The FM Tube Modulator Jimi Loved

His guitar slung across his back, his dusty boots is his cadillac.

Moderators: VelvetGeorge, BUG

Post Reply
Tek465b
Senior Member
Posts: 207
Joined: Sat Jan 10, 2015 9:43 am
Just the numbers in order: 13492
Contact:

Re: The FM Tube Modulator Jimi Loved

Post by Tek465b » Mon Feb 15, 2016 9:34 am

Yeah its cool (h)

Actually there is 14 digital I/O so we could do 14 PWM output without the timers.
Problem is its alot of mathametical thinking and coding + loss of frequency and phase accuracy/jitter. (theres a good reason to using timers)

There is 6 pwm output and 3 timer, so the PWM output are paired for 2 PWM output per timers.
Like i said above i have to use the 16 bits timer1 for the timer interrupt(that make for very good frequency accuracy) so we are left with 4 PWM output.
Good news is that there is other atmega chip like the 2560 with 5-6 timers and 15 PWM output, that is fucking great :D.

With this one we will just need to duplicate the 3 analogwrite lines for 3 extra PWM output.
(just copy and paste).
The main problem is we use a 8 bit 16Mhz processor, 6 channel audio signal processing(depth, offset, frequency modulation) thats ressource heavy on this little thing(its far from a 32-bit ARM cortex running quadcore 1ghz lol). That the only limit. The more channel and resolution and signal processing we do, the maximum operating frequency decrease(the slow it gets).

Also i did another simulation with a better simulator and i can get up to 40hz(instead of 15)? So maby in real world we can actually do 20hz easily with 6 output :D.

User avatar
Eb7+9
Senior Member
Posts: 105
Joined: Wed Mar 09, 2011 1:30 pm
Just the numbers in order: 13492
Contact:

Re: The FM Tube Modulator Jimi Loved

Post by Eb7+9 » Mon Feb 15, 2016 8:06 pm

geez, I get dizzy just reading this stuff ...
remember, you're talking to a guy who didn't really go farther than Finite State Machines in digital class

:P

ok, I think I get the "we need timers" part ...
mind you, there are a number of patents out there (see Wayne's last one)
where they inject noise in the oscillator to produce random jitter

so, maybe not a bad option to explore ... (assuming infinite time to check every angle out)

but anyway, I'm just trying my x2 jFET-oscillator rig at the moment (using three amps)
and finding that just one channel running is better (esp the slow one ... @1.67hz)

two channels and it's already sounding too chaotic for me
just one is perfect ...

I'm sure with the band-limited TANK circuits things might be better in a multi-channel way
but I won't be trying that I'm afraid // I'll leave it to someone else to (fully) explore

so, thanks for looking into this,

it was just a passing thought ...
~jcm
modern VT circuit analysis and modeling: https://viva-analog.com/product/ifmta-book-pdf/

User avatar
Eb7+9
Senior Member
Posts: 105
Joined: Wed Mar 09, 2011 1:30 pm
Just the numbers in order: 13492
Contact:

Re: The FM Tube Modulator Jimi Loved

Post by Eb7+9 » Mon Feb 22, 2016 12:40 pm

so I finally got something boxed up
quick demo, just to give an idea what the class-A jFET version can sound like :

https://youtu.be/4pMuMS7g34I

the audio is kept raw for obvious reasons // panned hard R-L

recorded the with two amps to see what the experience would feel like
(worse case for noise, etc ...)

I'll just say it sounds way more pronounced in the recording
but still pretty cool in person ... not unlike a Univibe set on low, let's say

even with tightly matched jFET's the unit sounds a little wobbly (un-even)
which is not a bad thing in itself, adds chracter ...

the Source circuit trimmers are very sensitive here ...
even though they're 1k's set around half-way
indeed, they could be tweaked for more or less steady oscillations

I extracted signal, as before, using a Y-cable in the FX loop of a first Crate V50
then thru the CTPrev2 ... (with several volts p-p signals)
and to the front of a second Crate V50 (both modded/matched)

with matched mics (e609's) .. that's my noise test

as Dave has mentioned before
another way would be to go off the speaker

//with the advantage of LowZ for noise immunity
but possible noisy ground as well ...
which is cured with a chassis isolated jack anyway//

and then from the output of the CTP to a separate channel in recording console

in THAT scenario we would yield better noise performance
that's one big reason

AND, also to gain the ability of working with cranked out amplifiers

where the CTP chain doesn't interact dynamically within the circuit
like I'm using it here

