Triode Emulator

[tca]

it helps to look at the right set of transfer curves ... have a look here:

http://www.lynx.bc.ca/~jc/transferCurvature-TubeSimulation.html

I will read it most carefully. I've started to make the calculations to look for the possibility of getting a cascode version of the Fetzer Valve circuit. My idea is to put a constant current source at the drain of the Fetzer circuit, most like the mu-amplifier, but maintaining the 3/2 exponent.

Thanks.

[tca]

it helps to look at the right set of transfer curves ... have a look here:

http://www.lynx.bc.ca/~jc/transferCurvature-TubeSimulation.html

it's not about gain but more about gain curve ...

Hi Eb7+9, I do like your schematic!



I don't know how you got |VP|/2 (beautiful insight, though). I've redone my calculations for your case and obtained the value of  |VP|/5 for that voltage for an exponent of 3/2. Your circuit is very inspiring: "it's not about gain but more about gain curve ..."

Actually this voltage is given by |VP|*(2-n)/(4-n) where n is the exponent. If n=0 I get your result as an extreme case. Neat!!!

In the case of the Fetzer valve circuit for Vin=0 there is not a solution for k (can anyone check this) and, obviously, the value k=0.84 plus/minus 50% is a good starting value for this parameter in this case for that particular circuit. The bias voltage for k=0.84 is 0.08*VP, almost zero for all practical implementations.

[Davelectro]

I've started to make the calculations to look for the possibility of getting a cascode version of the Fetzer Valve circuit. My idea is to put a constant current source at the drain of the Fetzer circuit, most like the mu-amplifier, but maintaining the 3/2 exponent.

That would be awesome. I hate the low voltage restriction while working with Fetzer stages.

[tca]

In the case of the Fetzer valve circuit for Vin=0 there is not a solution for k (can anyone check this)...

Just want to correct this statement that I've made earlier (which is wrong)  . For the Fetzer valve circuit, which has Vin=0 when no signal is applied, the value of the bias voltage is z=(1+k)^(-3/2), (which is not the case in Dimitri work, z=1/2), the exact value for the constant k in this case is the solution of the equation

(1/(1+k)^(3/2)-(-1+sqrt(1+4*k))/(4*k^2))=0

which is k=0.78644.

[Eb7+9]

I don't know how you got |VP|/2

well, simply put // when there is no degeneration resistance (ie., Source circuit feedback via resistor) then the total usable input voltage range is "obviously" Vp to 0 ... and so choosing a bias voltage of -Vp/2 (same as |Vp|/2) simply translates the transfer curve so that an equal portion lies on each side of the y-axis (ie., evens out the x-axis range) ... in "this" case we maximize headroom, and that's all // in theory, that's the only purpose one would choose that applied voltage ... but in practice we can set it to whatever we like, which is really just another approach to generalized "biasing" // ie., setting the quiescent current (the first and foremost purpose of biasing) ... in the practical case there is a benefit to biasing for a lower current (one is noise) ... in another, it shortens the positive envelope range and lengthens the negative (that's the part I'm chiefly interested in here) ... since the positive side leads to a clipped waveform nothing is lost beyond the clipping limit, while on the negative side we can give the tail an extended range // this becomes important once we feed that signal to the front of a tube gain stage or another jFET gain stage as it determines the range of dynamic compression produced as a result ...

at least one forum member figured this out last year after I posted, where he applied the idea to produce a fully SS amplifier that mimic'd dynamic tube-like compression // I didn't specify exactly how to do this 'cuz I wanted to give folks room to figure out their own ways of making use of the concept ... the other cool thing about all this is that the exponential input diode-like loading function of Triodes isn't required in order for this dynamic mechanism to take place // in fact, I found that Plate and Grid curvature both "add" in the same direction to produce the compression effects // but, more so, the Plate curvature tends to dominate in this picture (!!) ... a long-standing assumption about what's required for full emulation is that both be present // indeed, peeps were for a long time missing the "what" part in the emulation picture ... ignoring the fact that not all Triodes are equal to begin with, and then the idea of emulating "a Triode" was never properly defined to begin with ...

