The FM Tube Modulator Jimi Loved

[Eb7+9]

transmission line impedance levels ... ?!
50 ohms AC Driving Point Impedance on each feed ...

... well, whattaya know

yielding almost constant 120 degree phase shift between outputs
from about 100Hz to 10kHz

I think I'm sold ...



thanks much Dave ...!

http://www.lynx.net/~jc/50-50.jpg

... moving on then

[daveweyer]

Alright, that's more like it. That is very similar to what I measured 45 years ago using multi-trace scopes.
It looks like you are on track to make an improvement in the original design; I have to wonder if the improved version will be too good, if you know what I mean, as most of us in the audio and tube fields have found over the years, there is often something valuable in the raggedness of these simple designs which gets lost in too much perfection technically--guitar players will certainly attest to that.
Baldwin could have used both sections of the tube in parallel though and made a great improvement--maybe they did, I'd better check the unit to see.
The "split load" phase inverter was often chosen for its near-perfect amplitude matching between phases, especially if no gain was needed in the stage (gain of 1) , as I mentioned they could have used a different tube, a high transconductance regulator type, which would have given them a very low driving impedance on each phase. Alternatively they could have used another cathode follower and gotten very low driving impedance, similar to what you'll get if you use an OTA.
Perhaps they just didn't care, or they found that it didn't really make any difference to the effect they were after. Now days we'd imagine it was just to save a few pennies, but back in 1961 those kind of efficiencies hadn't been considered so all encompassing.
At any rate you have the first important piece of the puzzle, the real secret to how this thing works.
You might also look at Brad Plunkett's patent, he was after the same thing but using different phase alignments. (Brad was the engineer at Vox who designed the Wah Wah, he was also Bob Hovland's boss and my friend and teacher.)

[Eb7+9]

just for fun I did some small-signal analysis of the 2-transistor WAH a few years ago, from that I came up with a parametric equivalent running on op-amps ... allowing AC conditions conditions to be varied (via trimmers) without having to worry about DC shifts (eg., tweaking the vocal properties) and also providing high-headroom operation ... I'll admit I'm still in awe of the basic WAH design // it almost shouldn't be, cuz it's so weird electronically speaking

there is NO doubt in anybody's mind that the tube version of the Chora-Tone will have the most organic, full bodied, MOJO ladden response of any of the architectures available for embodying the system ... no contest there IMO

so, a word about why I'll be using "linear" types of technology here ...

maybe it's because I went to school and have that tendency many schooled clowns have, but I know that once the small-signal response of a circuit is modeled or understood then we have the choice of porting over THAT response to another type of circuitry, either with a different large-signal response (as in going from tube to jFET's, say ...) or going to one which is very neutral dynamically (linear gain blocks ... op-amps, OTA's, etc ...) ... or DSP

my personal tendency is to go with the pure plain-jane "linear math" equivalent circuit first - and that's done using linear gain blocks (no MOJO) ... and then, if I see it fit I'll funk it back up with MOJO by finding some kind of carefully biased class-A (single-ended) equivalent to make it happen with ... I've done this with compressors, and phasors ... it's pat of the process I like to take sometimes

in this case the pull to go in the linear direction is even stronger
to this end I'd like to point out a few key points that Wayne emphasizes in his description ...

"... The phase-split components of the original signal are next amplitude modulated at a low frequency rate" (column 1 bottom line) ... "... continuous phase variation ..."(column 2 lines 10 and 17) ... "... by virtue of the action of the oscillator itself there is applied to the grid of the triode 61 a generally sinusoidal sub-audio signal" (column 9 line 24) ... "... changes the operating point so that the small signal audio frequency voltage gain varies over wide limits" (column 9 line 31) ...

to me, the text seems to describe a set of continuously varying operations ... and so, I'm inclined to mimic what the text says in a more direct way, electronically speaking ... doesn't mean I have less respect for that fact that Wayne made his invention run on less-linear gain blocks // ... I say, let's see how far off the "linear" version will be in terms of "effect quality" ... or maybe we can get closer to ideal performance going this route ...

we won't know unless we build both (or more) styles of design

my goal here is to first build a purer-sounding "equivalent" system to Wayne's ... just for sake of curiosity, stripping the MOJO away and leaving only ... a more purely (gulp...) MATH response !!