(hey, there's gonna be lots of room for more experimenting later ... which i will leave to the rest of you to do))

since I'm not running at cranked levels here I'm eliminating the issue for now ...
and focusing on the bare bone aspects (time constraints, etc ...)

to be honest,

the only problem was a little bit of hum from the CTP's on-board 22vdc supply line
still, I had a pretty effective line filter on there ...

which means this class-A circuit is particularly sensitive to supply ripple

in the final opto-PWM version this will not be an issue
since I'll be using op-amps throughout the signal path

point taken,

---

ok, I'll say it again ... this version was difficult to build
not to mention it can take a minute or two of diddling the start switch to get the oscillators going
like rubbing sticks ... archaic to the max // kinda cool actual

and, I wasn't happy with the CMOS-opto "slowed-down" start/stop circuit performance, so I disabled it ...

so, it's hardly a stage friendly device at this point
even the (easier to build) BJT version would have these issues I suspect

so, there's no great advantage to going BJT other than having more ease at matching devices for the modulator
not sure if the modulation quality would change though ... I suspect it would be minor

still, I encourage other DIY'ers to check out that version of the circuit

---

on the brite side,

Tek is burning me a couple of "3-phase LFO" programmed pro-mini PIC boards
and as soon as I get those I will draw up a proto-board that will cradle them,
alongside all the necessary accouterments: ladder network, amplifier stages, etc

I may include the option of running the unit in mono
ie., with a dry/wet mix control ...
and output gain amp, so that the unit can be run at unity gain (as a plain ol' stomper)

we'll see if that 10~12db of insertion loss means eating too much noise later in the make-up stage ...
~jcm
modern VT circuit analysis and modeling: https://viva-analog.com/product/ifmta-book-pdf/

User avatar
Eb7+9
Senior Member
Posts: 105
Joined: Wed Mar 09, 2011 1:30 pm
Just the numbers in order: 13492
Contact:

Re: The FM Tube Modulator Jimi Loved

Post by Eb7+9 » Tue Feb 23, 2016 3:21 pm

good tutorial on the Arduino "Pro Mini" ...

https://learn.sparkfun.com/tutorials/us ... o-mini-33v
modern VT circuit analysis and modeling: https://viva-analog.com/product/ifmta-book-pdf/

Tek465b
Senior Member
Posts: 207
Joined: Sat Jan 10, 2015 9:43 am
Just the numbers in order: 13492
Contact:

Re: The FM Tube Modulator Jimi Loved

Post by Tek465b » Wed Feb 24, 2016 7:57 am

Nice clips :D.
Great tutorial.

This tutorial show for the 3.3v version, we use the 5v version.
Not a big difference but, you have to get the 5v serial/usb converter not the 3.3v. forgot to tell you that :O.

User avatar
Xplorer
Senior Member
Posts: 2470
Joined: Sun Apr 19, 2009 5:27 pm
Just the numbers in order: 7

Re: The FM Tube Modulator Jimi Loved

Post by Xplorer » Wed Feb 24, 2016 10:04 am

i'm sorry i can't participate, this is rocket science for me, and it's impressive, but speaking of rocket science, here's music from space, recorded by Appolo 10 , while in the dark side of the moon ... who knows, it's perhaps Dave's Tube modulator ;) in feedback process, it sounds quite like it anyway !!

https://www.youtube.com/watch?v=bjLZBrQ ... r_embedded

https://www.youtube.com/watch?v=_QYRVCq ... r_embedded

daveweyer
Senior Member
Posts: 713
Joined: Wed Oct 29, 2014 9:36 pm
Just the numbers in order: 13492

Re: The FM Tube Modulator Jimi Loved

Post by daveweyer » Wed Feb 24, 2016 3:31 pm

Even if you can't directly participate, it is a remarkable thread to follow. I mean here is a guitar modulation device being engineered right before your very eyes, in both analog and digital versions, and you get to watch it develop from an embryo to an actual functioning device, with all the twists and turns, false starts, and empirical learning processes that such devices are built upon.
I think this is a wonderful exercise, especially if you never knew what kind of thinking and effort went into providing you with a device capable of transforming your guitar sound.

User avatar
Xplorer
Senior Member
Posts: 2470
Joined: Sun Apr 19, 2009 5:27 pm
Just the numbers in order: 7

Re: The FM Tube Modulator Jimi Loved

Post by Xplorer » Wed Feb 24, 2016 10:23 pm

yes it is ! even if it's a heavy level, you can feel this scientific process, i love it

User avatar
Eb7+9
Senior Member
Posts: 105
Joined: Wed Mar 09, 2011 1:30 pm
Just the numbers in order: 13492
Contact:

Re: The FM Tube Modulator Jimi Loved

Post by Eb7+9 » Tue Mar 01, 2016 9:31 am

thanks for the encouraging comments gentlemen,

just a short update here ...

