The FM Tube Modulator Jimi Loved

[Eb7+9]

oh geez ... sorry for that typo



I appreciate your comments Dave,

I've thought of slowing down the switching transitions in the ring counter design using caps on the gates (the obvious approach) but then these would be adjusted and optimized for a specific operating speed ... it would do away with the instantaneous "chopping" effect ... but when speed variation is introduced that would all start to fail and some chopping would appear at some point in the speed range

(can't win ...)

I'm not sure how a four-quadrant multiplier would smooth out transitions either as multiplier circuits are dead fast as well ... maybe you could supply a reference to a synth circuit and I'll have a look

I'm in amp repair/modding mode these days ... in fact, working on a unique amp design that resembles nothing under the sun // ie., finishing the last unit that a good friend started before he passed ten years ago // you might have known him, or of him ...

http://www.cbc.ca/news/canada/manitoba/ ... d-1.636051

and I'm also developing a new (high headroom) optical WAH circuit ... and possibly approaching one of the big design houses with the idea of manufacturing one of my "emulator" concepts ...

a bit much on my plate right now ...
likely I'll be getting back to the CTP monster this winter

hope you're feeling better after your accident Dave ... thx for chiming in !!

best, ...
~jcm

[Tek465b]

Four quadrant multiplier. Thats funny Dave got this idea too..

Remember a few week ago when we where chatting on facebook? i talked about using quadrature IQ signal to do SSB modulation(frequency shift)? exactly what i was thinking about.
for the forum reader(frequency shift = continuous phase shifting..).

Split the audio signal by 90 degree(just re-tune the lattice or use all-pass filter), split one of the arduino 50%PWM output signal into 90 degree using flipflop(that will be our quadrature clock). the flipflop can also be used as squarewave frequency "clock" divider AND 90 degree shift at once..for each extra sub-channel
Then you add or substract the signals output from the mixers for upper sideband or lower sideband selection(frequency shift up or down).

One clock output only is needed (fixed 50% PWM, variable frequency)and we can use as much "channels" as we want. just extra flipflop and 1 extra quadrature modulator(four quadrant mixer) for each channels.(+ the audio mixing/output circuitry)

But i also dropped the idea because those mixer are fast, and since we drive em with low frequency squarewave clock(this give me a bad feeling)...(i think, but i could be totallly wrong there) I still beleave it could work as-is, i just don't feel comfortable with it..
Or we could use integrators for a trianglewave quadrature clock?
So instead we could use the frequency shifted audio signal from the previous channel(instead of the original audio) through another latice/all pass filter and another quadrature modulator for additionnal sub-channels(this allow us to re-use the exact same original trianglewave quadrature clock signal for each channel)(ok i probably sound stupid there, maby not soo much)...
1khz + 2hz = 1002hz
1khz + 4hz = 1004hz
and
1khz + 2hz = 1002hz
1002hz + 2hz = 1004hz.
same result just different way to get there.

But then we could also simply move away from the squarewave uC oscillator.
Then use one variable frequency LFO with sinewave output.
We just need to split it's output by 90 degree again using lattice or all pass filter(for the quadrature clock).
We then have the same 2 option, multiple clock divided by 2 for each channel(using mixers i guess?), or re-use the last stage audio signal that is frequency shifted and run it through the lattice/all pass filter then quadrature modulate(frequency shift) it again.
Then we have a 100% analog solution. with 1 oscillator and many channels.
All we need is 1 extra lattice and 1 extra quadrature modulator for each additional channel. Bam.. i love myself loll.

Phew i hope all this make some sense?

Let me know what your thinking about the idea. Still sound like a cool idea to me, and am glad Dave got the idea too.(i know its alot more circuitry and IC then we'd like, but i still think its worth checking into.)

[daveweyer]

No, I don't think it's stupid at all.
It is very "analog synth" like in conception. It seems entirely possible to me to use integration to get the wave structure you need, and cheap too. Remember that synths used all kinds of waves to get various envelopes from the four quadrant multipliers. Square wave to pseudo-sine with two components.
The old idea from the first divider organs; get as many frequencies as you need from one clock, if you start with a high frequency clock you can have a lot, and all tunable with one component, in the organs usually a slug in a coil.
There are a lot of possibilities still in the analog realm worth thinking about.
Let the ideas flow........
My friend Bob Hovland, the engineer at VOX, always wondered why Wayne didn't use 90 deg phase shift implementations in his design. Brad Plunkett, Bob's boss at VOX, actually did in his version of the frequency shifter.

