A nice tube emulation article and a question

[11-90-an]

On my ongoing curiosity about tubes and such, I came upon this interesting article and wanted to share and ask a question... 

https://www.schematicsforfree.com/archive/file/Audio/History%20of%20Audio/The%20Tube%20Sound%20and%20Tube%20Emulators.pdf

In the tube emulation circuit proposed there, there seems to be something called an "AX7"... I don't know if this is an IC or something else as my google search yielded no results... anyone with an idea on what it is?

[duck_arse]

without following the link, I'd guess 12AX7. especially when discussing valves.

[marcelomd]

Yup. 12AX7. Probably the most used tube for guitar (pre)amps.

https://en.wikipedia.org/wiki/12AX7

[11-90-an]

yes but...

from the article,



I dun think you would use a 12AX7 with op-amps to emulate a 12AX7 

[Gus]

Put the patent numbers at the end of the article into a patent search. This one seems to show more of what could the circuit

https://patents.google.com/patent/US5133014A/en?oq=5%2c133%2c014%2c

try another number you find in the writeups like this
https://patents.google.com/patent/US5636284A/en?oq=US08%2f202%2c369

This one might be more helpful
https://www.freepatentsonline.com/5434536.pdf

[Rob Strand]

Back in those days Eric Pritchard was up there with his work on Tube emulation.
He's got a patents on this stuff.

... which Gus just  posted links to ...

[teemuk]

Pritchard has a habit of being very cryptic about details of his intellectual property. Sort of "epoxy gooping" of ideas...

AX7 (e.g. figure 6) refers to a non-linear network that replicates non-linear characteristics of 12AX7 triode. Pritchard presents several such networks in his patents, which are also somewhat cryptic and take a moment to decipher. Most rudimentary network is a resistor-diode "ladder" that forms a piece-wise waveshaper by introducing specific ratios for the resistances involved. The patent literature, by the way, uses sort of diode-resistor symbol combination to present (complex) network circuits in much more simplified manner.

It's a circuit block in general; not quite a diode per se but almost. Just with more triode-like non-linear resistance characteristics.

The simplest "Pritchard triode" is an opamp stage with shunt diode in input (simulating grid diode) and a forward-biased non-linear network ("diode") in output. When output signal amplitude gets high enough the diode reverse biases, seizes conduction, and clips the other half wave.

Patents present the rudimentary concept and wealth of other ideas including several power amp emulation schemes and dynamic filters.

[merlinb]

there seems to be something called an "AX7"... I don't know if this is an IC or something else as my google search yielded no results... anyone with an idea on what it is?

Says in the article:
"The triode tube emulator is a combination of non-linear networks built into an epoxy encapsulated hybrid SIP and application specific components. This SIP has 15 pins on 0.1 inch centers along its entire length and is 0.71 inches high and about 0.125 inches thick. The application specific components provide the gain elements and the grid conduction components."

[11-90-an]

Another interesting thing...

https://gabevee.tripod.com/sstube.html

[dschwartz]

Another interesting thing...

https://gabevee.tripod.com/sstube.html

It reminds me to the bjt Vulcan..
I wish it had a more extensive explanation, otherwise, it looks like a buffered common emmiter stage

[iainpunk]

Another interesting thing...

https://gabevee.tripod.com/sstube.html

It reminds me to the bjt Vulcan..
I wish it had a more extensive explanation, otherwise, it looks like a buffered common emmiter stage

its a P-channel MOSFET, this configuration does have gain from both the FET and the BJT, and they interact because of the DC coupled nature of this gainstage set.

cheers, Iain

[dschwartz]

Another interesting thing...

https://gabevee.tripod.com/sstube.html

It reminds me to the bjt Vulcan..
I wish it had a more extensive explanation, otherwise, it looks like a buffered common emmiter stage

its a P-channel MOSFET, this configuration does have gain from both the FET and the BJT, and they interact because of the DC coupled nature of this gainstage set.

cheers, Iain

Oh i see
Thanks!

[Vivek]

Ha Ha

"Recently Eric Pritchard started down the circuitous path.........."

[amptramp]

A lot of these tube simulators miss one aspect of the simulation: interelectrode capacitance.  In tubes, they are small, on the order of a few pF and there is little variation with electrode voltage.  In FET's, they are larger even for (RF FET's) and the capacitance varies significantly with drain voltage.  For bipolar transistors, the capacitances can be larger and they also vary with voltage.  Even if you managed to get the DC characteristics you wanted, you might find a difference in dynamic characteristics.

Want something to sound like a tube?  Use a tube.

[Fancy Lime]

Another interesting thing...

https://gabevee.tripod.com/sstube.html

It reminds me to the bjt Vulcan..
I wish it had a more extensive explanation, otherwise, it looks like a buffered common emmiter stage

its a P-channel MOSFET, this configuration does have gain from both the FET and the BJT, and they interact because of the DC coupled nature of this gainstage set.

cheers, Iain

The image is mighty blurry but I read that as an n-channel.

Andy

[iainpunk]

Another interesting thing...

https://gabevee.tripod.com/sstube.html

It reminds me to the bjt Vulcan..
I wish it had a more extensive explanation, otherwise, it looks like a buffered common emmiter stage

its a P-channel MOSFET, this configuration does have gain from both the FET and the BJT, and they interact because of the DC coupled nature of this gainstage set.

cheers, Iain

The image is mighty blurry but I read that as an n-channel.

Andy

yeah, i misread that MOSFET, its a N-channel indeed, i shouldn't posd while drunk

i think it uses the inherent non linearity of a mosfet buffer.
makes me wonder about IGBT sound quality, they are supposed to have super high gain

cheers, Iain

[dschwartz]

It's easy to fall into the "tube emulation" rabbit hole without considering the fundamental thing...it just has to sound good..
After 15 years analyzing tubes, emulators, tubes alternatives, jfets, mosfets, cmos inverters, etc..and designing more than 20 distortions and preamps, i came to the conclusion that it doesn't matter if it simulates tubes down to the minimal detail....with the right gain structure, the right filtering and voicing and the right clipping symmetry and shape, you can get the most awesome tones and feels.. you just need a good pair of ears and good knowledge of what affects what and the rest is like painting.. it's an art and getting into the tube emulation it's like trying to replicate the mona lisa, while forgetting that you can do even better than that.

[PRR]

...i think it uses the inherent non linearity of a mosfet buffer....

Which is about zero.

Without knowing the resistors, there is nothing more to say about that drawing.

[teemuk]

...i came to the conclusion that it doesn't matter if it simulates tubes down to the minimal detail....

True. Not even all tube amps feature those "minimal details".

Take PPIMV for instance; it prevents output pentodes from clipping and also neuters all power amp circuit's dynamics like sag, dynamic bias shifting and screen compression. Yet most people are totally content with the feature and would rather utilise one instead of a modeling amp that actually overdrives more "tube-y" than a mere phase inverter stage, has the "minimal details" so to speak.

Pritchard early circuits, for instance, were mocked for sounding pretty poor but that had nothing to do with poor tube emulation, just poor voicing. Working together with guitarists and receiving artist input has since improved "voicing" of his designs, and tone.

Marshall's early history is a good example of modifying design based on artist feedback and effectively turning a bass-heavy "flubby" Fender circuit into brighter and "tighter" Marshall.

[iainpunk]

...i think it uses the inherent non linearity of a mosfet buffer....

Which is about zero.

Without knowing the resistors, there is nothing more to say about that drawing.

well, if its biased near the linear region, instead of in the middle of the saturation region, there would be quite a bit of nonlinear behavior (clipping)
yea, i'd like to see the resistor values as well.

cheers, Iain