(heh heh ... y'all scared now ..?!)

I also need to take this route first to confirm that I'm interpreting the design thinking properly ...
once we've done that, then we can start considering alternative class-A design means

there are other personal goals that I like to meet when doing something like this // one is achieving really good noise performance (while we're at it) ... in this regard, I've got the front end sussed out, and I think I can keep the OPA2134 op-amps running at close to their minimum noise ratings ... I really like the transient response and distortion specs of this IC, and so I think - at least as boring old op-amp designs go - I think I'll be able to give this thing a good level of spectral and transient fidelity ... by going op-amp I can bring down the driving point impedances to lower values possible with triodes, possibly enhancing the function of the all-pass lattice ... these are possibilities here

as of yet, I m even considering something with low-enough DC offsets so that we could run the unit without coupling caps (and phase distortion) at both the input and output ... 'tis an advantage of going "op-amp" after all // and I like to make use of it when I can (something that is not open to commercial manufacturers ...)

---

I'll leave you now with a cool idea that popped into my head before crashing last night ...

as some may know, I've done quite a bit of tinkering with true-stereo Phasing and Vibrato ... either with bulbs and photocells, matched opto-couplers, and also using PWM techniques ... my latest stereo design uses both PWM and opto-couplers ... the idea is basically to have an anti-phase replica of an LFO to play with and drive a second channel in anti-phase

simply... I just realized that my separateLFO-OTA approach makes it possible to build the "mirror-image" replica of the Projector ... I have unused anti-phase currents available that I could use to drive a complimentary set of OTA's that would put out the same signal in reverse rotation ... so, if output side were to put out +, -, +, -, + phase sweeps across 5 octvaes, the other would be -, +, -, +, - ...

something the world probably hasn't heard yet

I have a feeling I'm just going to have to go ahead and try that out ...

---

FYI, to give an idea of the complexity level
each modulator uses 3 OTA's, 3 op-amps, 3 diff-pairs (6x PNP), 3 PNP mirror outputs

and so a 5-octave unit would need 15 OTA's, 15 op-amps, and 45 PNP transistors
the front end uses 3 op-amp IC's (2 duals, and a single)

so, not too bad as a single unit // too much as a dual...

yeah, I guess two boards will be the way to go then
easy to link to appropriate lines, yada yada


ok, enough yakkin' // off to working on the PCB layout ...
~jcm

[daveweyer]

I was disappointed in the anti-phase idea with this particular circuit; I have two identical units and so it was easy to set them up in mirror image, one thing that was most obvious was the garbledness of the overall sound-- it's really so complex of a phase shifting phenomena that the ear cries out for more of the original sound. You'll hear it when I get Xplorer to post some samples, it is VERY wavy and churning; it really doesn't sound all that good by itself except on extremely dense distortions of single notes, like a fuzz output, or a guitar synth. On those tones the band limiting actually makes them sound like muted brass horns or big organ reed pedal pipes, just with a lot of movement.
I've made thousands of sounds with these units and have a pretty good library of successful ways to utilize the things.

Now in a simple stereo recording, with the main signal on the left and the modulator on the right, the mind is inevitably drawn into what seems like a parallel universe, some mystical, haunting, endless spacial secret world, and after a while of listening, the brain stops hearing the churning of the unit as mechanical artifact and surrenders to the endless space it creates--it's such a mind bender.
For one approach to using this unit I put a Leslie speaker on the left on slow rotor, and use the tube modulator on the right; now this approach IS very satisfying, two spacial modulators, but not mirror images.
I'll provide some examples of this approach.
Using them in series is also a trip, especially if you're trying to play along with the pulsing of the units.

I have also found their noise to be very useful, also their hum, if you can believe that. I have made long recordings of the unit itself, with no input signal, just the endless movement of hum. I have pitch shifted these slightly to musical pitch and then applied them as low drones for layers applied over the top.

