Vacuum Tube SPICE Models

[ja2dhc]

Originally Posted by kevinkr

JA would you mind posting the files here? (zipped)

Hi, ! Attached,

[Ray Waters]

Thanks, all, for the models. Looking through the list, I didn't see a model for the 6DJ8/6922 or the 6N1P. Is there a possibility that these are part of the set but just omitted from these ZIP files? Or does someone have the expertise to generate them? If I had these two additional models, that would cover all of the tubes I generally design with.

Ray

[ja2dhc]

Originally Posted by Ray Waters

I didn't see a model for the 6DJ8/6922 or the 6N1P.

The European name of 6DJ8 is ECC88 which you can find in the list.
6N1 is not listed.

[Ray Waters]

Thanks, I should have noticed that myself.

[jazbo8]

Originally Posted by spirakos

The files LTspice supports are
Schematics(asc)
Netlists(cir,net,sp)
Waveforms(raw,fra)
Discretes(bjt,cap,dio,ind,jft,mos,res)
Symbols(asy)

I did change ^ with ** but an error wakes up
The error says ' Multiple instances of "print" '

It seems like you are trying to open the .inc file directly... Please refer to the image below, it should be self-explanatory of what needs to be done... If not, please consult the user manual for LTSpice.

[chanmix51]

Hi,

Here are some of the models i use : https://gist.github.com/chanmix51/6947361

Cheers,
Grégoire

[jazbo8]

Originally Posted by chanmix51

Hi,

Here are some of the models i use : https://gist.github.com/chanmix51/6947361

Cheers,
Grégoire

Thanks for sharing, but it appears that most of the models in your collection with a few exceptions are based on the Koren model, which unfortunately are not accurate (in particular the pentodes/tetrodes) - sorry to sound like a broken record . I will just show an example of the 6V6 here for illustration purpose. The other Koren-based models also suffer from similar issues to varying degrees - the spacing of the grid curves, the knees, etc. Don't even get me started on the grid modeling...

6V6 vs. the Tung-Sol datasheet (the -30V Eg curve does not even show up on the plot...)

[chanmix51]

Hello Jazbo8,

That's true, the 6V6 model is one of the worst I could find. Since all those models come from different sources, some others are somehow better.

If anyone had a 832 (GU32) pentode spice model, I'd be interested

Cheers,
Grégoire

[jackinnj]

If you "step" the iteration, not rushing to just arrive at a "solved" solution, you can find Koren parameters which model pentodes with reasonable accuracy through the knee. Watch how the errors change every time you step the process. Moreover, there are many ways to look at the error you're minimizing, and it's necessary to use the correct one!

It's probably easiest to just model these devices in the area in which they will be used (Va>=50V or Va>=75V), however.

[jazbo8]

That's beyond my pay grade... I am looking forward to see some new methodologies and software tools for pentode model building.