12AX7 SPICE Model Comparison

[überfuzz]

jazbo8 - Could you explain how you set up the simulation of the tube models in your test runs, those you attached in your first post.

Thanks!!

[jazbo8]

jazbo8 - Could you explain how you set up the simulation of the tube models in your test runs, those you attached in your first post.

Thanks!!

Here you go:

[überfuzz]

Okey, thanks!

When I tested the tube models I have I didn't include any resistances. Are there pro and cons for that? A more general question, does the saturation voltage/plate Voltage give any information that an anode characteristics diagram wont give?

[jazbo8]

The reason I include them here, was to get a closer match to the image from Loudthud using a real tube. But, when plotting the plate characteristic curves like the datasheets, the resistors are not needed.

[loudthud]

A quick note, the plot I posted was with a 33K grid stopper. The X voltage probe is attached on the generator side of the stopper. If the stopper is removed, a lower plate voltage can be obtained.

[überfuzz]

Oh by the way, maybe someone fancy the plot of the anode characteristics I made. The plot is somewhat bloated, but anyhow her it goes.

[Rock Mills]

LTSPICE runs on Mac OSX 10.7 up in 2016

Personally I find the popularity / incompatibility of LT-Spice to be a major obstacle. For those of us who haven't used Spice before but would like to learn how, not having a Windows box is a deal breaker.

Old thread, I know, but I don't want anybody to miss out on LTSPICE for lack of a Windows box. Haven't tried the Mac version yet -- been running it on my Mac in a virtual XP machine on VirtualBox (free, http://virtualbox.org), along with SEAmp, TubeCAD and other non-Mac software. Works great! Helps keep my tube-geeking (somewhat) separate from my 'normal' life. VirtualBox lets you run just about any OS on any hardware.

Agreed, SPICE tube models give, at best, a crude stab at grid behavior, but are good enough to reality-check the basic operation and response of your design. You can fiddle things and see what happens. Clipping onset and the harmonic consequences kick in at a reasonable approximation of reality. Needy tone stacks drag gain down realistically. Tremolos tremolo. Change B+ and watch those starve/over-fed OP tubes react. Given the option, I'd rather do the basics without the solder fumes. Keep an eye on what you're pumping into the next grid at noon gain/volume, build it, and cautiously push it to the edge. Warm up the iron and proceed. Great tool to explore what all those bits are doing in there, and what changing/adding/removing them does.

Tube models that could closely approximate behavior at the extremes are the holy grail, I suppose. The 'not well understood' +grid behavior probably qualifies as chaotic (too many variables). Akin, perhaps, to fluid dynamics, requiring copious cycles to simulate (like weather). Could be done, and I might be willing to let a box chug for three days or more ....

I think look-up tables of recorded, averaged, real tube behavior, though voluminous, will simulate faster and perhaps be more predictive than best-fit curve math(s -- apt because several types are required). Computers are great at memory access, less so at complex multivariate math. Plot actual production tube traces, not old datasheets. Individual NOS tubes, if you're anal. Models are great. But....

In the end, you play it and use your hind-brain and fingers to wrangle it.

[Old Tele man]

Close enough for "ornamental" work!

[überfuzz]

I allways start with spice simulations where i tap in the voltage feed by the rectifier. Then the fiddeling starts. When im dead sure that i got everything exactly the way i want it i start the build. When its done a protest is filed by my ears. So now the modding starts.

Works like a charm! Everytime.

[Old Tele man]

My "tongue-in-cheek" analogy -- Equations provide theoretical values, SPICE provides approximate values; but, the ears provide exact values.