FET simulations of tube circuits

[michael_ibrahim]

Hi people,...

I posted earlier in another thread about FET simulations of Tube circuits, so I thought I'd post the kinda thing I was talking about. There are 3 circuits shown below:

1) Top right: Tube circuit to be simulated.
2) Top Left: my FET simulation.
3) Bottom: standard simulation



The pot in the second circuit would be used to get the correct bias (a pot of approximately the same value as the plate resistor is the best choice, so in this case a 250k trimmer will do just fine,... and it would probably be set to about 150k). For the 'standard' circuit, the pot does the same thing (bias setting), but since you want to get approximately symmetrical swing (as in the real tube circuit), the trimmer would be set to about 2k. the resulting circuit is this very different to the tube circuit you're trying to emulate, and this is shown quite dramatically in the simulation results:



The 3 frequency response curves are:
1) Red: Tube circuit
2) Blue: My proposal
3) Green: the 'standard' emulation.

Just to show you how close this is to the Tube curcuit, here's a closeup. It's within 0.5db of the Tube response:



Hope you find this interesting!

[B Tremblay]

Have you built it yet?  If so, what were the FET pin voltages when using a 220k resistor?

[michael_ibrahim]

No, I haven't built a tube amp sim with this. I'm working on a few real tube circuits at the moment, and I came across the post, so I got thinking,...

The pot should be adjusted so that the source voltage is about 4.5V for a 9V supply. The drain would be about 2.5V, I guess,...

Oh yeah,... building a complete simulation like this will result in WAY too much gain and clipping. A Recto, for example, has about 110db of gain at full tilt,... but the guitar signal in the recto won't be clipped until it's amplified by about 40db, since the voltage supply is so high (400V). But in the case of the FET emulation, the 9V supply means that you need about 30db less gain to get the same about of distortion. In most cases, replacing the first tube stage with a voltage follower would do the trick, with a bit of extra frequency shaping to make up for the missing stage.

[B Tremblay]

I just finished breadboarding both FET versions for comparison.  I added a standard 100k-A volume pot to test the stage as a booster.

When using the 220k drain resistor and 2N3819, I biased the drain to half the supply voltage.  I was using a hot new battery that measured 9.53v, so I biased near 4.76v.

The sound was a bit constricted and there was noticeable distortion present.  The sound quality improved when I increased the drain voltage to 6v, but was still not clean.

I then tried the drain trimmer version, which produced a full, clean sound.

Out of curiosity, I replaced the 2N3819 with a J201.  When biased to 1/2V, the J201 was louder and still somewhat distorted.  However, the mis-biased sound was not present.  Using a J201 drain trimmer version produced results identical to the Fetzer Valve.

Code Select Expand

Here are my pin voltages for 220k drain resistor version:

2N3819   D: 4.74, S: 4.49
J201     D: 4.70, S: .678

These are the readings for the higher bias version that sounded better:

2N3819   D: 5.95, S: 4.60

And now the voltages for the drain trimmer version:

2N3819   D: 4.73, S: 3.24
J201     D: 4.73, S: .365

I hope this information is helpful.

[puretube]

I always simulate that first (upper left) standard stage with 320V
for a 2N3819...  

[ragtime8922]

Let's get REALLY unique and design an all tube circuit that simulates a transistor!!! :idea: ... :lol:

[puretube]

you want Ge or Si  :?:

[michael_ibrahim]

Hey, why not use an ancient all-valve computer to simulate a valve  :wink:

[Eddie]

How do I change the thunderchief first JFET stage into a voltage follower? SchematiThanks a lot!


Eddie

[RedHouse]

B Tremblay
I just finished breadboarding both FET versions for comparison

Quick question B Tremblay, did you use a proto-board or perf? I proto'd the Fetzer, Sparkle Boost, and Thunderchief and got ...none... of them to pass any signal.
(yes I had 4.5v bias)

Proto boards are known for high capacitance and I was wondering if the proto-board was futzin the circuits.

I have like 25 2N5457, 10 J201, 3 MPF102 in my stash and none seem to work so it's really got me wondering.

[Ge_Whiz]

So, joking apart, our conclusion here seems to be that the 'new' circuit gives great simulation results but doesn't sound right, yes?

This begs a few questions about the value of simulation, I think.   :?

[Ge_Whiz]

A typical proto-board capacitance would only cause significant problems with RF circuits, a few MHz and above, not audio, unless ridiculously high gain.

[smashinator]

Let's get REALLY unique and design an all tube circuit that simulates a transistor!!! :idea: ... :lol:

I've tried out a couple of tube amps that did that pretty well.  All of a sudden it became apparent why they were on sale.    :lol:

[B Tremblay]

Quick question B Tremblay, did you use a proto-board or perf? I proto'd the Fetzer, Sparkle Boost, and Thunderchief and got ...none... of them to pass any signal. (yes I had 4.5v bias)

I used a little RadioShack breadboard, the same one that was used in development of all the runoffgroove.com projects.  Many other builders have successfully built the projects you mentioned, so my guess it that you may be making the same error over and over.  FET pinouts, possibly?

[B Tremblay]

So, joking apart, our conclusion here seems to be that the 'new' circuit gives great simulation results but doesn't sound right, yes?

Well, for what it's worth, that's my conclusion.  I highly encourage others to perform the same tests.

[puretube]

So, joking apart, our conclusion here seems to be that the 'new' circuit gives great simulation results but doesn't sound right, yes?

Well, for what it's worth, that's my conclusion.  I highly encourage others to perform the same tests.

I don`t dare drawing a conclusion here, but want to give to think, that I honestly wasn`t joking about having simulated a couple of my tubecircuits
(a.o. the "Tube Zipper") as described earlier, (2N3819 MicroSim model instead of 12AX7, with >300V, allcaps & resistor values like in real life),
and it quite decently resembled the real thing I get on a scope...

I may have to add, that I did the simulation years after having built the actual things  

[michael_ibrahim]

Hey B Tremblay,

You've got a great point about the bias. For this circuit, the bias should be setup so that the drain voltage is half way between the supply voltage and the source voltage. So if the source is at 3V, and the supply is 9V, the drain should be biased at 6V.

I did a quick test of the frequency response of this circuit in real life, and it's pretty close to the sim,...

[puretube]

fine observation - Ri of a 12AX7 in standard configuration is ~68k...

[RDV]

Let's get REALLY unique and design an all tube circuit that simulates a transistor!!! :idea: ... :lol:

Too late, Peavey did that already with the VTM60!!



RDV

[B Tremblay]

For this circuit, the bias should be setup so that the drain voltage is half way between the supply voltage and the source voltage. So if the source is at 3V, and the supply is 9V, the drain should be biased at 6V.

I wondered if that might be the case, so it's good to know I wasn't too far off.

Between the 2N3819 and J201 when biased in this manner, I prefer the J201.  Neither are clean, but the tone of the J201 sounds better to my ears.  However, the character of the distortion isn't that impressive.

Could someone else please breadboard it and post their findings?  I'd rather not be providing the only findings for this circuit!