BTW I loved the typo!!!

[daveweyer]

J.C.
I just checked out the amp link you posted. Such a very sweet remembrance. I hope they give me something like that. Can you share anything about his amp design?

I don't mean to get repetitive, but if you get the time, please check out the transfer curves for the depletion mode mosfet LND150. Those curves look an awfully lot like a triode (even though the output characteristic looks like a perfect pentode), with a very nice ramp to complete cutoff of -20 volts. Between 0 and -2 volts on the gate you get a drain current of 5ma down to milliamps, and as you go more negative, down to microamps before complete cutoff.
It cuts off faster than a triode, but there is still room to get a smooth turn on before the drain current slope gets much steeper. This is a Supertex product, with a 500 volt rating, so it can be operated exactly like a tube, with negative bias, also meaning local feedback of the inverse current variety for further linearity. It looks to me that this may solve (with a little work) the transfer of the Wayne idea to the SS realm, and save you a thousand parts.

[Eb7+9]

Dave, I checked out that MOSFET device, it looks pretty typical other than the voltage range ... still a MOSFET in character, and not that different than jFET's ...

let's backtrack a bit here // sorry if this long-winded

Wayne's first (original) patent, which is embodied in the Baldwin Chora-Tone Projector, operates a three-phase ring style oscillator using 12au7 triodes ... that style of ring oscillator can be built using any transconductance device, ie., BJT, jFET and MOSFET transistors, and also triodes and pentodes - where it all started ... ring oscillators are also built using CMOS inverters and op-amps ... one thing voltage-mode ring oscillators "all" have in common is their fixed-speed character since the cycling rate is based primarily on capacitance found in the circuit // either parasitic (in the case of high speed oscillators) or discrete ... in order to vary the speed we need to be able to vary three capacitors at the very least, no way around that ... this cannot be done continuously or electronically (if a solution did exist it would be out there for sure) ... if you recall, somewhere on page 7 of this thread (viewtopic.php?p=456218#p456218) I started figuring out how to make a speed controllable current-mode version of the ring oscillator and I succeeded in taking a translinear gain block and using it to make the oscillator speed vary continuously ... that was on the simulator, I did draw up a PCB but haven't yet populated it / main reason I sidestepped that effort was because of the level of hardware complexity ... then I built the first version using jFET based ring oscillators as you'll recall ...

Wayne's ring-oscillator-modulator circuit operates no differently than any of the other transconductance based variations ... the audio signals (labelled A, B and C in the Baldwin schematic) are piped into the low-Z node of the three ring devices which are also acting as a AC current buffers where the input impedance (set by 1/gm in all cases) varies up and down with bias, in ring fashion ... now, depending on which kind of transconductance device we use to embody the ring oscillator-modulator a different grade of headroom comes along for the ride ... as I simulated and measured on the bench, the (matched/graded) 2n5457 jFET version only gave about 1volt pk-pk headroom before clipping // Vgs(off) was measured to lie around -1v for those devices ... I assume that the original (12au7) version gave better specs than that as the unit was originally intended to be used at the speaker output of a main amplifier ...

so back to your MOSFET, what we need is a good spice model for that one - but I haven't been able to find anything reliable (yet) ... I have no problem building a high-voltage equivalent of the 12au7 circuit using this MOSFET but now we are stuck with the original issue (mine) of having a fixed speed ... is that ok with you ?! I agree that we'd be playing with a closer "analogue" of the original VT circuit, I'm sure anybody in this forum could pull it off ... in my opinion, we need to know beforehand what to expect at the headroom level // and that's where SPICE models are needed ... if anybody finds them please post here ... a thourough study would compare the headroom of the original 12au7 version to the MOSFET ... in the case of the jFET devices |Vp| is the limiting factor, I suspect the same holds for MOSFET's ... question is, what's the turn-off voltage of these LND150's ...?! data sheet says -1v to -3v ... which is okay I guess, won't give hear the headroom of the opto version ... I just wanted to point that out as I think it's a major variable in this invention ... question is, will these HV MOSFET's give us better performance than what I did with matchd jFET's ...?? again, depedns of Vgs(off) ... datasheet says we can do no better than 3volts with LND150's

