"1 Octave Down" pedal project

[ElectricDruid]

The "buy or build" argument needs to take into account relative costs versus benefits too.

There are many, many **WAY** overpriced boutique fuzzes and overdrives out there that you can *easily* build yourself and tune to your personal idea of fuzz/drive heaven for less than they sell them for, and with twice the satisfaction.
At the same time, there are some *amazing* and very low cost digital pedals out there that do a brilliant job of some of the things that analog electronics will never be any good at (like pitch shifting and some delay effects).

So if you just want to get the job done well, get a EH pedal. If you want to build an old-school flip-flop octave down divider, go for it, but "caveat constructor", to hack the latin.

T.

[swisher]

Of course I could buy digital octaver and there will be no tracking problems, but:
1. I want to build it myself
2. I don't want any knobs (just one setup)
3. All closed in enclosure with special painting
4. That satisfaction...

I've done 80% of U-boat (I don't have all parts). This project looks promising so I will try it

Thanks for all answers!

[Mark Hammer]

I'm not so sure such a unit can be produced in a set-and-forget manner.  Not unless you know you will always use it with guitar X, and that it will never have anything else between the guitar and pedal to change the signal in any way.  These things CAN be twitchy, and require adjustments to eliminate, or at least reduce, the twitchiness.  If you are blessed with the circumstances that allow you to anticipate only one set of conditions, more power to ya.  But I'm not entirely convinced such a circumstance actually exists.

Still, kudos for pushing on ahead instead of buying.

[swisher]

I need this effect for one tone only. Octaver as first effect in chain, then some gain from preamp - and that's all. Maybe some delay, but probably not. I can use it with one guitar if it will be needed.
I didn't tell this but I'm satisfied with OC-2 sounding (and his bad tracking). But if it's possible to build something better, even a little, I want to build it

[Fancy Lime]

Hi swisher,

the thing with "OC-2 type" bass octavers is that they need to switch between an inverted and a non inverted signal "exactly" on the positive or negative peak. For that they need a peak-detector circuit, which is not a trivial thing to design if one is dealing with input waveforms that are not well defined. In practice, the OC-2 uses an envelope follower and compares the input of that envelope follower to its output using a comparator. (In fact it does that twice, once for the negative, once for the positive part of the signal because the flip-flop needs a set and reset signal.) So what that circuit really detects is when the original waveform drops below its own envelope. If that works with theoretical infinite precision and the input is a wave with exactly one peak point (as for example a sine wave), then the circuit detects exactly the peak. Problem #1 is: There is a difference between praxis and theory and it is bigger in praxis than in theory. In a real device, there needs to be a finite difference between the two signals that are being compared and therefore the comparator will switch a little after the actual peak. Problem #2 is: The less pronounced the peak of the original waveform is, the longer the delay between actual peak and peak detection gets. In the extreme case of a square wave input (which does not have a defined peak), "peak detection" happens on the zero transition. Guitar or bass signals can be seen as something between sine and square wave. The OC-2 deals with this problem by putting a 3rd order Sallen-Key low-pass filter in front of the peak detektor. the problem there is that this method is only effective over some frequency range. The deeper the input signal, the less effective is this method. That is why OC-2 type bass octavers do not like distorted (i.e. more "square-wavy") input, especially on bass.

So, in conclusion: Peak detection is difficult and there is no method to make it work perfectly with imperfect inputs, like a guitar.

The alternative would be something that relies on zero-transition detection instead, which is much much easier to do reliably for a guitar signal. Valve Wizards U-Boat and Slackers Slacktave use this method. With that you can basically just boost the $#!+ out of your input signal before you feed it to a comparator or better yet a Schmitt trigger, and get really good tracking at any volume. With this method, a compressor up front helps, with the OC-2 method it does not. Of course you then get a different sounding output than with an OC-2. I am working on a method to use zero-transition detection to detect peaks, but that is a bit more involved and I'm not sure yet if I will get it to work at all. So, I think it is possible to make a better tracking bass octave. But it will not sound like an OC-2 or it will be much more complicated. Things get much easier if you are OK with something synth-sounding like the Slacktave that uses the output of a ripple counter directly. Add some tone-filtering and that can sound quite nice. I think the old MXR M88 Bass Octave did just that (except with a flip-flop instead of a ripple counter) but I am not sure. This method of course kills all playing dynamics on the octave signal but thats not necessarily a bad thing and happens anyway when you stick to the usual playing caveats for analog octavers.

Good luck with your project and keep us posted,
Andy

[Kennt82]

Maybe some ideas from here? "M51134 OCTAVER"  It's not the Meatbox... but kinda like it.
http://www.diystompboxes.com/smfforum/index.php?topic=101580.0

[anotherjim]

When the harmonics are stronger than the fundamental, even zero-crossing detection is fooled, since the harmonics can add extra crossings. This more likely early in a guitar notes life. The same strong harmonics create extra peaks. I keep thinking there must be a way to combine xing and peak detection, but every way I think of doing it fails. Our brains know what note is played because of the harmonic relationships, but the electronics are going to be fooled everytime.

As Andy notes, low pass filtering helps but ideally should track the fundamental or else it's at best a 1^1/2 octave range - which brings us back to the problem of pitch detection again.

If you can wait for something to average the signal enough to get the fundamental frequency (like your guitar tuner does), you have a chance, but you will always have some delay before you hear anything, which would be longer the lower the fundamental pitch is.

[Fancy Lime]

@ Jim

Well, yeah, one problem is that at the beginning of a note, the strings movements are quite chaotic before they settle down to oscillate periodically. For a bass played finger style this is much less pronounced than when playing with a pick (guitar or bass). Using a Schmitt trigger instead of a comparator to detect the zero crossing and putting a low pass before that should help. I tried to post that in another thread a minute ago but the image server seems to be down. I'll try again later. That thing also uses zero-detection to find the peak. Not sure if it works though, just an untested brain-child so far.

Also: even if our guitar signal was a perfect sine wave from the nanosecond the pick leaves the string, we would still run into an uncertainty, which is quite similar to some quantum phenomena. A frequency is only defined for a periodical oscillation but during the first oscillation, there is no periodicity yet. Consequently anything that analyses the frequency to produce another signal is always going to be a little late for the party. Damn physics...

So, I guess "perfect" tracking is simply impossible, both practically and theoretically. But I do think there is room for improvement over the designs that have been around since the dawn of time.

Andy

Edit: Now the upload worked:
http://www.diystompboxes.com/smfforum/index.php?topic=118550.0