The FM Tube Modulator Jimi Loved

[Eb7+9]

btw, for those interested in OTA's ...
from Don Sauer, the father of the 3080 and 13600/13700 devices

http://www.idea2ic.com/LM13600/LM13700.html

check this out ...>>>

http://www.idea2ic.com/TaleOf2Invention ... story.html

may we see these new generation OTA's in our lifetime ...!!

AB_BIAS LM3080

1) measure supply current = 2.045uA 2.0uA

2) Measure Max output current = +/- 2.016uA_pk +/- .97uA_pk

3) Measure Max Vin extrapolate = +/- 196mV_pk +/- 51mV_pk

4) Measure Offset_V = +/- 2.7mV_sd +/- 1.6mV_sd

5) Measure Offset_Iout_%Imax = +/- 1.36_% +/- 3.2_%

6) Measure input noise_rtHz = +/- 110nV_rms +/- 56nV_rms

7) Measure 1% THD output current = +/- 1.50uA_pk +/- .334uA_pk

Measure 1% THD input V = +/- 146mV_pk +/- 18mV_pk

9) Meaured Rgm = Vin/Iout = 97KOhms 51KOhms

10) Total Output Shot Noise = 159pA_rms 151pA_rms

11) Signal to Noise Clipping = 79.0dB 73dB

12) Signal to Noise 1% THD = 76.4dB 63.5dB

[Eb7+9]

at this point, I'm considering heading back to more modest ideas
something closer to the original circuitry perhaps ...

but I've got it in mind that the only way is by using (matched) jFET's //
maybe I'm wrong there ... if I am, then we're in luck

I will start looking into that again

... well, more as it comes
~jcm

[Tek465b]

yeah it look a bit intimidating, but the solderpad are not overly small and are not extremtly closely spaced like ive seen in the icom ic-28h. had to replace an inductor in the dc-dc converter in one of them(and remove/reinstall some other component to test and to get some room to replace the inductor). it was a 2 sided through hole plated pcb (under a metal shield)and it was a pita. Just cleaning the holes and putting back the componenst and shield on the board took quite a while...

[Eb7+9]

thx for having a look Tek,

indeed, working on super-tight commercial devices these days requires utmost care ...
takes good soldering skills

I'm leaving as much as room as I can reasonably on this proto-board
ie., to probe and mod (Dremmel + bridging) if need be later //

which reminds me, ... I need to get a new tip for my Weller

I'm finishing the layout today and gonna throw some bucks at it
should have a board by the end of the week

once the files for this board are sent off, I'm gonna get down and match a bunch of 2n3904's and 2n3906's
the not-so-exciting part of the conquest ...



I'm also gonna take a look in parallel to the jFET approach mentioned before
(which means finishing up on the characterization work for my lots of 2n5457 jFET's) ...

~jcm

[Eb7+9]

I checked out the jFET constant-RATE oscillator, with circuit values a little different from the Baldwin schematic ...

with jFET devices that are matched (ie., identical Threshold voltage)

by altering Vth for the three devices (ie., 1.37v, 1.27v, 1.47v)

conclusion, the circuit also works in the +/-0.1v un-matched case, but performance is clearly uneven ..
producing unwanted performance ...

or not ... (?!)
maybe cool // who knows ... (that's the mystery of analog for ya)

but to have symmetric operation (as per the patent) we need to match jFET's for Vth by much better than 100mV ...
more on this Vth matching later

not an impossible task ...
of course,
going this route would yield to a way simpler/smaller circuit than the OTA method I'm doing here first ...

>>>

back to the jFET solution shortly ... looking quite appealing at this point
especially now accepting/assuming that the 3-phase modulation waveforms aren't supposed to be perfectly sinusoidal

~jcm

[Tek465b]

is it me or output pin 12 on the 3xOTA the trace is incomplete?
the pin 5 output is connecting to the output inductor. but pin 12 output seem to be floating instead to go to the next inductor.

