method? Frequency would be variable in this case so I think a flip flop would still preserve the initial wave.

You can delay PWM using dividers or RC time constants, as long as you treat the rising and falling edges separately:
http://www.edn.com/design/test-and-measurement/4337382/Delay-line-has-wide-duty-cycle-range

[jaapie]

i've been fooling around with this rudimentary dome filter for a few months and thought i'd share it. It's not perfect, but it seems to work pretty well for guitar frequencies and only needs one quad opamp.



i've tested it, and it works on the breadboard. hope it's useful for someone!

[Strategy]

I've heard the Juergen Haible (RIP) Frequency Shifter and it's amazing. It sounds similar to a ring modulator but at low frequency settings sounds like a bizarre phase shifter. All of JH's effects circuits were configurable as either standalone effects or synth modules. I did the Tau Phaser as a really big pedal. If you are not against big enclosures, you could do the project that way.

The JH frequency shifter project requires some component matching effort as well if memory serves. not a project for the faint of heart, but what a high end thing you get at the end.

Strategy