Tek has been supremely generous with his time,
and is now in the process of mailing me a pair of programmed Arduino Pro-Mini boards

once I get them, I will set up a test jig on the breadboard
and confirm that all is as it should be (... not that i have any doubts)

I do this just to make sure, before I draw out the board that will house the pro-mini
maybe in the future (likely in fact) that will get revised to hold a standalone 16-pin DIP chip

we'll see ... when we get there

once the digital side is vetted, I will turn my attention
to the signal path again and try implementing the variations I described in my last post

---

I was gonna go to the city to work, take a break from all this
but weather has been sh*ttier than &^%$
... and I have to camp out in my van when I go

working all day in the rain, climb into a cold soggy van to sleep
and do it all over again the next day ...

or, ...

stick around home a little longer till stormy season clears and keep tinkering with this stuff ...

hmmm ... :scratch:

~jcm
modern VT circuit analysis and modeling: https://viva-analog.com/product/ifmta-book-pdf/

Tek465b
Senior Member
Posts: 207
Joined: Sat Jan 10, 2015 9:43 am
Just the numbers in order: 13492
Contact:

Re: The FM Tube Modulator Jimi Loved

Post by Tek465b » Thu Mar 03, 2016 2:24 pm

Eb7+9 wrote:once I get them, I will set up a test jig on the breadboard
and confirm that all is as it should be (... not that i have any doubts)

I do this just to make sure, before I draw out the board that will house the pro-mini
The board should be to your mailbox this friday or monday(maby before).
Yes this is standard practice, its soo easy to forget or misplace a bracket loll. Or have the wrong variable type.
I honestly dont have any doubts about the code, but experience told me that whatever how many time you check and re-checked a code, little bug can still happen(wich usually is an easy fix). But this code is quite small and i really have no doubts about it.(You just need to avoid huge ammount of ripple on the Speed control pot signal, consideration when you design the pcb/power supply).
Every time it has a speed change, one of the 510 step get reset back to 0. I had to do that, because if the counter is at 12300 and a speed control change happen, it set the interrupt at lets say 600.. then it will need to go up to 65536 and then overflow back to 0 then back to 600 before the trigger(that mean quite a LONGGGG stop/hanging delay before the PWM get "modulated" again). So what i did is each time there is a change in speed(interrupt value) the counter get reset back to 0.
So big ripple on the speed control = counter go back to 0 each 50ms but without long hanging/delay..
(in normal/easy term: the frequency slow down just a little bit if the speed control voltage keep swinging up and down"AC/Ripple") But that should not be an issue(low ripple on a 5v pot.. is not a big challenge)

I heard weather gets better in the next few days.
It was -36 with wind factor yesterday night, tonight is going to be cold but apparently tomorrow it should be alot better.
But then.. its going to be WET.
That remind me of a few years back before i got my shop, my syster got a camaro and we did a (el cheapo) quick body / electrical / mechanical restaure job... laying on our back on ice and water.. woohoo.. lots of fun.(never ever again)

daveweyer
Senior Member
Posts: 713
Joined: Wed Oct 29, 2014 9:36 pm
Just the numbers in order: 13492

Re: The FM Tube Modulator Jimi Loved

Post by daveweyer » Thu Mar 03, 2016 5:09 pm

It's still its of fun for me to follow the thread. There's more once this thing is up and running.
Do you think the tube modulator didn't exist because nobody wrote about it in their book about Jimi?
Seems like it's getting more real every day.

Tek465b
Senior Member
Posts: 207
Joined: Sat Jan 10, 2015 9:43 am
Just the numbers in order: 13492
Contact:

Re: The FM Tube Modulator Jimi Loved

Post by Tek465b » Wed May 18, 2016 7:03 pm

Wow finally the forum is back. phew.
The code for the CTP LFO is working (and at 32khz) ;) :!:
it do produces a running time shift that just keeps on going.(frequency shift) its awesome.

Also i did some reading on the CTP patent and its operation principles.
Look like its not that hard to understand finally.