anyway back to the main idea here, things changed when I found Wayne's 1970 patent ... where he uses NPN BJT transistors for the ring oscillator but instead of piping signal into the (low-Z) emitter ports, he uses those nodes to drive incandescent BULBS to shine on isolated photo-cells to perform linear time-variable mixing ... it's a patent and we don't know if he ever built a working unit, I can only assume that he did // we'll probably never know for sure ... doesn't matter tho, all I did was take the idea of using photo-cells to mix signals with and use Tek465b's AVR board idea to drive optocouplers, bypassing the fixed-speed NPN driver module ... now, one could argue that there's a certain level of complexity in using an AVR board, but with the advantage that their PWM modules can drive LED's directly ... personally, I regard this as moderately complex in terms of hardware // certainly an acceptable tradeoff in obtaining speed variability ... the other advantage here is absence of headroom limitation, there is none with photocells ... and the limits in the Lattice circuit is set by op-amp voltages alone ... this is win-win as far as I'm concerned ... beyond what Wayne's original approach did (?!) I'm assuming that's the main reason he proposed a second approach // still using the basic thinking idea ... as I pointed out previously, higher headroom translates into better S/N when used at the output of an amplifier, or in a wide-range FX loop ... why I'm so obsessed with it in this re-design

Tek, I don't know about using SSB modulation and all that // and what about dynamic headroom ?! radio message signals tend to be hard-limited for this reason if I'm not mistaken ... I re-read what Dave said above, I'd like to see schematics for the VOX stuff you mentioned ... but seems to me all this frequency shifting stuff is just begging for major headaches ... we're talking high circuit complexity for one, it's easy to make it sound simple in wording // if you think I'm wrong here please supply more info, incl. a drawing or something ... there is no magic with multipliers, in this capacity they would operate equivalently to switched gates as I did in my last version of the circuit, so far the driver signals haven't changed, still PWM ... I think I follow what you're saying Tek but worry you could be introducing a whole whackload of secondary problems to deal with later (otherwise I think the DIY world would be all over this by now) -again, pls correct me if you think I'm wrong ... your idea of using integrators to turn a square-wave into a triangle wave doesn't work unfortunately (yes, crossed my mind) ... if it did we would have fixed height square-wave to triangle-wave converters, but those don't exist // there's a mathematical reason behind it (causality) ... only in a closed loop config (typical FX LFO circuit) do we avoid output swing variation when using an integrator ... that's why I agreed to go the AVR route and asked you to program it as a triangle-wave modulator, in my mind this is why (ultimately) the standard PIC structure's existence and uses are justified - ie., to do things we can't do with analogue circuit design ... please send me a drawing of your SSB idea, I'd like to check that out a little closer ... again, if this were a viable solution I'd think we would have seen that before somewhere else in audio gizmo land // I've never heard of anybody doing this ... maybe you'll be the first ... the easiest approach would be to use a Johnson counter and then find some kind of pulse shaper, but again that flies in the face of causality when speed control is introduced ...

Gar Gillies was a great man, he is best known for creating the American Woman tone that the Guess Who had // the Herzog was the name of the unit used there, simply a modded high-gain Champ variation ... he was the maker of Garnet amplifiers, one of two major Canadian amplifier brands, the other was Traynor ... when I was growing up Garnet and Traynor amps were regarded as "practice" amps and you stepped up to a Marshall when you were ready ... though, a couple of exceptions come to mind: Pete Traynor's Custom Special and his mega huge Super Custom Special (the loudest production tube amp at the time, which some say made Ampeg come out with the SVT) and the Garnet "BTO" , Gar's answer to the SL100 ... to me Garnet BTO is the Kodyak bear of 100 watt heads, sounds more agressive and balsy than a Marshall ... quite an amazing piece // Gar told me that a little over a hundred of those were made ... the Herzog is still being made in Winnipeg by Pete Thiesen (with Gar's permission) ... Pete sent me the last amp that Gar was working on, thinking I was the guy to complete it // ... I spent a few years thinking about how best to celebrate Gar's legacy in this design but had to put it aside due to work, homelessness and a couple of motorcycle accidents since and have only recently started looking at it again ... the main problem there is deciding what to alter and not (out of respect) // Gar had the power supply and output stage all wired up ... by the time Gar stopped making amplifiers he'd been trying to emulate what Mesa was doing with their 1x12 high-gain designs in a model called the Enforcer ... but I didn't want to go that obvious route, instead I decided to make use of Gar's most famous design and find a way to incorporate it into this chassis ... in the process the amp will have a very cool "option" not seen in any other tube amp // I'm still working on the logistics, which I'm afraid must involve cad drawing the chassis as it is now and sizing parts, etc ... yup, it's a party!!

