The U-235 Suboctave Generator - a new octave down pedal

[Freppo]

I'm releasing a new DIY project today - The U-235 Suboctave Generator
It's an analog semi-clean octave down pedal with a twist...

Original design based around a Phase Locked Loop.

I hope you like it!

It's avaliable as a PCB, but I'll share a vero layout aswell soon.

More info over at www.parasitstudio.se

Cheers
/ Fredrik, Parasit Studio

[Fancy Lime]

Hi Fredrik,

nice! I've been waiting for this one. Sounds fantastic. I must admit I'm somewhat mystified by PLLs. Will you post the schematic too? I have a feeling that theres a lot of fun to be had with PLLs if one understands them.

Keep up the good work,
Andy

[R.G.]

There is indeed much fun to be had with PLLs. You might try a search of this forum for "phase locked loop" to see a few examples.

Phase locked loops are an analog means of locking the output of a voltage controlled oscillator (VCO) to an incoming signal. The heart of the PLL is the phase detector; this is the part that compares the incoming signal's phase to the phase of the VCO to produce a filtered-DC error signal. The error signal is fed to an amplifier and drives the voltage control input of the VCO in the direction (up/down) that makes the VCO frequency closer to the incoming signal.

All that sounds complicated, but it's the frequency domain analog of what an opamp does. In an opamp, the input differential amplifier picks out the difference in instantaneous voltage between an input signal and an output signal from the amplifier. This difference is an error voltage, which is then amplified by the rest of the opamp and converted to a voltage output which is then fed back to the input in a direction that minimizes the difference in instantaneous voltage between the incoming signal and the output voltage. Another name for an opamp could plausibly be a "voltage locked loop".

There are a few critical things for successful PLL use.
> prefiltering and cleaning up the incoming signal so there's not a lot of hash for the loop to try to follow; the phase detector can be confused by some things in the signal like strong harmonics and extra zero crossings
> the phase detector, which is crucial, needs to present a very clean difference of the actual incoming frequency and the VCO; as with so many things, you can pick any two of fast, good, and cheap
> the phase detector output filter; this determines how smooth the error signal from the phase detector is, and how much wobbling is passed on to the amplifier and VCO, as well as the loop damping, for things like how the output frequency wobbles when it gets a transient change in input frequency
> the amplifier's ability to cleanly amplify the error signal for the VCO
> the VCO's frequqncy range; it's not going to lock well if it can't produce the frequency it's required to produce.

As a practical matter, the CD4046 PLL produced useful answers to most of these issues long ago. It and its successors have a wide range VCO, and the choice of two or three detectors, one of which can guarantee lock over the 1000:1 frequency range of audio. The 4046 and successors are the right starting point for effects PLLs. I'd bet a modest sum that there's one in this pedal.

[anotherjim]

I built this mad thing a while ago...
http://www.diystompboxes.com/smfforum/index.php?topic=113380.0
A desktop box rather than a stomp box. However, tracking is not it's strong point - that was low on the list of what I wanted.

Going against the true PLL design goals that RG explains above, if VCO range is limited (or forced) outside of the incoming signal fundamental and the phase detector can see the harmonics, the VCO output can be strange.

[ElectricDruid]

Good work! It's nice to see something a bit different!

Tom

[Fancy Lime]

Agreed. DIY is not a National Park, so it's a good thing some people leave the beaten path. BTW: I vaguely remember reading somewhere that the DOD meatbox uses a PPL but I cannot find the reference anymore and cannot find a schematic. Is that my imagination or is that so?

Andy

[Freppo]

Great explanation on PLL by R.G.
Yup, this pedal uses a CD4046. 

Good work! It's nice to see something a bit different!

Tom

Thanks Tom

Hi Fredrik,

nice! I've been waiting for this one. Sounds fantastic. I must admit I'm somewhat mystified by PLLs. Will you post the schematic too? I have a feeling that theres a lot of fun to be had with PLLs if one understands them.

Keep up the good work,
Andy

Thanks. I'm glad you like it

I'm sorry to say that I will not be sharing the schematic for this pedal (not for a while anyway), but I plan on doing another CMOS workshop about Phase Locked Loop chips, and how to set them up as a frequency divider or frequency multiplier. So that people can breadboard something similar and play around with it.

Agreed. DIY is not a National Park, so it's a good thing some people leave the beaten path. BTW: I vaguely remember reading somewhere that the DOD meatbox uses a PPL but I cannot find the reference anymore and cannot find a schematic. Is that my imagination or is that so?

Andy

I believe that the meatbox uses a OTA, but i'm not sure. It was long ago that I look at the schematic.

[EBK]

but I plan on doing another CMOS workshop about Phase Locked Loop chips, and how to set them up as a frequency divider or frequency multiplier. So that people can breadboard something similar and play around with it.

