U-boat. NEW Smooth sub-octave / synth effect!

[merlinb]

To my knowledge, no one has ever designed a DIY smooth sub-octave effect. The best we have is the chopped OC2, or other commercial schematics. I though I would set this to rights by designing a new smooth sub octave effect, presented herewith!

The U-boat is slightly simpler than the chopped OC2 as it uses only eight opamps, two of which are just buffers. It uses a zero-crossing method to convert the incoming sinewave to a suboctave. This gives slightly better tracking than the OC2 because it is much easier to make a zero crossing detector than a peak detector, but the result contains more harmonics. The OC2 uses two out-of-phase peak detectors to drive an SR flip flop, whereas the Uboat uses one balanced zero crossing detector (highly accurate) and one peak detector. By switching the feeds to the flip flop you can either get the smooth zero-crossing suboctave effect, or a synthy, fuzzy, more glitchy peak detector effect. Like all my effects I have used electronic switching, though it is not essential of course.

The only control is a blend knob, which allows you to mix the sub-octave effect with the original signal, in any proportion. There is also an internal trimmer for setting the max volume of the sub-octave.
There is a switch for selecting either the smooth sound or the synth sound.

Here's a bunch of quick clips I made (played directly into sound card) with the blend knob at different settings. My favourite is the 12-oclock setting, as this gives a convincing funky bass guitar sound... There is a little bit of fuzz audible, but this can be removed by increasing the caps in the second low-pass filter. The values I used were a compromise between getting a smooth sound on one setting, and a really thick synth sound on the other. I may increase them though, we'll see. Unlike most sub-octave demos you hear, I do actually let the notes ring out to show the tracking, which is near perfect over the whole fretboard for at least five seconds.
w w w.freewebs.com/valvewizard2/Uboat.mp3   
(remove the spaces from the www part. I have to do this or freewebs redirects you back here)

Here is a quick schem to give you an overview. There are bigger schems, PCB layouts and photos here:
http://s81.photobucket.com/albums/j207/merlinblencowe/U-Boat%20Sub%20Octave/

[earthtonesaudio]

Sounds great!

[jrod]

Sounds pretty cool!

I love enclosure. The rust is a nice touch!

[theundeadelvis]

Awesome, awesome, awesome!

[p_wats]

Nice!

[Derringer]

quite inspiring merlin!

awesome pictures and clip!

[R.G.]

Neat work.

I pre-apologize for the kibitzing that follows. I hope you can use some of these.

I notice you gate either a true or an inverted signal to the output to create the output. Did you try a switch-hitter for this? I think it would save an opamp package and do much the same function.

On another note, you might try making the second of the inversions in each pair be about half the size of the first. It would add a more string-like second harmonic of the bass tone as it would make it more sawtooth-like as a result; probably help hide some of the harmonics that are still in there from the alternate full wave rectification. I think you could make this switchable with a couple of resistors.

I did a side step on that approach a while back. Did the usual lowpass front end, squared up the signal (which amounts to zero crossing detection, there's an edge at each zero crossing) then frequency multiplied for eight octaves up and summed to a psuedo-sawtooth, organ style. Worked fine, but had to be paired with another side chain to restore the envelope, as the result was always fixed amplitude.

And on yet another note, I think you might be able to phase lock a 100X signal to the switching on the flipflops that do the crossover and inversion detection and use that to run a switched capacitor low pass for a really sharp harmonic cutoff and get rid of almost all of the higher harmonics. Probably at the point of diminishing returns by there, but it's possible.

Another dodge to get better tracking is to heavily compress the input signal and use this compressed signal in a side chain only for the frequency determination, not for output. It significantly extends the range of tracking because it brings the small signals up to preserve the level the detector has to work with. As a side effect, a noise gate on the compressor can be used side-chain style, to mute the output signal when the compressor signal gets low, and drop out low level mistracking in the main signal path because the compressor will still have signal when the main path is way down in level.

[soggybag]

Great work, sounds as good as any octave down I have heard. I like the box. The rust is a nice touch.

[merlinb]

I notice you gate either a true or an inverted signal to the output to create the output. Did you try a switch-hitter for this? I think it would save an opamp package and do much the same function.

A switch hitter (silly name) needs one opamp and one FET, the same as what I have already. However, you'll notice that the in-phase and inverted signals are both sent to the zero crossing detector; this balanced operation gives twice the detection sensitivity with no sacrifice in headroom, which is why I used it.

On another note, you might try making the second of the inversions in each pair be about half the size of the first. It would add a more string-like second harmonic

This is identical to turning the blend knob to a medium setting- yeah it does sound like a plausible bass guitar at some settings!