cheers ...
~jcm

[Eb7+9]

like I said, I haven't given up on the opto versio, I'm just looking around for something that might do better ... esp. in the hardware dept. ... as it is we'd have five modulation PCB's each with their own AVR board, a triplet of matched opto-couplers and the corresponding output BP filter amplifier stage ... another board could be used for the latice and final mixer

tho ... as we discussed also // at the end of the day I would be very happy working with an AVR board that can put out 4 or 5 sets of 3-phase clocks ... with the option of having them run sync'ed or not ... with master speed variation, and not ... but I know it's a tall programming order on your part Tek and don't have the resources to compensate you with yet (maybe we can do a kickstarter for this project)

the only thing involving hardship would be the purchasing of an small lot (maybe x50) of opto-couplers and going thru the trouble of matching them into "close" triplets ...

at this stage I think this would be the best we could do without involving too many extraneous hassles
~jcm

[daveweyer]

Wow, nice post! Full of ideas and hurdles.
If I am not mistaken, what you are looking at in the mosfet is how much signal you can input at the source without clipping the top of the waveform off. One thing you would need to look at is the effect of inverse feedback at the source/cathode. Since the biasing resistor tends to work opposite of the signal on the inverting node, the amount of signal which can be applied at the gate is increased, and the voltage swing on the source/cathode is also increased. This has the effect of raising the amount of voltage which can be superimposed on the non-inverting node. What I have observed say on the 12AU7, is that it takes about 10 volts to shut the thing off, and by that time, the distortion in the gated signal caused from being larger in amplitude than the bias voltage is essentially inaudible. So there is a little roughness at the lowest levels of transconductance, something you wouldn't get with a lamp and photocell, but something which is not very important either.
It looks to me like you could get maybe 5 volts of swing with the mosfet using a similar sized source resistor, but that would have to be calculated by the transconductance of the chip, which is probably higher than the tube.
I asked Bob Hovland (the engineer at Vox, JBL, and Marantz) to share his thoughts on the matter and he advised:

"Check out pages 2 and 3 of the LND150 spec sheet, I attached it again
so it might be easier for you to find.

The gate is ESD protected, so to me this means you don't need any
back to back zeners on it to protect it. The spec sheet shows the
gate can take ± 20 vdc, so you should be able to get it to cut off
kind of like a tube; however, it may have a bit sharper cutoff curve than
a tube.

Best,

Bob"

So from his experience I take it that we would be in the ballpark with the mosfet. It boils down to the fact that a longer ramp to cutoff means less matching, and more dynamic range. All we really need is that the gated signals don't "pop" on, and that we have a reasonably stable amplitude between the three gated signals.

As far as frequency shifting goes, I have gotten some range change by changing resistors, probably enough to satisfy the needs of the musical world. Listening to speeds above 8HZ seems to get more experimental and less musically useful for "songs". I just don't know how important frequency changing is with such a complex signal to begin with. Switching a few caps, which is a pretty simple procedure, could work much better now than in Wayne's day because of the incredible size reduction for those larger values, that is if one really needed that option. I'd love to see the comparison in parts count and complexity between these various ideas.

I know from experience that everyone was trying to get into the frequency shifter game by 1970, but that SS devices were the only allowable option because of cost and complexity. Tubes were out forever according to the engineers, as they went about their illusory challenge to replace them in every application. That's why Wayne went to optical, and every other device/engineer I saw back then except for Don Foster's "The Foster Freqy" a unit I worked on with him before he died which translated the signal into RF and did the modulations there. It was however SS.

According to Bob Hovland, the closest thing to a tube we have in the SS world is that mosfet. I'd love to try it myself but have no time.
Maybe this winter will have a few spare hours, but then..........

