The FM Tube Modulator Jimi Loved

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

Moderators: VelvetGeorge, BUG

Post Reply
User avatar
Xplorer
Senior Member
Posts: 2431
Joined: Sun Apr 19, 2009 5:27 pm
Just the numbers in order: 7

Re: The FM Tube Modulator Jimi Loved

Post by Xplorer » Fri Nov 20, 2015 8:45 pm

Eb7+9 wrote:Wicked ...!!

:o

thanks for preparing and posting the clips Dave & Adrien,
I can say I feel like I'm in the same ballpark at this point ...


... and, the addition of a Master SPEED control might just be the cherry on the cake
... plus, the stereo option ... firecracker

ok, ... back to the Current-Controlled 3-phase Sine-wave LFO design
the end is finally in sight ...

~jcm

what else could one ask for ? speed controler, ( depth also i guess, already included ) , and stereo option ...
Just perfect. good luck !

you'll maybe also like this one, a more classic chorus i'll still build, the onboard Juno 60 synthesizer chorus, applied to the guitar it's really great. but very different from the natural sounding chora tone . coincidence, this chorus original only has two speeds too, but some diy guys did a speed control for a pedal version.

https://soundcloud.com/gligli/juno-60-c ... econd-demo

http://gliglisynth.blogspot.fr/2014/05/ ... no-60.html

Eb7+9
Senior Member
Posts: 104
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 » Fri Nov 20, 2015 11:54 pm

that's pretty cool for synth work ...
can you email me a schematic of that thing pls ?!

what I like about the Choratone is there are no BBD's here ...
and so, should expect way higher headrooms, firstly ...

higher than all CHORUS boxes these days, since they're using what's available for bucket brigade IC's
(ie., nothing like the SAD1024 and SAD4096 of yesteryear)
even then ...they are noisy

I think we can get way better noise specs with the Chora-Tone architecture

---

btw, here's the paper that I'm studying

http://people.biology.ucsd.edu/gert/pub ... 08_osc.pdf

in particular I'm trying to figure out what we can do with the circuit module of Figure 3
it isn't clear to me how you would chain these together in a ring "as is"

... so far, I can get it to work using sandwitched level-shifters
but the waveshape is not sine-like enough

... I need to characterize the module a bit better

... and, maybe 3 diodes is not the way to go here

interesting problem ...

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

Eb7+9
Senior Member
Posts: 104
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 » Sat Nov 21, 2015 12:44 am

don't know why this didn't jump out at me when I went thru the reference list

https://docs.google.com/viewer?url=pate ... 489878.pdf
modern VT circuit analysis and modeling: https://viva-analog.com/product/ifmta-book-pdf/

daveweyer
Senior Member
Posts: 603
Joined: Wed Oct 29, 2014 9:36 pm
Just the numbers in order: 13492

Re: The FM Tube Modulator Jimi Loved

Post by daveweyer » Sat Nov 21, 2015 12:26 pm

If these things will work at very low frequencies, and you just want independent sine waves to operate gain blocks, then it appears you may have found your variable frequency solution. Fascinating patent.
Have you figured out how to arrange these in order to get 120 deg lagging phases?

Eb7+9
Senior Member
Posts: 104
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 Nov 23, 2015 1:32 am

glad to be able to write this post, hope y'all enjoying following this project development ...

to recap,

one thing that mystified me about Gilbert's Translinear circuit of Fig. 3 (Loizos) was how the ring is coupled together later ... so, I contacted one of the authors and he recommended using capacitive level shifting after I told him about the approach I was taking ... anyhow, caps wouldn't work at our low speeds // unless we use stupidly large caps ... so, scratch that ...

besides, Gilbert's paper clearly shows a loop of those two Translinear blocks DC coupled to each other
I figured, that's the only way that ring could work for us (at super low speeds)

the author also suggested NOT using that module, period ...
even tho the text makes it sound like a viable engine to work with // ... that's how I read it anyway
I found this all a bit confusing

then, I noticed Gilbert's patent in the ref section
which encouraged me to SPICE the "bad-boy" module on its own a bit ...