the key to understanding this is to undo the thinking we were taught in school and textbooks which says the large-signal gain curve "should be as linear as possible"// in fact, one needs to go quite the other way here to get it ... to be honest this biz of applying a voltage directly to the Source/Cathode terminal is not a completely new idea in itself, since tube hi-fi "extremists" were sometimes known for biasing preamp tube Cathodes directly with batteries back in the day (mainly to get lower noise specs than provided by bias resistors) ... tho, what is new in my case, is the idea of applying that biasing approach to a jFET device for the purpose of offering a greater order of curvature (2 instead of 3/2) to play with // going beyond what Danyuk is doing in his paper ... indeed, this biz about exploiting transfer curvature for wave-bending purposes is challenging a long standing paradigm in signal processing and thus further requires a conceptual shift in order to appreciate what this can do for us // indeed, in isolation it seems to really not do much except produce a deformed wave ... it's what happens to the next gain stage "as a result of this action" that is key here, once THAT deformed signal traverses a signal cap to the input of this next stage (and obviously not a linear stage either) ... this is something I discovered in '94 when I started doing my polynomial based modeling work on Triodes, which is fully described in my tube amp book IFMTA ... (see Glass Audio 2/98 for an introductory write-up)

http://lynx.net/~jc/Maillet-GlassAudio98.PDF

I'll say it again, getting hung up on the 3/2 power thing is not very useful here once we understand how the negative "lobe" elongation can be used to shift the DC (baseline drift) average of a signal (the first term in a Fourier series) to progressively turn off a successive gain stage // not only that, but pretty much all tube stages do NOT operate in the 3/2 power mode anyway since they typically include degeneration resistance in the Cathode circuit, something I think Danyuk missed in his argument ... in other words, peeps tend to get hung up on the wrong part of the emulation thing as a result of overlooking this greater aspect ... it's not so much the power function as much as the "concave" nature of the transfer that we "want" to make use of once we know how things work in tandem // the curvature order only gives us more curvature per amount of input signal, that's all // and even then that doesn't really matter ... we "primarily" seek to use a device that has similar transfer concavity in the SS realm, that's all really ... as I mentioned before, Danyuk came very close to figuring the whole thing out but unfortunately his approach also doesn't work well in the context of cascading gain stages ... he only looks at things from a stand-alone point of view

when I so decide, I will reveal my MOJO Booster circuit and fully explain what this mystery MOJO business is all about and how exactly this circuit can be used to duplicate and manipulate the dynamically degenerative mechanism of cascaded common-Source Triode circuits // it will also suggest how the nay-sayers of MOJO were no more those who copied/applied text-theory by rote and never gave much thought to how transfer curvature can affect/determine dynamic-dependent Transient responses ... I guarantee some will flip out once they realize what can be done with this idea // in particular, one single-stage circuit in front of a good tube amp is enough to send things into an exciting dynamic synergy - which I've done at jams by using other player's rigs (axe and amp) and inserting this pedal directly in front of the amp ... it's the first of in a two-part series of interactive SS/tube-amp synergistic circuits

[tca]

http://lynx.net/~jc/Maillet-GlassAudio98.PDF

Thanks for that. Just start reading it.

[tca]

Eb7+9, I'm starting to understand what you are saying, the value of k, and hence the value of the exponent, is not really important, as long as it is high enough.
The next plots show the values of the normalized input voltage (biased at 0.5), the non-constant part of the drain current and the FFT of the output signal for various values of k.






The harmonics  2th, 3rd, etc start to appear at roughly k=.67 (another "magic" value) and the relative percentage remains almost constant for higher values of k (this partially justifies the comment made earlier by stm: "...also liked the case where Rs = 1.0 * |Vp|/Idss, as it sounds fuller than the original guitar signal.").

The plot for k=.01 does not contain 2th harmonics because the input voltage is not high enough.

Note:
This plots were not made with a simulator.

[BubbaFet]

 JC (Eb7+9), I sure do wish you great financial reward for your discovery/realization.
However, I fear that you are revealing too much without any means for adequate recognition/compensation.
I guess you might have to settle for a tribute page on Wikipedia as the 'dude who finally killed the tube amp'.
You deserve better. I say this as the guy who has built a SS amp based on your technology.

[tca]

'dude who finally killed the tube amp'.

I guess that is an over statement.

As the work of Eb7+9 shows, the only way to understand these things is to work out the math, e.g., http://lynx.net/~jc/Maillet-GlassAudio98.PDF

For adequate recognition/compensation only time will tell, but I do think that the best way to get all those things is to share knowledge.

