Announcement

Collapse
No announcement yet.

Yet another op-amp tube emulation stage - simulated grid current & bias shift

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Yet another op-amp tube emulation stage - simulated grid current & bias shift

    I was messing around with LT-SPICE and came up with a simple emulation of a prototypical preamplifier circuit. The top circuit is a basic 12AX7 gain stage consisting of an input circuit and a Marshall-type cathode follower, such as is often used to drive a tone stack, running off of 300V, simulated using Duncan Munro's 12AX7 model. The bottom circuit is a opamp circuit running off +12 and -12, using the official TL072 SPICE model. They look very different, yet produce nearly identical output in silico:



    Red lines indicate similar breaks between subcircuits in both examples. The tube circuit uses a single 12AX7 for the input gain stage, while the opamp circuit uses 2 TL072 stages as a single stage cannot provide sufficient gain with a full 20KHz frequency response. The green box shows the subcircuit which provides an approximation of grid current draw.



    Here's the output! Red trace indicates the input signal. Green trace is the simulated tube circuit, and the blue trace is the simulated opamp circuit. There are obvious similarities here - grid bias shift produces asymmetrical clipping and dynamics, the output is almost identical and I bet it'd sound pretty much the same.

    So a simulated tube amp, and a simulation of a simulation of a tube amp

    The CF tube stage is actually, surprisingly, pretty much a symmetrical square wave clipper. Pretty much all the delicious even-order harmonics come in because of bias shifting! At least in my simulations. Probably in real life, as well! The TL072 circuit with 2 Zener diodes in the feedback loop gives pretty much the same output for the same input voltage, except that it's level shifted somewhat and is scaled down by a factor of about 25 in voltage. Because the TL072 can't swing 300 volts! The funny green-box circuit provides an approximation of grid current draw here, scaled up by a factor of 10. This requires a 0.22 uF coupling capacitor instead of a 0.022 uF cap, to give identical bias-shifting effects, but allows the use of resistances 10x smaller in this circuit to avoid what might otherwise become serious noise problems. The 3.9K resistor is a simulation of plate resistance which the actual tube stage has, as opposed to the near-zero output impedance of an opamp. It's also bad bad bad to run an opamp into a weird load without series resistance, I've found they tend to oscillate if I do that.

    If anyone wanted to build the opamp circuit, I'd recommend bypass caps on the ICs, an input protection network for that first noninverting stage, all that usual stuff. This wouldn't be a high-gain thing by any means, but put a tone stack circuit after it and you might have quite a 'tubey' sounding preamp! And a higher-gain tube circuit could totally be simulated this way, I think, as well.

    So . . . interesting? I think so.

  • #2
    Originally posted by Morris Slutsky View Post
    And a higher-gain tube circuit could totally be simulated this way, I think, as well.

    So . . . interesting? I think so.
    Well, that's a good start but there are a couple of things you need to add to your simulation. One is the subtle (at low levels) and not so subtle (at large signal levels) distortion that a 12AX7 adds to the signal. You can see it in the X-Y scope photo below. Horizontal axis is the input (no grid stopper) at 500mV per div and the vertical is the output at 50V per div. Add to that the effect of B+ regulation. When you hit a guitar amp with a big signal, the trace sinks down as the B+ sags and changes slightly. Small amounts of this sag sneeks through the coupling caps and change the bias on the next stage.
    Attached Files
    WARNING! Musical Instrument amplifiers contain lethal voltages and can retain them even when unplugged. Refer service to qualified personnel.
    REMEMBER: Everybody knows that smokin' ain't allowed in school !

    Comment


    • #3
      Hi Morris , it's *very* interesting indeed, thanks for sharing.
      I think the road to tube independence in the future goes in this general direction, the other path being pure digital simulation.
      I, for one, am not taking the latter one, no Sir !!
      Of course different people will have different ideas as to exactly *what* to emulate from our beloved tubes, I for one do it in a different way, but hey !! , that's the fun part of it !!
      A fellow experimenter in SSGuitar made an impressive triode emulation ... using 19 Op Amps, no kidding.
      There was quite a lively discussion over there.
      I think Mr Pritchard is the current leader in this field, take a peek at his patents.
      It would be very interesting if you could sometime Protoboard this schematic, hook a Guitar there and post some MP3.
      Thanks again.
      ps: *Loved* the other "schematics" posted just below yours, on the same page, he he.
      Juan Manuel Fahey

      Comment


      • #4
        Thanks for posting a trace! Is that for this same CF circuit? I know that modeling work is only as good as the models.

        Here are the voltage and current traces for my modeled output stage, the tube one and the opamp one:




        The horizontal axis is input voltage. The cyan trace is the output (cathode) voltage for the CF stage, while the green trace is the output voltage of the opamp stage, scaled up by a factor of 24 and shifted to match as closely as possible. Surprisingly, the opamp stage seems to clip more softly. This is probably due to the zener curve's knee being all stretched out because of the relatively low gain of the opamp stage. The red trace is the input current for the tube stage including the grid bias resistor and the grid conduction current, while the blue trace is the input current for the opamp stage divided by 10, the sum of the current drawn by the input impedance of the inverting follower and the diode/zener bypass network. So it's pretty close up to scaling.

        Your trace seems to show a very soft clip as it nears the conduction saturation. That could be waveshaped in if necessary and maybe it would be necessary. Probably replacing the 2 back-to-back zeners with a more asymmetrical combination and perhaps more diodes and resistors could do it.

        Comment


        • #5
          That's pretty cool stuff Morris!

          Have you put a guitar through it?
          How about scaling the op-amps feedback/diode network and using the loop around tube stages! Ooh! Man! That might be too tubey!
          Peace,
          Cru

          Comment

          Working...
          X