http://www.lynx.net/~jc/translinearGm-R ... n100nA.jpg

using TINA-TI component models, we get a differential small-signal voltage gain of about 5v/v at DC ...
(the paper claims 3v/v, guessing because they might have used diode connected transistors in their case)

to compare, the op-amp version of the oscillator (shown way above) has a gain of x2 per stage ...
and no loading between stages ...

in case this is not clear to some, the circuit, and therefore the whole oscillator, operates in a differential manner, as opposed to single ended referenced to GND ... voltages on one side go up while their counterpart on the opposite leg go down, and vice versa for currents ... it turns out that the place where the level shifter takes its signal can also be the place where the output current converter can take its signal from

Also, notice that we 're working with device models for common "hobby-DIY" parts, the ones I'll be using myself ...

1n4148 Si diodes, 2n3906 PNP and 2n3904 NPN Si transistors

the idea is to approximate as closely as possible the same circuit we will be bread-boarding and building later ...
for those interested, it then gives us a chance to compare simulation results to their real-world equivalent
...
to see how full-of-sh*t the simulator was all along, or not ... but I digress ...

now, if the bias current is stepped from 1uA to 120uA, voltage gain is maintained (except for frequency response) :

http://www.lynx.net/~jc/translinearGm-Rstage1-120uA.jpg

it confirms the theory presented in the paper ... namely that the differential voltage gain
does not vary with bias current ... ie., Gm changes inversely to diodes resistances, and at the same rate

to understand how this is relevant to our oscillator goals here ...
when we stick a capacitor across there you get a RC time constant that varies directly with current
since the voltage gain stays the same, the loop gain will not vary but the internal time constants will ...
we get a variable frequency oscillator

I found that the gain stage preserves this constant-gain property down to about 20nA ... which is pretty darn good
seen here with bias current swept from 10nA to 120nA (ie., 0.01uA to 0.12uA) :

http://www.lynx.net/~jc/translinearGm-R ... -120nA.jpg

this gives us an idea of the usable bias range (minima)

now, if we tie these guys together in a loop ... Bases to Collectors ... then, for one, a stable DC
operating point cannot be achieved ... and this makes the circuit quite a bit more difficult to analyze IMO

and, for the life of me could not figure out if that was part of the idea
so, to answer that I inserted a differential level-shifting stage ... one with gain built in
... this effectively removes the ambiguity //
and, possibly what the authors meant by THIS circuit requiring high supply headroom

I sim'ed many combinations, some with 2 or 1 diodes ... but went back to the three diode case


so, the basic Gilbert Translinear Gm-R module gets augmented to this :

http://www.lynx.net/~jc/translinearGm-R ... 015JCM.jpg

the horizontal arrows show the In/Out flow of differential voltages in the ring
while the down arrows show the current outputs that will be used to drive the bias input of OTA's ...

three of these are tied together in a (DC coupled) ring according to the diagram in Fig. 1 of Loizos' paper
and looks like this :

http://www.lynx.net/~jc/CC-3phase-Osc-2015JCM.jpg

there are now three currents (IS1, IS3, and IS5) that control the oscillation RATE

these can be operated from a single MASTER current echo'd thru matched or scaled current mirrors

aside from it's overall straightforward design, it does require at least 9 matched device pairs of NPN and PNP devices
which is not such a difficult thing to do with BJT's ...

http://users.ece.gatech.edu/mleach/lowtim/part2.html

in all the total number of transistors making up a complete oscillator block (incl. sources) would be 27

the bias current in the shifter stages is set arbitrarily to 1mA
and the same for the Output current converters (OTA's blow at 2mA, so anything below that would be ok)...
choosing 1mA makes it easy to read the data afterwards, and help see if the diff-pairs are being swung wide enough, etc ...