Cheers.

[tca]

...a fully SS amplifier that mimic'd dynamic tube-like compression...

Eb7+9, can you pin point this entry? Thanks.

[Eb7+9]

I can't recall where BubbaFet mentioned finding his working example, plus the exact details weren't revealed if I recall // doesn't matter, this is an open ended problem that should be explored individually - as a result more circuits ideas will crop up ... I can give working hints when I get a chance // for now it's sufficient to say that we can produce the desired result by simply sticking a working circuit in front of a Common-Cathode gain stage (front-end of a tube amp) or any Common-Source jFET gain stage based circuit ... simple enough !

as far as making dough with this idea, turns out this is not my lifetime for making that happen it seems // either I get a straight job and secure some comfort, or gamble my health in exchange for my rewards of discovery - can't have it both ways unless some form of patronage gets set up (paypal?!) ... I was destitute as a math student in university, pacing for weeks to solve a single problem, year after year // same thing with last 20 years playing electronic fantasy // honestly, I much prefer living on this side of the fence, I feel fortunate despite the many hardships along the way ... certainly, I have barely paid rent for years building my Bass preamps and other inventions but let's face it this old jet plane has had its belly rubbed real hard on the ground many times and I'm a little banged up by now (LOL) // sometimes it's not fun, but then again my sense of discovery seems to come out more when I'm not doing too well in the comfort dept. // that's just how it is for me ... I'm into the hunt for what it is, sheer thrill of discovery, and the pleasure of sharing ... but I should say, this Triode transfer emulation stuff turned out to be more than just fancyful academia in action // this is an idea that has a very fun practical side if you're a guitar player ... you'll see what it can do once you start playing with it // myself, I prefer not to look at this as tube amp replacement therapy but rather enhancement type stuff ...

if anyone feels like they could setup a donation plan for me to continue surviving that would be nice, but me and money don't seem to jive in that sense (my marketing skills simply suck too much), plus the way things are going I think I'd be waiting a long time to see any return // it's too dog eat dog anyway once you try getting into all that) ... life is finite and I have no expectation to make money from my discoveries (some of which has been many times plagiarized over and over by one of the stars on this forum) ... ultimately, that's not so important to me, I'm not here for that reason ... one thing I have thought of doing is releasing the MOJO Booster idea in the form of a kit // I'm trying to get that done before I get evicted next month ... we'll see if I can pull it off

without trying to sound too immodest I'll just say this, plugging the MOJO Booster in front of a tube amp leads to a pretty neat synergistic effect, and that is real fun for me // the idea of emulation leading to an additive dynamic push of sorts ... indeed, playing with dynamic compression goes beyond the typical block-mode compressor circuits we all know // not that they (character type circuits) are BAD or anything, for sure they have their place - but this is a little on the newish side ... for one, there is no "trailing" effect via the diode-and-capacitor charge/relaxation combo commonly seen in these types of circuits, as a result it's very natural feeling ... I had one guy race back in the club from his cigarette break after giving me his rig to play a tune on wondering "wtf" I did to get that response ... those little situations make it kinda fun, it's not just a bunch of pretentious blather but real-working stuff ... sonic rocket fuel in a little box ...

tca, I commend and thank you for your (real time) plots // they suggest bone fide class-A Triode circuit action ... I bought a Nano DSO a few months ago and just realized I shoulda got the dual-trace version // with that I plan on demonstrating the idea in action in terms of a time-varying transfer profile in a somewhat similar way to how these guys did their testing of non-linear circuit transfer:

http://www2.hsu-hh.de/ant/dafx2002/papers/DAFX02_Moeller_Gromowski_Zoelzer_measurement_nonlinear.pdf

in essence this type of analysis would constitute a substantial departure from taking the (ubiquitous) Steady-State analysis approach we have all been taught in school // I could go into details why this is so, but that's a long story ... in a nutshell this is a new approach to characterizing Transient response activity in signal processing circuitry // it took many years after leaving school for me to ponder just exactly why all they teach in school is small-signal (System) theory ... I'm parked outside a cafe and it's getting cold // I gotta cut this short, but when I get a chance we can get into the nitty-gritty of all that other stuff ...

cheers // and thanks for following me down this rabbit hole
~jc