[Tek465b]

We did alot of progress latelly.

No more matching needed. No more opto-coupler, no mosfet transistor .
I will let JC explain the technical detail when its ready. Still some testing to do but i think we are real close.

[Eb7+9]

Hello again everyone ...

as Tek said we've been doing some cool work lately
here's an update ...

the last version of the single channel CTP circuit saw me attempting to do the
triple-channel mixing through PWM controlled Gate resistances
causing "stepping" noise to ride along with the mixing mechanism
... which otherwise worked fine !

recall this was simply an alternative to the optical version,
which required matched cells to operate properly

I did this to see if I could match the overall behavior
and do away with the problems involved in opto-cell matching ...
(similar hassle as dealing w jFET's)

for that reason alone the idea of using switched analogue (bi-directional) gates is tempting
because matching is now built into the IC by design (with four gates per CD4066 chip)

part of the problem lies in the fact that there is no theory on this stuff

switched capacitor circuits are commonly studied
but not switched resistance

even though MXR used the idea in their Envelope Filter in the 70's
I know of no audio DIY who's done any theory or basic experimentation on the subject

I felt it was something I should do, and now could with tek's help ...
heck, I needed those answers for myself anyway

so, I got Tek to write me some new code

what we did was knock a couple of birds with one stone
by designing a stereo phasor based on a well known approach
ie., using two Phase-45 circuits side by side running in opposite phase

the idea in the end is to have two perfectly sounding and identically behaving channels
I've come pretty close using my optical matching techniques, but it's never perfect

http://www.lynx.net/~jc/pwm01.jpg
http://www.lynx.net/~jc/pwm02.jpg
http://www.lynx.net/~jc/pwm03.jpg

since op-amp based all-pass filters (what P45//P90 phasors are made of)
have a very predictable Bode response, that is,
Magnitude and Phase response versus frequency
I could use my DSO-QUAD in spectrum analyzer mode to infer the equivalent resistance
by seeing where the characteristic notch lands

we did this by programing an Arduino Pro-Mini board to spit
out a minimum duty cycle of 0.39% ... ie., 1/256 ... at 32khz

in doing so I was able to play with the OFFSET control
and manually set the dutycycle to verify static operation first
this comes in handy later when I try to "avoid" a certain range if needed

this is similar to the external BIAS control in my P45 mods
which I refer to as a "COLOUR" control ... same deal here w OFFSET on the PWM

by going to the maximum 100% and minimum 0.39% statically
I was able to create a notch at 40khz and 40hz, respectively ...

http://www.lynx.net/~jc/pwm04.jpg
http://www.lynx.net/~jc/pwm05.jpg

since the phasor cap value is 0.01uF in this test
and we know the notch occurs where the equivalent resistance of the switched gate
is equal to the reactance of the capacitor
we can infer Requiv by using the following equation:

Requiv = 1/2piFC

which leads to 400 ohms (at 40khz) and 400k ohms (at 40hz)
showing that a 1000:1 resistance ratio is possible in an 8-bit setting

which is very much in line with several optical methods I used with NSL32-SR3 Silonex cells,
see here for details:

http://www.lynx.net/~jc/NSL32-SR3modeling.html

in the process of building my Stereo Phaze prototype I was discovered that the CD4066
is not the best choice for the job since it has a topology that leads to increased clock feedthrough
I just noticed that yesterday

by switching to the simpler CD4016 bi-directional switch IC I was able to do away
with the noise and generate a full-range stereo phasor operating cleanly ...

almost flawless behavior, as good if not better than using optical means
so - it's a milestone for me (maybe for the DIY community)

further tests and variations need to be made to see if I can go beyond the 0-5volt
environment and use the idea in high voltage settings

a PWM'd-gate Univibe is in the works as I'm always into checking out
alternative ways of making the Vibe happen ...
my opto-Vibe for example is all current-mode and operates using matched opto-couplers
this will make a radical alternative as well

as far as applicability to the present CTP research project
first thing will be to open up the PWM version (shown above)
and swap the 4066 IC for a 4016 and see if I get disappearance of noise
I spoke about there ...
and thus yield a better behaving mixing function ...

the only thing that might bite me there is a slewing issue with the comparator circuit
I'm using to convert Arduino 0-5v pulses to -7.5v/+7.5v ...

we'll see

recall, the optical version doesn't have that problem

so, if I'm successful it means that component matching will be done away here
and will make it much easier to build a full deal 4~5 channel CTP circuit ...

