jh_scanner_vibrato

JH. Interpolating Scanner
and
Scanner Chorus / Vibrato

In the 1990s, I have designed an "Interpolating Scanner", a CV-controlled linear crossfade over a certain number of VCAs, to be the center module of my JH-3 modular synthesizer. You can read all about it here, including a block diagram, schematics using now obsolete SSM chips (not recommended any longer for building), and a glimpse of its widely varying applications such as a voltage controlled crossfader, a wavefolder with dynamic breakpoints, a tracking generator with voltage controlled breakpoints, or voltage controlled non-harmonic modulator.
Here are some sound samples:

Illumination (excerpt)
OB-8 Pad sound into EMS 8-Octave Filter Bank (Clone). Filterbank outputs are crossfaded with Interpolating Scanner. (It also goes into a panner, and - obviously - into a reverb.)

Polysynth "enhanced" with Interpolating Scanner
The Interpolating Scanner is used to create a complex nonlinearity, basically acting as a parametric distortion device.
A sound from the OB-8 is fed into the Scan (!) input of teh Interpolating Scanner, while the 8 signal inpots are fed with DC voltages to set the piecewise-linear distortion courve. The Sample starts with the dry OB-8 sound; then the Interpolating Scanner is switched it.
Very short sample in .WAV format (mp3 is horrible here).

NEW: Waveshaper
First part of the :wav sample is dry signal (OB-8 sound), second part is the same thing run thru the Interpolating Scanner set up as Wave Shaper, with one Breakpoint set to a fixed value, and 5 breakpoints set by 5 slow running triangle LFOs.

Right from the start, the Interpolating Scanner has also been intended for an emulation of the Hammond^(R) Chorus / Vibrato.
(Hammond is a Registered Trademark of Hammond Suzuki. "Hammond Chorus / Vibrato" refers to a mecahnical / electrical device that's found in vintage Hammond organs that predate the Hammond Suzuki company. I'm using the words in order to describe these vintage devices only, and am in no way affiliated with the Hammond Suzuki company, who own the registered trademark of the name "Hammond", and still build organs under that name.)

The Hammond Chorus / Vibrato consists of two main parts: A tapped LC-delay line which provides a delay of up to 1ms for frequencies up to ca. 5kHz (the "Line Box"), and a mechanical / capacitive device (the "Scanner") what continuously crossfades from one tap of the Line Box to th enext, and then back again.
There have been many attempts to emulate the unique sound of the Hammond Chorus / Vibrato, using different modulation waveforms, and a combination of frequency and amplitude modulation.
I'm using a direct approach, a "physical modelling" approach if you will: The huge "Line Box" is emulated by a smaller version, which is still a LC-filter. It's just transformed to a lower line impedance that makes it possible to use unexpensive off-the-shelf 33mH inductors. The capacitors and termination resistors are re-calculated accordingly, as is the compensation for the losses along the line.
The mechanical / capacitive scanning is replaced with a fully electronic version - a 9-stage version of my Interpolating Scanner.
You need 9 stages because the Hammond Scanner had 16 positions, of which the last 7 were hard-wired to 7 of the previous 9, thus transforming a rotary motion into a linear scanning along the line Box taps. So despite its rotary implementation, Laurens Hammond has actually used a linear back-and-forth scanning right from the beginning! Also, the rotary motion does not introduce any sin- or cos-shaped modulation, as one might think at first glance. There is a sort of sine-shaping going on, but it's in the spacing of the taps along th edelay line, not in the scanning circuitry!
So all in all I think my emulation is as close to the original from its topology as it could possibly, theoretically be. There may be a few parameters that could be tweaked for optimal realism - and there may be quite some variation in original Hammond Scanner Vibratos from model to model, and due to ageing of components. But it's important that you get the right topology - that means, all the little side effects from crossfading like frequency-dependent phase cancellation, must all be there "automatically", by design.
I leave it to you if the version I've implemented and made sound samples with is realistic enough or not. What I'm going to provide is a PCB that allows you building your own version by tweaking certain component values, and a recommendation of what components I've used.

