Pentode Models Comparison
I have been trying out different ways to generate pentode models, out of curiosity mostly… Given the wide tolerance of vacuum tube specs, I don’t think the differences between these models would mean much, but nevertheless, I am hoping that through actual bread-boarding, testing, I will eventually find a model that best matches real life performance data.
Here is a summary on how these models were derived:
1) The McLean Model – this model comes courtesy of Robert McLean’s BeamPentodeModel Excel worksheet, Robert was kind enough to share his proprietary worksheet with me. Since Robert is well known as the tube SPICE guru, I have good faith in this model, although Robert did mention that the model is still well short of his own expectation.
2) The Eureqa Model – this model comes from Eureqa Formulize, a curve fitting program designed by the smart folks at Cornell, which uses sophisticated algorithms to derive equations given a set of data. I scanned the IV curves from the datasheet with CurveCaptor (which is great for generating triode models BTW), and imported the data into Formulize. The optimization process took about 10 minutes, the longer the program runs – the better is the fit. But for my purpose, I don’t think the extra precision is really needed. Close enough for rock & roll, as they say.
3) The Koren Model – this is the standard Koren pentode model using the same data set from the above curve capture process. Robert McLean’s KorenModel worksheet was used to generate the model parameters – this one comes in handy for another reason – the tube parameters could also be used in Chuck Doose’s power amplifier worksheet, none of the other two models can do that since they do not use the same set of parameters as the Koren model.
As mentioned, actual testing of the tubes in real circuit designs will be needed to verify the usefulness of these models, but at least it is a starting point…