I don't care how big the sucker gets in the end
the cell-matching was really the strongest constraint holding me back

you never know, I might be able to pull this off at some point this winter
depending on resources, etc ...

that's it ... in principle this project is still alive
I wish I could solicit help with funds but it looks like I'm on my own as far as development costs go
so, for now it's matter of how much stuff I can sell at my store, how many amps come in for surgery, etc ...
which is starting to happen again, now that word's out locally

http://www.viva-analog.com

hope you're feeling better Dave

thx for watching // best regards, ...
~jcm

[Xplorer]

Long life to this very cool super Advanced project ! thanks for sharing all this development.
Selling some prepared kits with a well deserved commission for you, Dave and Tek on some group buys could help you perhaps ?

[Eb7+9]

Hi again,...

just found your recent video on YT Dave,

https://www.youtube.com/watch?v=U5VjZ5YFiH0

I noticed mention of the CTP and the work we're doing here
cool view of your lab !

the bit on Jimi's Woodstock-WAH is fantastic ...

so cool to finally find out how it was done, esp. the "power roach" resistor
(LOL)

great to see the real (10k) Fuzz-Face interface resistor also in there ...
years ago Mike Fuller told me it might have been a 22k ...

I always wondered

thx to George and whoever for getting this forum up and running again
hope you're feeling better George ...

---

electronics wise, it's been a busy fall/winter for me ...

I recently did a bunch of PWM-gate theorizing and found some practical refinements which has paid off
with these resources I was able to design and build a PWM based Stereo Phazor and PWM based active filter/wah ...

both exhibiting smooth behavior with good noise specs ...
circumventing the matching aspects I typically see in my other opto units

with the help of Tek465b I've been able to get a duty cycle minima of 0.39%
and get a very high range of resistance variation (400ohms to 400k easily) ...
I was able to confirm these numbers from frequency response plots

I felt I needed to look at the technology again following my attempted PWM version of the single-channel CTP ... (which I felt wasn't as good as the optical unit as you might recall - mainly due to strange noises)

in the process of doing the other two PWM circuits I found out how to eliminate the aliasing noises
I was getting in the PWM-Wayne unit

actually, simply/accidentally by using a CD4016 instead of a CD4066 gate chip ... (luck!)

so, a pretty fruitful detour of sorts

---

fast forward a bit, more math based design work ...

a few weeks ago I started taking a different approach to doing a 2/4 stage jFET phasors
fans of the MXR Phase45 and Phase90 might want to take a look here for details:

http://viva-analog.com/the-paradigm-shifter/

(a PCB for that project will be released in the next few days ...)

as I mentioned, I also built a PWM active filter this week ... a DUINO-wah if you will

basically it's an MXR Env Filter with three LP/BP&HP outputs ... w precision foot control,
with variable "Q" and variable loop gain (into an oscillator eventually) ...

kinda trippy / sometimes a WAH, sometimes a SYNTH
vid on that coming soon ...

---

I've also been asked to re-do my bulb-based Stereo-Vibe // all class-A ...
inspired by the Magnatone 480, but done in compact solid-state form

I'm currently finalizing that design (after a year of mulling it over) and drawing a board for it shortly
it's based on the Univibe, but operates differentially and does away with the low-speed losses
the original Univibe had ... something I've been pondering for a long time

---

now, back to the CTP ...

everything seems connected in my little world ... the stuff I did recently with the PWM testing
gave me an idea of what I can get away with as far as "simplicity of build" is concerned ... and not

like I said, and to recap again, I've visited three approaches so far :

(i) a direct jFET emulation of the modulator/oscillator ... which requires trios of matched jFET's to work well, and with limited headroom as main drawback

(ii) the optical mixing version ... which requires trios of matched opto-isolators to work well

(iii) the PWM mixing version ... which I will re-visit with a CD4016 and see what I get ...