[daveweyer]

JC, interesting train of thought: I've wondered all along what might be simpler than Wayne's idea of using the oscillator stages for gates, to me that was the real genius of the circuit--it just reduced the complexity and parts count by such a factor. If I were responsible for the test circuit right now, I think I would cautiously build a one oscillator prototype on the bench and check out the modulated waveform vs various devices' characteristics--just to get a real good idea about how important matching and transfer characteristics are.
Of course that's kind of the old way of doing things, it's pretty handy for substituting real devices though, say from a variety of part types which you can get from suppliers.
Anyway, thanks for sharing your thought process as you go, it's interesting from a lot of angles.

[Eb7+9]

is it me or output pin 12 on the 3xOTA the trace is incomplete?
the pin 5 output is connecting to the output inductor. but pin 12 output seem to be floating instead to go to the next inductor.

good observation Tek,

all OTA outputs are floating as is //
but in the end they all get connected together to drive that channel's (common) output tank circuit
that is, C/L and variable R as volume control for that channel ... all three L-C-R in parallel

ok, in this case we're dealing with a single "octave" channel
(one of five per the patent, or four per the Baldwin CTP) // sorry if that sound confusing

a single octave channel with stereo (in phase LFO, and out-of-phase LFO) outputs
my enhancement of the architecture (cuz it falls out naturally here)

the reason why I decouple outputs is so I can first trim out the offset for each OTA individually,
before connecting all three OTA outputs together to their common load


the idea:

by setting the voltage across the output load as closely to 0volt as possible
I'm basically infering that the OTA puts out close to 0mA with no input signal
obviously this needs to be done for each OTA inependently from the others
since we could have all three OTA's putting out a non-zero current at idle that
all cancel each other out algebraically ...

we don't want that, ... rather, we want all three to be balanced in their own "minimum distortion" zones
and THEN have their output current summed together

that's what the trimmer leading to the + inputs is for ... Offset Null adjusting

again, there are two sets of outputs in this "low-octave" stereo version
by low-octave I mean the one running at 1Hz and with a 560mH/18uF tank (see Baldwin schem)

those on the LHS (pin 5) for the main in-phase output, say ...
and the ones on the RHS (pin 12) for the inverse PFO-phase output
(which may or may not get used at first - again, those outputs are for yielding a stereo version of the effect)

notice that pins 5 all lead to their own pad just below the IC
those get bridged (individually) to a corresponding pad just below it ...
and that lower pad and other two same go in common to the tank load, etc ...

the ones from pins 12 go to a funny trace shape that I can simply bridge with a blob of solder ...
they achieve the same bussing purpose and lead to a second (identical) common tank+pot load
each side of this stereo unit has its own output volume pot

resulting in the same thing on both sides //
one using a jumper going between pads, the other a direct blob between traces (no pads) ...

again, it's all for dealing with the Offset Nulling stage of the build

>>> all OTA's require this Offset Correction in general // esp. when using them in "linear" mode

to be clear, these need to be corrected independently

the way I would do this, is short one output to the load, null its offset ...
un-solder that bridge, and close the next one ... adjust offset for second LHS OTA ...
un-solder that second bridge, and close the third ... adjust offset for third LHS OTA ...
done with Offset Trim x3 ... bridge all three LHS OTA to the common load ...
>>> now ready to work with minimum distortion

do the same on RHS if we want the other set of OTA's to play
and provide second (stereo) output

---

sorry if this sounds confusing

I'm just sticking as many options as possible since we're prototyping at this point
once (if) we're happy with things we can re-design a more compact board, etc ...

but for now, I'm making it hack/surgery friendly ... as simple as possible to work with
thx for checking it out

btw, the PCB card was sent yesterday AM and should be in tomorrow's mail I'm hoping

~jcm

[Eb7+9]