So I wouldn't worry too much about anti-phase units, at least on this particular circuit. Of course, as always, be my guest if you need to try it. It looks like you're off to some extreme fun.

This thread should be very exciting for many of the forum members as they can see the development take place right before their eyes. This may be very educational for all of us. What a concept, a publicly developed electronic engineering project............

[Eb7+9]

I was disappointed in the anti-phase idea with this particular circuit; I have two identical units and so it was easy to set them up in mirror image, one thing that was most obvious was the garbledness of the overall sound-- it's really so complex of a phase shifting phenomena that the ear cries out for more of the original sound.

quick question about that Dave,

if I understand correctly, all you would have done to the second unit was reverse the modulation order ...?!

unless you had a way of sync'ing the oscillators I imagine that the oscillators were not phase-locked
and so, I can imagine these two units together, with eight out-of-sync LFO's, could make a somewhat confused sound

of course, I'm just guessing here

in my case I would have two available outputs from each of my 4 or 5 LFO's
and these in-phase and out-phase outputs would be locked by nature (as one being a direct inversion of the other)
so, the modulation would also be locked (in anti-phase) for each octave band the same way

(same principle was as in my stereo phasors, and other true-stereo units as well (eg., Mutron Bi-Phase)) ...


In so doing, I imagine likely creating a more "focused" effect ...
a different sounding beast altogether if my presumptions are correct

---

the idea of modulating AC hum sounds intriguing to say the least ...
I need to hear me this !!

---

just a last minute thought here ... what happens when centering the BP sections around the E notes on a guitar ? ... since there are five (octave) E's on a 24 fret axe ... maybe this is a logical choice // curious to see what happens when we set up the Cora-Tone this way

[Xplorer]

This thread should be very exciting for many of the forum members as they can see the development take place right before their eyes. This may be very educational for all of us. What a concept, a publicly developed electronic engineering project............

it is !!! i didn't dare to interrupt this amazing thread, you guys seem to have a lot of fun ! can't wait to hear this.
no problem, iwill post your clips when you're ready.

JC, the stero vibe concept applied to the tube modulator, wow ! yeah ! go for it !
i don't know if anyone else can fully understand this level of electronic like you do, i sure can't, but it's still fun to read

[daveweyer]

JC, can you imagine the fun I've had all these years with these two crazy boxes, and all the various things I've tried between them and within them? I don't know if there is any arrangement I haven't tried, including using the same oscillators for both sides. Now maybe you'll think of something I didn't, and that would be exciting for me and probably everybody else who loves this kind of stuff.

These contraptions came along at a very transitional time in the history of electronics and changing markets. They have all but disappeared from the face of the earth; they were expensive to produce and difficult to make, and their social addition to the wants and needs of liturgical music was limited to the understanding of the church members who existed in the smaller markets and had to endure the limited range of the Baldwin model 5 organs which where plentiful in those arenas. The model 5 was a tube type organ with 12 master oscillators using Eccles Jordan flip-flops to generate the lower octaves, and subtractive synthesis to generate various tones. Needless to say, it had a very artificial quality because all the tones were phase locked, it didn't matter how many tones you added with the organ stops, they all added up to basically one tone with various parts of the spectrum added back in.
Enter the Choratone, a box Baldwin spent a lot of money on in order to spread the tone spectrum on their existing lower end products. As I mentioned before, I discovered this unit in a church in North Dakota when I was a technical representative for the Baldwin Organ Co, traveling Montana and North Dakota, repairing these things and setting up church organ installations. It was being used wrongly in the church because the technicians who did the installation didn't know how to use it (i.e. no stereo system, just one signal source going through the unit and connected to a speaker).
But I was hooked immediately.
Transistors were momentarily to replace all those old tube organs, and with them, the Choratones were thrown out because the cheapness of transistors allowed multiple sets of tone generators--the chorusing was inherent in the system. By the time I got to Hollywood in 1967, the Choratones were accumulating in Baldwin's factory repair center in City Of Industry California, not far from Los Angeles, and where I did a stint for Baldwin before joining West Coast Organ and Amp Service.