Yes, please! 

[Philthy]

Awesome

any update on the New Wave relase date?

[Freppo]

I will be released in about 2-3 weeks if everything goes well.

[Fancy Lime]

I'm sorry to say that I will not be sharing the schematic for this pedal (not for a while anyway), but I plan on doing another CMOS workshop about Phase Locked Loop chips, and how to set them up as a frequency divider or frequency multiplier. So that people can breadboard something similar and play around with it.

Jättefin! I prefer a workshop explaining the principle over a schematic any day! Also: kudos for your CMOS workshop series in general. Learned a lot from that. Easy to follow, yet complete enough to even understand details.

On a side note: Bummer this forum has a bazillion smileys but no "thumbs up". Maybe l₃ or _ξl ? Anyway: *thumbs up* is what I'm saying.

Andy

[kinski]

Very cool, Freppo! I noticed in the build doc, that there is a detune trimpot. Does this set the pitch of the detuned output? If so, that would be great to mount externally.

[Freppo]

Very cool, Freppo! I noticed in the build doc, that there is a detune trimpot. Does this set the pitch of the detuned output? If so, that would be great to mount externally.

The trimmer doesn't change pitch gradually (that would have been sweet!).
It's a sudden change from a normal one octave down when turning the trimmer, so it's a set and forget kinda thing.
Had some trouble getting consistent results with a fixed resistor from chip to chip..

Cheers
/ Fredrik

[R.G.]

I prefer a workshop explaining the principle over a schematic any day! Also: kudos for your CMOS workshop series in general. Learned a lot from that. Easy to follow, yet complete enough to even understand details.

Then you would really enjoy "The CMOS Cookbook" and "The Active Filter Cookbook".

It's a sudden change from a normal one octave down when turning the trimmer, so it's a set and forget kinda thing.
Had some trouble getting consistent results with a fixed resistor from chip to chip..

I hear an echo in here...  That's always going to be a problem.

Did you also sense the lock state on the PLL and mute it when not locked? Locking time, lock range, and phase jitter in lock are the normal issues to contend with once the PLL is running right. The frequency slew speed on a VCO limits how fast it can get to a new/suddenly changed input frequency, so the PLL output is putting out unrelated and slewing signal at full volume until it hits lock. So that stuff needs to be muted somehow. It will have a huge effect on how "polite" the pedal is in use.

There is a trade off in how fast the VCO slews and how well it retains lock. If the VCO slews very quickly, it follows little burble in the phase detector output really well, too. Phase detectors universally put out pulsed up/down pulses that the phase filter filters to a rippled DC for the VCO to follow. If you make the VCO slew FAST for quick lock, it also follows the ripple in the phase detector filter output well too. If you make the filter filter out the ripple too well, it also slows down slewing for initial lock. The solution that a lot of critical applications use (or used... this is 11970s technology) is to sense lock and change the phase filter's time constant or the amplifier gain, or both to slow down the VCO slew rate in lock.

So sensing and using lock is a biggish deal. The 74AC7046 (I think...) versions of the CD4046 make this easier with both a third phase detector and an improved lock sense.

I haven't looked at your pedal yet, so this may seem like a silly question, but did you phase lock and multiply then divide down to give the choice of octaves up and down simultaneously? You can also multiply up, then do resistor summing of the multiplied signal or a johnson counter to waveshape a sine or triangle of the base frequency instead of just a filtered square wave. I posted some info and a schematic link to this forum about that some time back. A single CD4024 can give you many octave-related outputs. You put the VCO output into the clock of the CD4024, and the loop locks whichever output you feed back to the PD input. The higher outputs are octaves up, the lower ones below the fed-back 4024 output are octaves down. You can also make this thing lock to just-intonation notes by making the loop divider be a non-binary output. There was a 1970s? 1980s? article in the UK Electronics and Music Maker magazine that was listed as a "Harmony Generator", doing either octaves or locked thirds, fifths, etc. I think Mark has this on his web site.

I also suspect that using a PIC or other uC with a built-in numerically controlled oscillator in a PLL to replace the VCO could be done. You would make the NCO track an up/down count of the frequency difference as detected by the PD. It's tempting to have the digital up/down output of the 4046 style PDs run right into a PIC input, but I have no idea how to make that be stable or what you'd use for overall loop gain or how to control it so it didn't jitter uncontrollably. Needs some thinking.

[Freppo]

Then you would really enjoy "The CMOS Cookbook" and "The Active Filter Cookbook".

I have them already. Got alot of ideas from the CMOS Cookbook.