JC, interesting train of thought: I've wondered all along what might be simpler than Wayne's idea of using the oscillator stages for gates, to me that was the real genius of the circuit--it just reduced the complexity and parts count by such a factor. If I were responsible for the test circuit right now, I think I would cautiously build a one oscillator prototype on the bench and check out the modulated waveform vs various devices' characteristics--just to get a real good idea about how important matching and transfer characteristics are.
Of course that's kind of the old way of doing things, it's pretty handy for substituting real devices though, say from a variety of part types which you can get from suppliers.
Anyway, thanks for sharing your thought process as you go, it's interesting from a lot of angles.

you keep hitting me with this idea Dave,

every time I give it another think ... maybe, I'm missing something here
maybe I'm not understanding right what you're saying
really not sure

rambling out loud here,

the way I see things, and the patent seems to state this clearly to my reading,
the idea relies on continuous gain variation (mixing), and not multiplexing /gating ...

my SPICE simulations with processed audio seem to confirm this (... very clearly I think)

if it were simply a matter of gating then I don't see how phase-shift would get produced ...
and, you are talking about gating "in perfect square-wave fashion" aren't you ?!

if so, then at such slow speeds simply multiplexing between three out of phase signals would only produce something (if any) in the very fast transition zones ... (again, my assumption here)

... BUT, that's just how I see it ...

NOW >>>

"if" we could somehow slow down the transition edges, and make them RAMP slowly
instead of flying up/dn real fast like logic gates tend to when operating in square wave fashion ...
almost so that the end of one ramp coincides with the beginning of another ramp in the three-phase system
THEN, I could see phase-shift happening similar to how the tube circuit does it

to me the key would be to get phase ramping for each section of the 3-phase system lining up end-to-end
so that it operates almost as a continuous conveyor belt

(why I'm stoked on this OTA approach, as it shows the 3-phase modulation waves lining up this way)

but, the problem with using gates ...

let's say you have a ring of logic gates (inverters) that have properly slowed down transitions, or not ...
Q: how do you inject the AC signal in there ...?

that's the part I'm not seeing here // maybe you can post a schem
maybe you're suggesting using the gate-based ring to drive OTA's... not sure
(please explain)

but, I admit ... deriving a ring oscillator using CMOS inverters is very easy // and very common in fact
fer sure,

I see that as an idea for a stand-alone oscillator ... but we have others as well
only thing, I'm not sure what the advantage would be ...

sorry, it's just not clear to me

---

now, last night I finished characterizing my collection of 2n5457 jFET's ...
253 devices in all ... crazy

anyhoo // I did a few simulations of a jFET based oscillator (similar to the tube version) on the computer and noticed that if I attach the ground rail to a floating voltage I can vary the speed of the circuit a little by varying the voltage of the rail which would normally sit at GND ... I suspect the same thing would happen with the tube circuits as well ... of course, back in the day peeps didn't have op-amps that could go negative in voltage

just saying ... in a jFET emulation speed control might pan out using this approach
we don't need that much speed variation anyway

this could end up being a big idea here ... (!!)

when I get a chance I will try building an oscillator on the breadboard now that I have almost-dead-on matched jFET's to play with ... and check the speed control idea on that

ya never know ...

---

thx for the suggestions Dave, of course I'm willing to consider anything ...
keep the ideas coming (and draw me something if you want)

~jcm

[daveweyer]

JC, You are right on about the gating angle, I only use the term because Wayne uses the term--that's how he sees the tubes turning on to allow the constant phase shifted signals to get imposed on the output of the oscillators. It's not like what we think of nowadays as a gate, it is a sine wave--so the tubes gate the signal from the time their cutoff voltage is reached until they reach cutoff again--and the gating looks like an amplitude modulated transmitter carrier, modulated with a sine wave.
We could think about square waves, it would work with those too but it would only be a pitch adding and subtracting system, not a pulsing LFO like the existing circuit. The amplitude variations have a lot of effect on the output signal.

[Eb7+9]

thanks again for that info Dave,

btw, as mentioned above - the research into building an OTA based version of the CTP
accidentally lead me to discover a way of making THIS happen :

https://youtu.be/t99MadCuSPg

in case anyone is wondering, the two holes on top are for (later) independently setting the HP and LP cutoff frequencies using rotary switches and caps

as is, both Freq's are currently "fixed" at about 340Hz ... about 40Hz above the Fender circuit values

comparing with the stock Fender circuit

https://www.youtube.com/watch?v=-k2Y3bF8mH8


anyhoo, I have (... surprise!) the jFET version of the CTP-emulator cooking on the bench // as we speak ...