Unfortunately, I don't own a Hammod organ - just a Korg CX-3 (old version), so the sound that goes into the Scanner Vibrato isn't that realistiv to start with.
But hear for yourself how the Scanner Vibrato changes this sound:

These samples have been created with an old version of my circuit, that used only two VCAs and a lot of CMOS switches instead of a full Interpolating Scanner.
I'm keeping the documentation for this up on my web site here, but I don't recommend building this.

The new version does exactly the same in terms of periodic modulation (i.e. the Hammond Vibrato application), but is not limited to these applications: You can scan along the (on-board) Line Box with arbitrary control voltages. (Random voltages, Envelope followers, aftertouch sensors - you name it.)

And you can use the same PCB and build a generic 9-Stage Interpolating Scanner for your Modular Synthesizer (+/-15V power supply) by simply omitting the Line Box stuff, omitting all the inductors in particular.

IMO - and I know this is shameless advertising! - everybody needs at least two of these PCBs: One as Chorus Vibrato, and one as a generic Interpolating Scanner. But then again, you may want at least 3 Interpolating Scanners in a modular: One for Oscillator waveform crossfade (a la RSF Kobol), one for Filter response crossfade, and one as a tracking generator.
Oh, I forgot voltage controlled LFO waveform crossfade. :)

At this point, I would like to thank Don Tillman. Based on my Interpolating Scanner idea, he created a very brilliant implementation of his own, which is more elegant than my first solution in several ways. Don kindly granted permission for me to use his implementation in my PCB project, which I will gladly do, as it needs less board space and a lot less current consumption. I've slightly developped Don's circuit further, mainly using emitter followers instead of diodes, and introducing emitter degeneration resistors on the transistors that perform the crossfade in order to linearize the transitions and get more independent of transistor tolerances. I've also used a TL431 voltage reference for precise control over all bias currents, turned the circuit upside down (current sinks instead of current sources), and drive simple transistor pairs instead of OTAs, for the VCA functions.

The driver stage for the LC delay line can be configured either as a simple voltage buffer, or as a variable low pass filter.
The latter may be useful with synthesizer signals which aren't band-limited like a Hammond organ.
The following pictures shows both options:

Marriage of Chorus/Vibrato and general purpose Interpolating Scanner, Tracking Generator, etc.

I'm convinced that everybody should have more than one of these devices, specialized for the different functions, because in a modular system, sooner or later you want to use the different functions at the same time. But upon special request, I'll sketch a way to even combine the Chorus/Vibrato device with a Scanner / Waveshaper module that has individual inputs. But be warned: This is not for the faint of heart!
The idea is that you have a 4-position switch with which to choose Chorus Vibrato 1 ... 3, and in it's 4th position, the general Scanner / Waveshaper function.
For this you need a 9-pole switch with 4 positions. Not just 8-pole, because there are 9 VCAs, and in the ordinary Chorus/Vibrato mode the 1st VCA input is always hard-wired to th estart of the delay line. In the ordinary Interpolating Scanner mode, there is no delay line at all, so the first input channel is just connected to where the first VCA input is (with the whole delay line missing on the PCB).
If we marry the two functions, we have to break that internal connection, and close it via one pole of the 9-pole switch when Chorus/Vibrato mode is selected.
Also, the resistor values to the VCA inputs are different in both versions. We solve this problem by soldering the smaller ones (from the Chorus/Vibrato version) into the PCB, and add eight 30k resistors and one 24k resistor in series with the on-board resistors when the Scanner/Waveshaper mode is selected. These Resistors can simply be soldered between the solder lugs of the big switch and the wiper connections of the input potentiometers.
That was the overview - now lets start in detail.

And here's the location of the Taps 1 ... 14 and Selected Taps I ... VIII again:

I noticed there is a tiny, but not unimportant design flaw - I'm surprised no one else mentioned it so far: The output amplifier gain is too high. That means, the output will clip hard and nasty way before the scanning transistors go into soft limiting action. Ok, there is a lot of ugly/hard clipping gear out there, so probably that's the reason nobody complained, but that's not the way I normally design my audio circuits. My apologies. - Fortunately, this can easily be fixed, by just changing two resistors. What has formerly been 13k, should now be 4k3. I won't redraw the schematics diagram, but here is a correction of the PCB layout (applies to all versions):