Did you also sense the lock state on the PLL and mute it when not locked? Locking time, lock range, and phase jitter in lock are the normal issues to contend with once the PLL is running right. The frequency slew speed on a VCO limits how fast it can get to a new/suddenly changed input frequency, so the PLL output is putting out unrelated and slewing signal at full volume until it hits lock. So that stuff needs to be muted somehow. It will have a huge effect on how "polite" the pedal is in use.

I have tried to combat all these issues in my design, and with pretty good results. I have a comparator set up before the 4046 input to mute the signal. It has some hysteresis and is configured to compare the input signal to a filtered signal (it's basically an RC filter between the inverting and non-inverting inputs) and that works really well compared to a traditional comparator with just a fixed voltage reference. The sustain is longer without as much sputtering decay as usual and it also works better with lower amplitude signals and doesn't mistrigger from noise as much thanks to the filter. The RC filter has a cutoff trimpot that acts as a voltage reference/touch sensitivity trimmer so the 4046 output can go from pretty gated and controlled (but still with a decent sustain), to very touch sensitive and crazy. But my goal with this pedal was to keep it "polite" so with the trimmer in a less sensitive setting you can't tell it's a PLL based pedal when listening to it (at least until the CV input gets used..). I've also cleaned up the PLL output with a 12db per octave second order lowpass filter.

I haven't looked at your pedal yet, so this may seem like a silly question, but did you phase lock and multiply then divide down to give the choice of octaves up and down simultaneously? You can also multiply up, then do resistor summing of the multiplied signal or a johnson counter to waveshape a sine or triangle of the base frequency instead of just a filtered square wave. I posted some info and a schematic link to this forum about that some time back. A single CD4024 can give you many octave-related outputs. You put the VCO output into the clock of the CD4024, and the loop locks whichever output you feed back to the PD input. The higher outputs are octaves up, the lower ones below the fed-back 4024 output are octaves down. You can also make this thing lock to just-intonation notes by making the loop divider be a non-binary output. There was a 1970s? 1980s? article in the UK Electronics and Music Maker magazine that was listed as a "Harmony Generator", doing either octaves or locked thirds, fifths, etc. I think Mark has this on his web site.

No. This pedal is octave down only, from the 4046 alone. No other counter or flip flop in the circuit, only a few op amp stages. But I have used the multiplication/octave up method that you are describing in other pedal projects, such as the 4015 Guitar Synth and the Into the Unknown. They use a CD4015 shift register (simply configured as 4 D-flip flops in series with the help on one inverter stage) to do both octave up (together with the 4046) and octave down. I love how versatile the 4046 chip is.

I also suspect that using a PIC or other uC with a built-in numerically controlled oscillator in a PLL to replace the VCO could be done. You would make the NCO track an up/down count of the frequency difference as detected by the PD. It's tempting to have the digital up/down output of the 4046 style PDs run right into a PIC input, but I have no idea how to make that be stable or what you'd use for overall loop gain or how to control it so it didn't jitter uncontrollably. Needs some thinking.

Sound like a cool idea

[PRR]

> this is 11970s technology

Wow! Almost 10,000 years into the future!! 

NO THUMB'S UP?? Try copying the url under the above (from WikiMedia).

> Got alot of ideas from the CMOS Cookbook.

Just for context: the story starts in RTL Cookbook (available as free PDF) and more in TTL Cookbook (??). We ALL had BOTH when CMOS Cookery came out, so Don didn't go over all the basics of logic again.

There's also a few things stone-age TTL still do better than garden CMOS. (And I like the simplicity of RTL, but you'd have to roll your own now.)

[anotherjim]

I'd guess that if you put an XOR frequency doubler between the VCO and PC, then the VCO should track at 1/2f. It would be a very clever to use the XOR PC in the 4046 for that, but I can't figure out how as the inputs would be already occupied. No other logic device in Frederik's design, although XOR can be done with op-amps & diodes.
https://www.cgs.synth.net/modules/cgs54v1_xor.html

[Freppo]

DIY kits are now avaliable at www.musikding.de
 

I'd guess that if you put an XOR frequency doubler between the VCO and PC, then the VCO should track at 1/2f. It would be a very clever to use the XOR PC in the 4046 for that, but I can't figure out how as the inputs would be already occupied. No other logic device in Frederik's design, although XOR can be done with op-amps & diodes.
https://www.cgs.synth.net/modules/cgs54v1_xor.html

Good guess Jim I'm using the xor, but no frequency doubling going on.
I don't want to give away too much, so I'll PM you the answer.

[Freppo]

I've added a verified vero layout for this project.

You can find it here - http://www.parasitstudio.se/uploads/2/4/4/9/2449159/u235vero.png

/ Fredrik

[Renegadrian]

Hope Freppo reads that or others may help...

Let's say I want to build this using the vero layout...how to get rid of the cv in?! which are the components that can be omitted?! Also, could a regular dpdt on on work, giving only 1 octave/2 octave?!