Can't wait for JC to come back from work and give you some more details. :D

User avatar
Eb7+9
Senior Member
Posts: 105
Joined: Wed Mar 09, 2011 1:30 pm
Just the numbers in order: 13492
Contact:

Re: The FM Tube Modulator Jimi Loved

Post by Eb7+9 » Mon May 23, 2016 11:25 pm

well, just a quick message to show our latest progress

Tek did a superb job figuring out the code for a 3-phase triangle wave (linear-time) LFO which includes a speed control range spanning 0.1Hz to 20Hz, a very useful OFFSET control, and of course a DEPTH control ... in the unit I built I also have a volume control in the final make-up gain block // necessary for unity gain operation since the phase shift lattice incurs about 10db of insertion loss

as a test, I piped in a 500hz signal with a 10 volt pk-pk amplitude and adjusted the volume control to produce the same at the output prior to onset of clipping // I don't need to remind anybody here that it far surpasses the performance of any BBD-based circuit in this regard ... preliminary tests on guitar (and Casio Organ patch) showed very good variable-speed pith shifting circuit, producing extremely good noise specs when running thru the FX loops of my two modded Crate amps ... by adjusting the supply rails even higher input headroom could be achieved if desired // this contrasts well with the versions I prototyped previously

below are a few pics to show what things look like ... as Tek mentioned he got the PWM signals to run above 30khz ... well high enough to prevent audible feedthrough to the output (alised or not ...) ... the thing is now operating flawlessly in every aspect ... embodying William Wayne's two patents referenced above in this thread

in case anybody is wondering, I have an LED wired to each PWM clock output and the three are displayed on the front in a triangle pattern, producing a rotating pattern as the oscillator goes through its cycles ... the visual pattern thus provides useful feedback on the effect all three control pots have on the PWM driver signals ... it's quite interesting to see in action ... FYI, these LED's draw less than 500uA of current each, and so do the LED's in the opto's // totalling less than 1mA on each (40mA max) PWM output ...

Tek's contribution and dedication to the project cannot be overstated // it's been an excessively pleasant colab, a first of the like for me ...

thanks for bringing up the forum again // cheers to Dave and all following this thread

more soon ...
~jcm
Attachments
AVRbrains.jpg
AVRbrains.jpg (95.57 KiB) Viewed 6069 times
CTP-PWMjcm2016.jpg
CTP-PWMjcm2016.jpg (400.08 KiB) Viewed 6072 times
PWM-CTPguts.jpg
PWM-CTPguts.jpg (262.3 KiB) Viewed 6072 times
Last edited by Eb7+9 on Tue May 24, 2016 9:38 pm, edited 8 times in total.
modern VT circuit analysis and modeling: https://viva-analog.com/product/ifmta-book-pdf/

User avatar
Xplorer
Senior Member
Posts: 2470
Joined: Sun Apr 19, 2009 5:27 pm
Just the numbers in order: 7

Re: The FM Tube Modulator Jimi Loved

Post by Xplorer » Tue May 24, 2016 7:43 am

:champ: :champ: :champ:
simply amazing, it looks very well built and advanced. i'm sure we'll love the sound

User avatar
Eb7+9
Senior Member
Posts: 105
Joined: Wed Mar 09, 2011 1:30 pm
Just the numbers in order: 13492
Contact:

Re: The FM Tube Modulator Jimi Loved

Post by Eb7+9 » Tue May 31, 2016 3:08 pm

just a reminder or two ...

the circuit shown above does not produce the complete emulation of theoriginal Wayne/Baldwin CTP unit that Dave introduced to Jimi, even thought i've labelled the PCB simply with CTP ... what I showed above is still only a single channel (wide-band) version of the full CTP invention ... what anyone would call a plain and simple Vibrato (pitch wobbling) circuit // William Wayne referred to his idea as a "Frequency Changer" ...

of notably interest, this latest build can process pretty large signals before incurring clipping distortion (set to the internal voltage limits of the signal path amplifiers) ... and so, for example, allowing really high S/N performance to be achieved when used in an old-school tube amp FX loop ...

so far I've focused mainly on bettering the performance of the modulator channel itself, before moving on the the full 5-channel CTP monster ... Tek465b's idea of using PWM signals, and then me accidentally discovering Wayne's other (1970) patent is what lead to where we are currently ...

again, the circuit shown above is the single-channel embodiment of the mechanism listed in William Wayne's 1970 patent, by using photo elements (in my case, Silonex NSL32-SR3 optocouplers) to accomplish the three-way mixing of the constantly dis-phased Band-Pass versions of the input (instrument) spectra // a mouth-full ... as opposed to using some kind of active gain modulator as I tried duplicating using jFET's and as in Wayne's 1956 Patent and the actual Wayne/Baldwin CTP vacuum-tube based oscillator modules ... my jFET circuit required very tightly matched jFET's but was still fixed-speed at the end of the day ...