==>>>

so, essentially, the game here is to mull out an arrangement that has the least damning constraints ...

any matching gets multiplied by the number of channels ... if we follow the Baldwin approach that means x4 // if we follow the patent it means x5 ... as I showed, matching jFET's closely in trios is not a piece of cake // as it requires many devices to start with ... the optical matching is easier to do, but still costly

so, even though I really like the sound quality of the single-channel optical unit matching opto's is not DIY friendly ... either way, matching should be out of the question

PWM-gate technology is something new that I've explored lately and I find the technique difficult to work with from a layout point of view ... and I'm hesitating going that way (for now anyway, even if the 4016 chip pans out in my previous PWM-Wayne unit)

==>>>

so, I started looking at a fourth option

(iv) a PWM-ota approach to doing mixing ... something I had to develop

the only drawback I can foresee is a light increase in noise, which OTA's are known for - that's the only thing I'm worried about here ... but the PWM-gate system makes noise too in the high resistance range limits (nothing is perfect here), so does the optical cells anyway ... so, why not try a PWM-ota system ... the DUINO part would generate speed/intensity controllable clock signals, ota's then perform the mixing operation and other tasks ... seamlessly, thanks to the advantage of playing with signal in current form

indeed, I noticed that the idea lends itself well to mixing (by adding currents directly, and without losses or attenuation) ... and the Band-Pass filters can be incorporated as load at the summing node as well ... so, in essence, knocking a few birds with one stone ... the mixing of all channels can be done easily, and with an increase in levels (maybe x10) at the back end

this might well be the light at the end of the tunnel I been looking for ...

then, I think that I might be able to come up with a complete CTP package that won't cost and arm and a leg in parts ... if possible (I have to ask Tek465b) I'm shooting to have one AVR controlling two channels, to save on board real-estate // each with its own set of OFFSET, DEPTH and SPEED controls - each tied to (and controlling) a trio of ota's via control currents ...

so the end product would have, say if a Baldwin style 4-channel unit, two AVR boards, and 6 LM13700 IC's to do the modulation (3 ota's per channel) ... I think this will be the easiest and cheapest to build ... with no matching of any sort // ...

so yeah, that might be it folks

I'll be checking this out in the next week or so
gonna have some cool stuff to lay on y'all ...
~jc

[Tek465b]

Oh nice, i just discovered Dave's youtube channel, its great .

Wondering what is the value of the capacitor under the pcb thats in parallel with the .01u ??

The PWM wah look very interesting, it pick up my attention.

The OTA version of the CTP look promising.
I really need to move to a bigger/faster chip for more channel(at that point i could do 4 channel or more with 1 chip). Its not a big deal for me to "migrate" the code to another chip.
There is 2 area of the code where i see improvement can be done too, i could do that at once.
am not gonna lie, its gonna be a little bit of work but i bet the result will be outstanding and well worth it with all the work that JC do.
I did the PWM-wah code at my cottage on my touch screen cell phone in about 3 hour(that include time to eat my diner)?...And i feel like i could have done it in 30 minute only. loll.

Its soo much fun to work with JC

[Eb7+9]

real pleasure working with you too Tek465b ...
thanks a lot for your dedication and attention to detail !

---

so, I was doodling last night and another idea just struckme
we're possibly looking at a fifth option now ...

(v) a PWM-triode approach to doing mixing ... something that might have more appeal to the folks here

for one, high voltage tube circuits have the best Signal/Noise
and the original Baldwin modulators ran on 12au7's

in fact, we could in principle clone the whole darn thing except for the modulator control parts

the idea is simple,

instead of running the original triode circuits as ring oscillator/modulator combos
now have the PWM outputs of an Arduino module generate the turn-on/turn-off LFO voltages ...
the modulation part would take place the same, the signal not be less the wiser for it ...
in the process maintaining the exact structure and impedances as the original embodyment

essentially achieving the original goal (as it were) of bringing SPEED and DEPTH control
to the original tube version ... or similar (mini-tubes ?!)
with level shifting this is easy to accomplish

in fact, modding an original unit with an Arduino control circuit
is yet another side possibility

umm,...

don't know why I didn't think of it before ...
anyway, it would be a very cool retro-modern way of doing the CTP

~jc

[Tek465b]

Here's the video for the wah:
https://www.youtube.com/watch?v=rX8M5-h ... e=youtu.be

[Xplorer]

love it ! great sound !!