But I knew something about the Choratones, and picked some up for myself, specifically for guitar, and soon for demonstrations into Jimi Hendrix' and many other guitar player's energized imaginations. I found it's uses musically endless, especially with being able to tweak it in various ways. And I loved the concept, even though the world of phasors and tremulants of various types would move a completely different direction.
So, after all these years, like my information about Jimi's amps and sounds, I decided to spread the word.
Maybe you'll make it into a valuable tool for guitarists which they can actually afford.
The tube type units will always have a sound of their own, as the grandaddies of whatever you come up with.

To comment on your quote from the Wayne patent; you'll notice that the system had changed from the original patent, gone are all the phase inverters, and the signals are now applied to the cathodes of the LF oscillator, the circuit having been streamlined substantially; the 120 degree three phase signals appear on the cathodes of the LF oscillators and the operation of the oscillator stages does the gating of those signals without interfering with the normal operation of the oscillator. This improved the system, and cleaned up the output, the gating action following the transition of the oscillator stage from conduction to off, not exactly linear, but sufficient so that there are no DC level changes in the joining of the three 120 degree shifted signals on the output. This was supposed to allow complex waveforms to go through the unit without little clicks as the gates shifted the signal back in time, and the DC levels where they appeared were different.
In other words there is a very quick "fade" of the signals from one to the next. I'm assuming you know all this, but just reiterating for clarity. I think that was your point in referencing the Wayne patent.
(I think I met Wayne back at the Baldwin factory when I was there for training)
Anyway.............

[daveweyer]

One more little clarification, the microsecond gate time I mentioned before is only a guess at how fast a 12AU7 can turn off a signal. In the circuit, the total open gated time will apparently be as long as it takes the tube run its portion of the cycle and to turn off or near off during the process of being cut off by the LC oscillator frequency. voltage. On the 1 HZ oscillator the total time should be a third of that.
The new method Baldwin used to introduce the signal into the RC stages was at the cathode, the non-inverting input--it saved them all those phase inverters and it put any extraneous capacitances and inductances into a very low impedance, rendering them unimportant in the equation. The modulated frequencies are introduced at a constant level relative to input so they continue try to turn the tube on and off with an AC current during that lower conduction zone, the cathode swing from oscillations in the 3 to 5 volt P-P range and the grid swing in the 16 volt P-P range using a Simpson meter as per their voltage chart. The 12AU7 should be off at about -24 volts, so whatever they mean in the patent when they say cutoff, it appears that it really means not conducting enough to matter. Since the drivers, the 6CM7s have no AC voltage shown, we'd have to measure the modulated voltage at full input to know how much of the modulated signal gets clipped at or near cutoff, if any.
As I also mentioned, I don't think the 12AU7 has a remote cutoff characteristic per se, but it does have a long range of cutoff voltage which are fairly evenly spaced as you can see in the manual--you could have a pretty large signal going through it and still not reach crossover distortion even at very low transconductance, this may have influenced the choice of tubes for this part of the application. If the voltages are in the ballpark on the schematic, the tube may just introduce some non-linearity in the modulated waveform at the points where it switches from one tube to the next.
With the torturing of the signal that goes on in the choratone, the little harmonic distortion is probably of no consequence.

[daveweyer]

And I guess it goes without saying that the modulated signals have an amplitude variation which is the same shape as the LF oscillator waveform portion, like a bunch of little humps glued together. The circuitry following the modulator outputs tries to ignore the LF content, in some ways resembling an inverted superheterodyne radio.
I still can see ways to gate the input signals without so much amplitude variation, don't know if that should be a distraction in the current line of thinking.

[Eb7+9]

thanks for that info Dave,
much to digest ...

