Multisim and Ultiboard

cancel
Showing results for 
Search instead for 
Did you mean: 

Why are generic tube models so sensitive to simulation error?

Well i have two models of the 12ax7 vaccume tube. The one on the left is the virtual component that i have set to the same parameters as the generic 12ax7 on the right.

virtualandgeneric.jpg

They behave exatly the same until the input signal is high enought so the signal out gets clipped. The Virual componet then continue to simulate fine but the generic make the simulation error!!!!

 

clipping.jpg

 error..jpg

 

Why? Whats the difference? All the vaccum tube models behave this way. Some are even worse like the pentodes. They must have exactly the right circumstances otherwise the simulation will error.

 

So why does the virtual componet work beyond clipping and the generic will make the simulation error? 

 

And is there any way to make the simulation go on without error with the generic tubes? 

 

As you can se the convergence assistant cant resolve the issue!

 

I would realy like to se distortion and clipping when simulating with the generic models but its not possible......

Message Edited by Tubis on 02-25-2010 09:47 AM
0 Kudos
Message 1 of 3
(5,146 Views)

Tubis,

 

It doesn't appear to have anything to do with the simulation settings, but it may be due to the model itself.

 

Looking at the model of the 12AX7, there is a divide by zero effect that seems to be happening... 

 

Looking at this part of the model:

 

B0  5 0 v=83.5*v(1,3)+v(2,3)
BIp 2 3 i=1.73242e-006*sqrt(abs(v(5)+v(5))/2*v(5)^2)

 

Due to the diode, it looks like v(1,3) and v(2,3) can be 0 or near 0 - making v(5) -> 0, thus making a divide by zero in 1 /2*v(5)^2 -> Inf...

 

I am not sure the details on this model, I will see if R&D can take a look and provide guidance. 

 

If you can make the existing Virtual Tube model match the specs of the 12ax7 closely, use that for now...

 

If anyone else in the forums has experience with Vacuum Tube SPICE models, please provide some guidance.

 

Regards

Pat Noonan

 

Attached is the circuit used for testing...

0 Kudos
Message 2 of 3
(5,090 Views)

I dont understand much of what makes a model tick. But the Virtual tube model works great to simulate gain, cathode caps freq response, miller effekt, onset of clipping, etc.

The only problem with it is the signal level doesnt stop at the onset of clipping. It just go on and on. You can get thousands of Volts of signal out.

And it would be nice to be able to use the "pentode" models. There is only at "triode" virtual model.

 

0 Kudos
Message 3 of 3
(5,058 Views)