we make note of the order of positions in the circuit for the output currents ...
and here are the final current output wave-forms:

http://www.lynx.net/~jc/CC-3phase-OscOu ... 015JCM.jpg

I think the 120 degree difference between waves is easy to see here (answering your question Dave ...)
and, this aplies to both in-phase and anti-phase groups // making synchronous stereo'izing of the effect possible
(something Wayne himself might have dug ...!) ...notice also the full 0 to 1mA output current range being spanned

I'll admit, part of my problem in all this was assuming that the Translinear circuit was going to be responsible for giving me a sine-shape to the wave-forms ... a realize my foolishness now, and that in fact, it makes perfect sense to see the wave-forms looking like they do ... possibly as they look in your Chora-Tone unit

I'm very pleased with the solution discovered ... !!
27 transistors and 18 diodes per oscillator is as good to hope for in a discrete analog solution exhibiting this kind of control

I can even see this circuit having wider application elsewhere, maybe in an e-bike ...(*!*)

... more on the circuit's performance shortly
~jcm
modern VT circuit analysis and modeling: https://viva-analog.com/product/ifmta-book-pdf/

daveweyer
Senior Member
Posts: 603
Joined: Wed Oct 29, 2014 9:36 pm
Just the numbers in order: 13492

Re: The FM Tube Modulator Jimi Loved

Post by daveweyer » Mon Nov 23, 2015 1:57 pm

Somehow I don't see that many replies coming to your post, maybe if this were a geek forum.............
I guess that leaves you and me, at least I find it fascinating to watch the evolution in thought.
Thanks for keeping us in the loop.

User avatar
C J H
Senior Member
Posts: 183
Joined: Thu Apr 14, 2005 7:43 am
Location: Sweden

Re: The FM Tube Modulator Jimi Loved

Post by C J H » Mon Nov 23, 2015 3:18 pm

I dig what's going on here.. can't contribute more than that since the discussion is about above my paygrade, hehe.. but I reckon the outcome will be pretty cool!

Keep up the good work!
"blah blah, wof wof!"

Eb7+9
Senior Member
Posts: 104
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 Nov 23, 2015 6:45 pm

thx for the good words

// just a quick update here ...

today I'm taking a break from this project and wrapping up my second Cascaded Skizlai-Pair mic preamp ...
tomorrow I head to the mountains for some end-of-season Chanterelle picking

when I get back I'll grade a bunch of 2n3904's and 2n3906's ...
and hopefully breadboard a working oscillator before the end of the week

---

btw, here are those results I promised

notice, that I'm displaying the Base-to-ground voltages, and only one set of differential output currents ...
note also, the 3 shifter stages and single output current interface stage remain biased at 1mA throughout

---

in the first case I kept the 3 capacitor values at 100uF and ran the 3 corresponding bias currents at the following values:
  • 2uA >> 0.14Hz
    20uA >> 0.23Hz
    200uA >> 1.3Hz
    2mA >> 12Hz
    20mA >> 122Hz
http://www.lynx.net/~jc/example100uF0.14Hz2uA.jpg
http://www.lynx.net/~jc/example100uF0.23Hz20uA.jpg
http://www.lynx.net/~jc/example100uF1.3Hz200uA.jpg
http://www.lynx.net/~jc/example100uF12Hz2mA.jpg
http://www.lynx.net/~jc/example100uF120Hz20mA.jpg

this technique requires knowing with some accuracy "all" cap values

---

in the second case I kept the bias currents at 150uA (set to 1Hz), and swapped out "common" capacitor values ...
  • 100uF >> 1Hz
    47uF >> 2.2Hz
    22uF >> 4.5Hz
    10uF >> 10Hz
    4.7uF >> 22Hz
http://www.lynx.net/~jc/bias150uA100uF1Hz.jpg
http://www.lynx.net/~jc/bias150uA47uF2.2Hz.jpg
http://www.lynx.net/~jc/bias150uA22uF4.5Hz.jpg
http://www.lynx.net/~jc/bias150uA10uF10Hz.jpg
http://www.lynx.net/~jc/bias150uA4.7uF22Hz.jpg