I'm taking a stab at the "linear"version first - for one, because it's easier to trim parameters afterwards ... it's also easier to scope, and interpret performance ... we can also play tricks to enhance overall fidelity (see below) ... the non-linear jFET emulation version is steeping in the back of my mind - and will probably by quite close topologically to the tube version

I guess it might sound like I'm going toward something way different in nature ... to explain a bit better, I'm just taking a guess that a "linear" approach might bring us closer to a more "ideal" rendition of the patent idea, or not (might be a total flop and not do anything ..) // either way, we'll find out ... once I see what the linear version does I think I will be in a better position to attempt a class-A equivalent ... hopes for that are up as my TINA-TI simulations (seen above) show the jFET 3-phase oscillator doing its job ... just a matter of double checking for sufficient transconductance swings, and also evaluating impedance levels in order to duplicate the injection of signal like the tube version does .... and then I could use the same op-amp front end, or do all that class-A as well (w maybe BJT Darlington pairs to bring the front-end Zout way down) ...

so, if successful all around let's expect two versions (I'm hopeful myself ...)

I should share this, funny how things go ... if you recall, the other day I mentioned how this processor reminded me of other tube-based modulator examples ... well, in the process of drawing the schematic for the linear prototype, I suddenly realized that I could use the same idea to emulate the Fender 6G series Vibrato effect, where inverse-phase LP and HP signals are progressively summed thru an LFO modulated long tail pair (ie., variably "mixed" in time) ... to implement it I settled on the idea of running the Univibe oscillator (for a more vintage feel) between 9v/-9v and setting the Darlington LFO output close to 0v by trimming the oscillator's bias voltage ... allowing me to pair that output to the same linear VI-converter and OTA arrangement I'm using in this first version of the Choratone emulator ... nice thing is I can independently set the LP and HP frequencies thru switch-able caps and not incur any mutual loading (as they do a bit in the Fender circuit) ...because there is no NFB like in an op-amp circuit I can simply swap the OTA inputs on one side so that the HP and LP signals add in-phase or in anti-phase at the output (something I've discussed with other curious peeps in the past, and finally get to test out easily by the flick of a switch !) ... also there are no jFET's to match or characterize and the headroom is high // so, easy and straightforward to build (if you have a scope) and it will work clipping-free on keys, bass, FX loops, line level signals, etc ... also, there are NO signal caps throughout // so, DC-infinity fidelity, and the circuit is designed for low noise levels as well ... a whole bunch of advantages fall out from taking this general approach ... looks so good I just had to draw out a PCB and send out to my board maker this AM ... (bonus ...!)

a very-cool offshoot of this work//

[daveweyer]

JC, will your linear approach be using the same idea, that of using a three stage phase shift oscillator to gate the signal to be modulated? If so, will the action be from rail to rail? The classical circuit works only because the tube grid, being driven negative by the positive feedback eventually cuts the tube off with the plate at the rail. When enough electrons leak off the grid to start the tube conducting again, the the plate starts going negative and the positive feedback drives the grid more positive unit the tube hits saturation with the plate at minimum voltage. This action repeats indefinitely, slamming from cutoff to saturation. If it weren't for the capacitors in the RC network, it would be a square wave.

Will your linear circuit operate the same way?

[daveweyer]

I'll try to have Xplorer post some audio samples of the Choratone tube modulator on his sound cloud.
(Sending him and all our French friends our thoughts and best wishes during this painful time in their country.)
I did a running commentary so you can read along to know what's playing where. If he gets that file posted please, please comment. You won't be interrupting anything and you might help JC with his new design.
Keep checking.

[Xplorer]

Thank you Dave for these words. yes i'll post the clips with pleasure, i need to focus on some other things a bit ...
yesterday it could have been me, i often go in these bars , and they killed 14 peoples 10 meters from my favourite pizzeria.
we're astonished, chocked, and the sad thing is that it becomes a habit now.
calling my sister, my friends, everyone calling each others just to see if they are alive, nightmare ...
And when you think that it occurs quite often in various places in the world, in Africa, etc ... and peoples aren't mobilized like we do for France today, it's sad.
we're all, like you, sad for the victims and their families, not only for the victims from yesterday.
i hope that russia, usa, and europe will exterminate daesh soon.
they may kill some frenchs or représentants of countrys who fight against this non sens super violent evil, but they'll never kill France. they will just earn they ticket to death sooner.

