music, guitars & other obsessions

An Overview of the Pentode Model Building Process

[2/13/2013 – For those interested in my latest finding, head over to this post.]

  1. Modified PaintKIP java app, which allows live manual adjustments of the pentode parameters, the process is relatively quick, but we have little clue about the actual error rate (the new version from Dmitry may fix that). The models are immediately usable and pretty stable. This is the quickest and most interactive method.
  2. Bill Elliott’s Derive program and data selection, a bit slower since it requires careful selection of the data points from the characteristic charts (both pentode and triode charts are needed) and data needs to be entered manually into the worksheet. And the worst part, the results often do not converge in SIMetrix, which requires additional troubleshooting. In LTSpice, they seem to behave much better, which is strange indeed.
  3. Formulize parameters optimization, while the parameter extraction process itself is very quick, multiple steps are required – the anode and screen characteristics need to be processed separately, then further adjustment to the screen current needs to be made before the model could be used.
  4. Teodoro’s Excel Solver, this is the simplest to use, but the screen current is simply a rough estimate based on predefined formulas using the parameters from the anode characteristics. So the overall error rate is unknown.
  5. PCTube App by Ayumi Nakabayahi, which I have highlighted in a separate post.

It really bothers me that Bill Elliott’s model does not function as it should in SIMetrix, since it is the one that I think encompass all the things I want to see in the pentode model development, will need to get to the root cause of this problem…

While the other methodolgies all provide satifactory and usable results, each has some shortcomings which can not be addressed easily (or may be I am just not trying hard enough…)

Here is a few screenshots of building the pentode model using Bill Elliot’s Derive program, it is very fast once I got the hang of the steps involved. This one is a keeper! I used 6CL6 for the test, and the model matches up well with the datasheet upto 0V, but above it, the mis-match becomes great. Since Bill’s instruction was limited to the zero volt grid line, I did not want to push my luck, as the model is already having difficulty convergening in SIMetrix, perhaps when things settle down, I will give it a shot, since I am very interested in the Class A2 operation.

Data selection per Bill's instruction

Data selection per Bill’s instruction

The model in action, the grid lines above 0V are not close to the datasheet

The model in action, the grid lines above 0V are not close to the datasheet


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s