150uA is not an overly small control current level for us to manage

say, if we assume +9/-9 supply voltages and the drop-out voltage of one current mirror diode
we get 150uA = (18 -0.7)/R ... so R = 17.3/150u = 118k Ohm
a 250k pot would work here (with plenty of room around 118k to play with ...)

so, the 150uA choice appears fine if we are aiming for a 1Hz "center" LFO frequency
(in conjunction with the common 100uF cap value)

we can use this group of "roughly known" capacitor values, and then trim the individual current mirrors accordingly
that is, to bring the numbers into the ideal alignment mentioned by Wayne in his patent

namely: 1Hz, 2Hz, 4Hz, 8Hz, and 16Hz ...

and then, these are still all scale-able "in proportion" via the Master bias current
thanks to the linearity of current mirrors (... a big benefit of current-mode control)

so, I would say this second set of simulations points to a final solution
this I will aim to confirm on the breadboard later this week ...

~jcm
Last edited by Eb7+9 on Tue Jan 05, 2016 7:51 am, edited 1 time in total.
modern VT circuit analysis and modeling: https://viva-analog.com/product/ifmta-book-pdf/

User avatar
Xplorer
Senior Member
Posts: 2431
Joined: Sun Apr 19, 2009 5:27 pm
Just the numbers in order: 7

Re: The FM Tube Modulator Jimi Loved

Post by Xplorer » Mon Nov 23, 2015 8:50 pm

have a nice chanterelles hunt ! ;:) and cèpes, girolles, morilles if you're Lucky :toast:

like most peoples except for tek, i can't help much except bringing enthusiasm. this quest seems to have kept you busy a lot, you didn't expect that didn't you ? ;) he he
very interesting researchs !

User avatar
C J H
Senior Member
Posts: 183
Joined: Thu Apr 14, 2005 7:43 am
Location: Sweden

Re: The FM Tube Modulator Jimi Loved

Post by C J H » Tue Nov 24, 2015 6:17 am

SORRY FOR GOING A BIT OFF TOPIC!

.. On a side-note, JC M, I've read elsewhere that you are somewhat of a vibe-guru?

I'm underway with building a Vibe (waiting for parts atm).. Pretty much straight off the old schems but have parts to allow for a more adjustable unit. Just one Q: what photocells, or what variables in photocells do you recommend for a vintage-sounding unit?

Good luck with the chantarelles! The rain has taken the last ones up here..
Jake
"blah blah, wof wof!"

User avatar
Xplorer
Senior Member
Posts: 2431
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 Nov 24, 2015 8:40 am

before shinei there was the univibe made by Honey, same enclosure, smaller logo, and a few different values. some say it sounds better. that's what i'm cloning. that's likely what Jimi played. two rubber bands on the pedal.

in any case or versions, for the cells, you'll need some cells that have a very low ohms to very high ohms range, at a fast rate.
i think i'll use some VT33N2, they're said to work nicely.

other than JC take on the univibe, and his great infos and researchs on his forum, you'll find a cool thread here. This one is about doing a pure replica. you should read it entirely. there's also a potential pretty beat up univibe said to have belong to Jimi, with gut shots of it.

http://www.freestompboxes.org/viewtopic ... 8270c1816f

the voltage is a very important thing too. you can see a report of various voltage points in this thread, from an original vibe.
not necessarily better to bring more voltage apparently, but just the right voltages. about 16 - 17 volts coming from the transformer, as a start.

this said, JC will certainly have the best infos about how a univibe works, and what should should be done to tweak it perfectly. the offset bias mod seems to be a must !

Eb7+9
Senior Member
Posts: 104
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 » Thu Nov 26, 2015 9:36 pm

C J H wrote:
Q: what photocells, or what variables in photocells do you recommend for a vintage-sounding unit?