[Eb7+9]

JC, will your linear approach be using the same idea, that of using a three stage phase shift oscillator to gate the signal to be modulated? If so, will the action be from rail to rail? The classical circuit works only because the tube grid, being driven negative by the positive feedback eventually cuts the tube off with the plate at the rail. When enough electrons leak off the grid to start the tube conducting again, the the plate starts going negative and the positive feedback drives the grid more positive unit the tube hits saturation with the plate at minimum voltage. This action repeats indefinitely, slamming from cutoff to saturation. If it weren't for the capacitors in the RC network, it would be a square wave.

these are important details Dave ...!!

I gave it some thought, and upon looking at the sweep curves I provided above (for 12ax7, 2n5457 jFET, and 2n5210 BJT) they show that much of the "transconductance variation" occurs very near cut-off, as opposed to being spread out evenly starting from Vin = 0 middle point

I had never considered this before, and it explains why/how a waveform that nearly (or totally) shuts off the class-A modulator stages would be desire-able, even necessary (if my understanding of gm variation is right here) ... I always assumed it to be more evenly progressive ... come to think of it, in the day there was such a thing as "remote cutoff" tubes // possibly intended more for such applications ... (?!)

btw, this might explain why nobody really designs/builds jFET-based AGC circuits where the front DC bias is shifted down to drop gain, as it would take the circuit too close to shutoff and therefore rendering it majorly susceptible to distortion via headroom loss ... so, if my assumption is correct it would explain why we need to see a choppy (rail-to-rail ?) LFO in the tube version as the modulation factor exhibits these near cut-off properties ... (correct me if I'm not interpreting things right)

either way, I think it's pretty clear (by quotes posted above) that Wayne identifies "continuous gain variation" as the mechanism responsible for creating his phasing effect ... this was my original assumption when I first saw the circuit, seeing that tube based LFO's like this one generally don't produce very clean sine-waves ... still, I see it as "aiming to do so" // something that we can do (more easily) using OTA's ... AND, in the process, maybe create a more "even" (rd. linear) modulation // just to satisfy my curiosity ...

I can still overdrive the V/I converter circuits to produce more of a gating effect, but I don't think it will produce the same "effect" because of the linearity ... but we'll see, I'm setting things so the option can be induced if desired // otoh, maybe the "choppy" quality of the effect is important here, from a "feel" point of view ... in using linear modulation we'll see if that component will be "missed" in the end ...

to get back to your question, YES ... I am starting from a separate 3-phase sine-wave oscillator (shown above) and feeding that to linear V/I converters and passing on the current output of those translators into the bias port of three OTA's ... fed by A, A + 120, A + 240 signals from the lattice network and inverters

as far as modulation goes // I think you need almost square-like waveforms to achieve wide range gain variation in the tube version ... but I can achieve the same by running a carefully tuned sine-wave on an OTA ... this is std. design // nothing out of the ordinary here ...

Will your linear circuit operate the same way?

I think so, but we'll see once I have it running,

I am very much mimicking the structure of the original circuit (adding an extra octave to the Baldwin Chora-Tone to reflect Wayne's Patent) ... except for how the modulation takes place, everything else is pretty direct emulation ...

to be specific about it again, all I'm doing is decoupling, so to speak, the oscillator-modulator section and using separate linear modulator circuits ... all in the name of tune-ability ... this way, I will be able to (potentiometer) select the output level of each of my 5-octave bands independently ... and also, will me able to adjust the modulation "DEPTH" on each oscillator ... while I'm here, why not test out these possibilities and see what variants on the main/original idea might yield us ... it's doable // just a matter of cost (which I'm ignoring for now)

as far as electronic circuit options being made, everything else in there will be op-amp equivalent gain blocks (simple, dist free, low noise) ... with gain tune-ability where it applies best ... and the option of either passive or buffered outputs into the resistive mixer circuit seen at the output ... I am drawing a board that will reflect all these options, and more ...

so far I've drawn the x5 modulator section (see attached) just to get an idea ... the only thing holding me back is the footprint for inductors (whether I use singles, and add multiples) // haven't figured out that part ... waiting for test coils to arrive in the mail

your insights are very much appreciated Dave ...!!

peace to all
~jc

[Xplorer]

very nice work JC , are you already at the point where you can do some early sound simulations, through the circuit like we did for the fuzz tests ?
can't wait to hear it !