And on yet another note, I think you might be able to phase lock a 100X signal to the switching on the flipflops that do the crossover and inversion detection and use that to run a switched capacitor low pass for a really sharp harmonic cutoff and get rid of almost all of the higher harmonics.
Another dodge to get better tracking is to heavily compress the input signal

My only grip with these things is that they add a lot of extra circuitry. Part of my aim when designing this was to produce a circuit that performed as well as the chopped OC2 (or better), but was simpler and would fit into a 1550B. Ironically perhaps, getting good performance by adding complexity is easy, whereas acheiving the same by simple means is the highest form of analog design (and the bit I like best!). As it is the tracking is good enough that I don't think compression would make a huge improvement, but it's something to think about. A tracking filter would certainly be a nice luxury, if only I had the space...

The rust is a nice touch!

Thanks, the distressed look was a bodge to hide the crummy etching!

[R.G.]

A switch hitter (silly name) needs one opamp and one FET, the same as what I have already. However, you'll notice that the in-phase and inverted signals are both sent to the zero crossing detector; this balanced operation gives twice the detection sensitivity with no sacrifice in headroom, which is why I used it.

More sensitivity to zero crossing is good, all right. I'll take a closer look at the z-c detector.

What happens with signals with a strong second and higher harmonics (alternatively, did you filter it down to help get rid of multiple z/c in the signal)?

My only grip with these things is that they add a lot of extra circuitry. Part of my aim when designing this was to produce a circuit that performed as well as the chopped OC2 (or better), but was simpler and would fit into a 1550B. Ironically perhaps, getting good performance by adding complexity is easy, whereas acheiving the same by simple means is the highest form of analog design (and the bit I like best!).

Yes - there is nothing so well suited to what it does that it can't have another bell or whistle added to it; the engineer's mantra.
Most people don't appreciate how very many double edged swords there are in analog design. Use it to fullest advantage and you have to use all the primary and secondary characteristics, perhaps more. In programming, they have the same concept: real programmers program right down to the bare metal.

As it is the tracking is good enough that I don't think compression would make a huge improvement, but it's something to think about. A tracking filter would certainly be a nice luxury, if only I had the space...

If the tracking meets your needs, then good enough is indeed good enough. I've always had some major discontents with tracking on bass synths, so a lot of my thinking on them goes to pushing the tracking envelope - which I guess is obvious from my comments.    I've used compression as a sidechain for detection, and it does help if done well. Filtering always helps, as you'll always run into the odd guitar which gives you more than one zero crossing per half cycle. This can be cured easily enough with filtering, but then you often get this combined with a guitarist who can't see why they should have to roll back the treble knob or adjust their playing style any.

[Mark Hammer]

Wit, you include "We Ain't Got Nothing Yet" by the Blues Magoos in the soundclip and nobody says anything?  Oh the shame of it.

[merlinb]

Wit, you include "We Ain't Got Nothing Yet" by the Blues Magoos in the soundclip and nobody says anything?  Oh the shame of it.

So that's what it is! I had no idea- just one of those catchy riffs that's found its way into my subconscious !

[frequencycentral]

So called 'crummy' etch looks great. Soundclip sounds great. But you've gotta sign up for dropbox to host your audio (photos are easier to manage than phootbucket too). http://www.dropbox.com/

[deadastronaut]

i love the look of that....the telegraph design etc....and the little bombs.....great!..

i can't hear it  at the moment, ive pulled my stereo/desk apart..redesigning my workspace....


+1 for dropbox.

[Skruffyhound]

Looks great, sounds great. Thanks very much. I'm enjoying your projects.

[merlinb]

i love the look of that....the telegraph design etc....and the little bombs.....great!..

The dial is based on the dials used in real Uboats. I based it on this copy of the one from Das Boot!
http://forums.watchuseek.com/attachments/f71/383364-does-anyone-make-affordable-yet-qaulity-u-boat-russian-diver-style-watch-armbanduhr-maschinentelegraph-u-boot-u96-u-96-uhr-watch.jpg

(I wonder what Grosse Fahrt means? I know what I *think* it means... )

[deadastronaut]

^ HA HA...grosse= big as far as i know....

i have a ships telegraph in my bathroom.(don't ask ) ..looks very  similar..but in english...


edit: got my stereo plugged in....sounds fantastic...great tracking...whats it likewith distortion, for heavy grungey stuff?...

[thedefog]

Great work! This seriously sounds awesome. Looks like I have a new project in the queue.

[jrod]

(I wonder what Grosse Fahrt means? I know what I *think* it means... )

HAHA! I wondered the same thing so I google translated it. Grosse Fahrt = Great Ride.

[auden100]

Sounds great, Merlin. Playing around with a perf layout. Just wanted to check if a j201 could replace the j112?

Thanks, in advance.