~jcm

[Xplorer]

WOW ! Beautiful !! loving this univibesque tremolo !

i guess it is the same kind as this brownface one too, right ? but your clips show it much better
https://www.youtube.com/watch?v=_qacyyzms9c

and what a beautiful sight you have, outside your window ! it must be inspiring while playing.

good luck on the next one, will you attempt to also recreate the whole feedback setup from Dave, thanks to a "chora tone pedal" ?

[Eb7+9]

not sure what you mean about Dave's feedback method Xplorer ...
please describe

for now, I'm replicating the Baldwin CTP structure only
might get some life out of it today // still awaiting inductors in the mail

~jcm

[Xplorer]

congrats again on your hard work JC, and i like this new kind of video you do now, over the soundcloud stuffs, it brings a new dimension when you also see the amp, the environment, the pedal ...

about the feedback or regeneration, i meant what Dave described, which creates these glass insect sounds like here :

https://soundcloud.com/xplorer80/dave-w ... sects-edit

here's Dave description :

The key to all that is to add regeneration. Everything I've done with all this sound experimentation has been with regeneration. It really doesn't matter how you do it, just get the outputs fed back into the inputs with some kind of control between so that you can get things right on the edge of going out of control.
I use a pathway on the Glass Insects something like this:
guitar to amp with a lot (variable) of distortion and gain, amp output signal to Ibanez analog delay, delay output to Digitek digital delay (2 sec adjustable) Digitek output to Hammond reverb springs (mixable), reverb springs to Lexicon pitch shifter, (sometimes in series, sometimes in parallel) Ampeg VT 40, (with LC filters) VT 40 out to speaker and choratone, choratone out to amp and speaker, speaker open to recording room with guitar for feedback to guitar and to Hammond reverb springs. On some of it I route the VT 40 out to a Leslie speaker on slow rotate. On some of it I use different routings, but the same basic building blocks. As Deniz plays, I control the knobs, select the signals which are fed to the next block, let feedback go or hold it back, allow one of the delays to pass and not the other, re-delay what Deniz has already played by blocking his input signal and allowing the first delayed signal to loop, then opening the signal from his guitar and let him play along with the loop, and then let that all become a loop of its own, and so on. That's how you can get those swooping sounds, and to get the guitar pluck to repeat indefinitely so you can get it to feed back all the way to screeching, which you then control with the LC filters to move the frequencies where it wants to take off.
The louder it gets, the more the Hammond springs resonate with the signal, and the more it feeds back itself into the signal. As the feedback gets to the choratone, it gates various frequencies in its rhythmic pattern, and the guitar pickups become a microphone which responds to the amplified resonances and it starts to play itself, the strings just becoming a drone which add their own resonances to the mix.
Add it all up and you have the first part of the Glass Insects. This is a simplified explanation, but the basic plan is the same. It takes two guys to play the thing, one guy to listen and react to what he is hearing, and another guy to try to control all that in a musical way, but that challenges the first guy. So it goes around in a circle, everything is going around in a circle, the feedback from man to man is intuitive, but it's there. It boils down to two guys trying to control a monster that is always on the verge of taking off in any direction on its own and that cannot be totally controlled, only herded in this or that direction until it changes its mind and either guy reacts.
It was all done live and recorded as it went. After it was done I overlaid some of it on top of itself where there were accidental places of good interaction.
The tube modulator was integral to the whole thing being resonant and unpredictable, and also creating that deep space.

[daveweyer]

First things first though. JC has done a lot of work so far and I'm envious of his energy and commitment.
It'll be a while before the regeneration discussions have any real meaning. I'll send some more clips of the CTP regenerating itself in the mean time, just to keep the enthusiasm going.