in this sense we've come a long way in our quest for variable speed, adapting modern thinking to the basic purpose of the original invention, and an easier implementation of the mixing inspired by the 1970 patent ... yielding an number of advantageous specs in the process

as I said above, I may well be exploring a tweak to this version of the circuit so that it produces mono-Phasing as well ... like the Univibe, I strongly suspect that the circuit has the potential of having separate Chorus/Vibrato modes (something not included in the original invention). we'll see shortly ...

of course, I will be exploring ways of embodying a full (4 or 5 channel) Band-Limited emulation of the original Baldwin/Wayne CTP circuit // only, this time with the ability to alter alter the frequency and intensity of each band-limited channel independently // pretty much a necessity all along ... doing full justice to Wayne's original "asynchronizing" concept ... again, thanks to Dave for explaining this aspect of the invention

I may also be exploring another alternative to using matched opto-couplers, seeing that we have PWM driver signals to play with // that is, by using analogue gates (quad bilateral switch 4016) instead of opto-couplers ... thus, doing away with matching "expensive" components altogether // ... again, the 32khz clock speeds will make all this possible

one step at a time, eh?!

btw, going back to earlier discussions we had on how the circuit operates, and having the opportunity to play with the "average" idling position of the PWM pulses through the OFFSET control shows that setting the PWM average to 0% gives the strongest picth-bending effect ... this produces more of a "gating" function in terms of cell aperture, as opposed to increasing the OFFSET above zero and giving the cells more of a modulated behavior ... in the process, echoing what Dave said about the "gating" function of the original Modulator circuit ... I can say that my concerns are now dispelled as I've had the chance to compare both modes of operation with the use of that control // thx to Tek465b for allowing the code to do this ... consequently, the OFFSET control can be mounted as an internal trimmer with little loss of function, though also handy as an external control as it can help subtle-fy the effect if needed ... the option is there

another conclusion to my testing shows that in its basic form the circuit can provide plenty of pitch shifting, maybe more than most people would want in typical situations ... part of my testing showed that circuit acting on an Organ patch (and my old B2) it goes way beyond what a Hammond Scanner can do ... and does so in a pleasant way

as a technical aside, the headroom limitation of the circuit is chiefly determined by the rail voltages feeding the op-amps used in the "lattice" part of the circuit, and also in the recovery stage ... in essence we can give this circuit as much headroom as the op-amps are designed to handle and put out ... which makes it ideal for placement at the output of a preamp circuit, and thus bringing vibrato/pitch-shifting into super low-noise (professional grade) performance ... ie., doesn't have to be placed before the preamp, as many pedals need to be ... it can also handle all kinds of strong sources that BBD-based picth-shifter typically cannot handle without incurring max 1.8v pk-pk headroom distortion // sources like active Bass and keybords are not a problem with this circuit (at least for those who want to go direct) ...

an interesting consequence to all all this S/N stuff ... made me wonder about the general designing FX circuits in a post-preamp environment // challenging the current "between guitar and amp" paradigm for typical FX pedal usage so-to-speak ... similar to the way the Baldwin CTP was meant to be used originally (ie., from a high-level speaker voltage) ... a little pondering shows that many pedals (even overdrivers and certain WAH architectures, etc ...) can be re-designed to work in the (unbuffered/natural) FX loop of a guitar amp ... here I'm not talking about the modern scaled-down-buffered-recovery style (made of low signal pedals) of FX loop // but rather a simple break in the circuit between preamp and output stage (in classic designs), as I have it in my modded Crate amps ... done chiefly with the purpose of bringing background noise down (drastically in some cases), and getting the overall signal path to operate at its best // especially in the case where multiple pedals, especially noise-prone ones, are being used ...

another thing I didn't mention is that the current version of the circuit has NO signal capacitors throughout // in fact the whole works is DC-coupled from beginning to end ... which means, supreme fidelity in one sense (ie., no cap is better than the best cap) and also absence of phase distortion in the low end, and yielding a full natural low-end spectrum ... all made possible by the fact that the circuit is low-gain in nature // measurements confirm little to no DC offset at the output, as a result the bypass switch produces no pop when toggling ... this arrangement thus allows a full DC-to-200khz bandwidth outside of what the Lattice circuit is doing, allowing us to focus in on the sound quality of the "main" idea and not the peripherals of the embodiment

~jcm
modern VT circuit analysis and modeling: https://viva-analog.com/product/ifmta-book-pdf/

Post Reply