Jake
I still have a large stash of vintage cells in my stock, if I ever have to build a bulb-based effector I'll use those // otherwise I (now) tend to go with an interface circuit that makes use of opto-isolators (either Silonex NSL-32's or NSL32-SR3's) ... I think the "bootique" cells that Smallbear have been reported to have good range (down to 1k at least), but have never used them myself ... I recommend getting a few (lot of at least 10) and selecting four closest to each other according to simple manual testing using desk light and multi-meter ... my old (original) website has all the info I think // ...

http://www.lynx.bc.ca/~jc/pedalsUnivibe.html

so you know, in my version of the Vibe I replace the preamp by a high-headroom mu-stage with volume and Presence control built-in // some feel it makes the Vibe sound and respond way better, can make it sound more open (less dark) if required ... I also increase the bias feed resistor to bring the min-speed down lower to about 1/3Hz ... all these mods have become well known by now

if you're interested in cloning the vibe there's one chap you should know about (n case you don't already) who asked if he could use my work as a starting point for a definitive treatise on the Univibe // his name is Brad Burt and operates under the name Classic Amplification ... he's the author of the "Forum Vibe", a design essay that he composed w some consultation on my part (it's based on my original work) // and was meant as a definitive work on the vibe ... I would ask him about cells

~jcm
Last edited by Eb7+9 on Tue Jan 05, 2016 7:28 am, edited 3 times in total.
modern VT circuit analysis and modeling: https://viva-analog.com/product/ifmta-book-pdf/

User avatar
Xplorer
Senior Member
Posts: 2431
Joined: Sun Apr 19, 2009 5:27 pm
Just the numbers in order: 7

Re: The FM Tube Modulator Jimi Loved

Post by Xplorer » Thu Nov 26, 2015 10:24 pm

sorry to interrupt the univibe discussion. Jake, good idea, we could perhaps start a univibe thread and copy paste what we already wrote here ? what do you guys think ?

so : some clip from Dave ! very cool !! if i may, i'm not sure to understand the relation between the electro magnet and the organ pipe ... can you elaborate on this please ? thanks.

https://soundcloud.com/xplorer80/ts-organ-thru-guitar

this tube modulator and feedback experiments .... i heard something that feels a bit from the same universe recently, in Apocalypse now ( not randy hansen playing it, but carmine Coppola. Randy does some cool feedbacks with his guitar in some other portions of the movie ) , very nice soundtrack.

https://www.youtube.com/watch?v=FfF3PfkdkGY

here's the description from dave :

Recording of Riverside Church pipe organ driving an electromagnet placed over open tuned guitar, right near the nut. Guitar processed through Hammond long springs, the tube modulator, and a digital delay, the stereo output signals also running through another modulator. We tried to adjust the magnet to get the guitar to resonate, but not too much—we got a little too close a couple of times. You can just hear the higher harmonics of the organ in a couple of spots—if the strings aren’t resonating too hard they will let a few higher harmonics come through.

daveweyer
Senior Member
Posts: 603
Joined: Wed Oct 29, 2014 9:36 pm
Just the numbers in order: 13492

Re: The FM Tube Modulator Jimi Loved

Post by daveweyer » Fri Nov 27, 2015 12:45 pm

Xplorer, the organ pipes were just recorded onto a CD, and we used that CD to drive the amp which in turn drove the electromagnet. The organ piece by Joseph Jongen had a long section of full organ in the same tuning as the open guitar strings, we thought that would get them going easily, and it did.
The point of the experiment was based on the expectation that something like a steel string would allow itself to be modulated by a signal of a different type--and in doing so the harmonic structure of one could be imposed on the other for a strange new sound.
It was a rather primitive experiment, but I was always amazed that you could hear the faint sound of organ pipes coming from the strings, especially in places where they were not resonating at their fundamental frequency with a lot of excursion.
It all goes to show that there is a world of sound yet to be explored right inside the good old electric guitar.
If anyone is interested I have more modulating experiments on tape.

Tek465b
Senior Member
Posts: 200
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 » Fri Nov 27, 2015 4:27 pm

thats is simply amazing :!
i dont post alot, alot of reading and i just try to keep my drooling at a minimum while doing so :)

Post Reply