Express Output Waveform Shots

[katopan]

Yes, I used the Munro NH12AX7 out of his dmtriodep.inc file. Thanks for the link. And yes, I very much agree about the pentode models. But they are good for indication as long as you're aware of their limitations. It did allow me to do a heap of testing very quickly and really see how this asymmetry was being generated. Also in seconds I could test the difference of changing the cathode bias resistor and the tail resistor. Much quicker than getting the soldering iron out.

I've used the simulations to hone in on what I now need to test on the real circuit for validation.

Thanks for following this along. Otherwise I'd be getting a bit lonely here in this thread.

[martin manning]

Have either of you guys looked into Derk Reefman's models? http://www.dos4ever.com/uTracer3/uTracer3_pag14.html

[jazbo8]

Have either of you guys looked into Derk Reefman's models? http://www.dos4ever.com/uTracer3/uTracer3_pag14.html

Yes, Dr. Reefman's models are definitely the best available currently, but you really need to have the uTracer to generate the models properly. My only pet peeve is that people are/will create "one-off" models based on a single sample in their collection - and those models could deviate quite a bit from the bogey shown in the datasheets.

[martin manning]

I too fear the proliferation of models. There is also the posibility that the u-Tracer data used to make the model is inaccurate, as I have found several ways that that can happen. That aside, the problem is to find a representative example to base the model on.

I see no better way to create a good model than to use real data, though, and that is the great advantage that the u-Tracer brings.

[jazbo8]

I too fear the proliferation of models. There is also the posibility that the u-Tracer data used to make the model is inaccurate, as I have found several ways that that can happen. That aside, the problem is to find a representative example to base the model on.

I see no better way to create a good model than to use real data, though, and that is the great advantage that the u-Tracer brings.

No problem if the models from the uTracer are used for one's own design, that's the ideal scenario and the simulation results can closely matched the real thing. In any case, I'm just waiting for uTracer v.4 to come out - hopefully not for too long - it's a great tool.

[martin manning]

The higher voltage capability of u-T4 would indeed be a good thing for extending the range of available test data, but that might be a long time coming, and it may have issues of its own to be solved.

One big improvement that Reefman offers is in improved modeling of grid current, which is down at the low end of the Va-k scale. u-T3 can generate the data to cover that area now.

[jazbo8]

The higher voltage capability of u-T4 would indeed be a good thing for extending the range of available test data, but that might be a long time coming, and it may have issues of its own to be solved.

One big improvement that Reefman offers is in improved modeling of grid current, which is down at the low end of the Va-k scale. u-T3 can generate the data to cover that area now.

So I may have to pull the trigger on v.3, hmmm... Apologies to Craig for getting OT, but Martin, what's your take on the TWE scope shots, anything jumped out at you?

[katopan]

No worries. All relevant and interesting for someone who's been using the Duncan models in Spice for years.

[martin manning]

what's your take on the TWE scope shots, anything jumped out at you?

I've only done a drive-by on this, but the cold-clipping and LTP PI generating duty cycle modulation (from Aiken) are familiar to me. I've also generated the over-driven LTP duty cycle modulation with SPICE... I don't think anything too fancy in the way of a tube model is needed for that.

[katopan]

Martin, could you please explain in detail your understanding of how the asymmetry is generated in the LTP PI?

[martin manning]

I don't claim to understand it completely, but I am familiar with the phenomenon from reading discussions like this http://ampgarage.com/forum/viewtopic.ph ... highlight= , and Merlin Blencowe has some discussion on overdriven LTP duty cycle modulation in both editions of his Preamp design book. In the second edition there is an illustration of how the selected bias point determines which side will increase in width. The Express' LTP is biased hot, which according to Blencowe will cause the inverting triode to clip harder on the positive side and increase its duty cycle.

[dave g]

Merlin Blencowe has some discussion on overdriven LTP duty cycle modulation in both editions of his Preamp design book. In the second edition there is an illustration of how the selected bias point determines which side will increase in width. The Express' LTP is biased hot, which according to Blencowe will cause the inverting triode to clip harder on the positive side and increase its duty cycle.

Ah yes, I remember that. He also says that the effect is reversed if the LTP is biased cold, and a center biased LTP will not have the duty cycle modulation.

I wonder then what an express would sound like with a center biased LTP. Replacing the 470R cathode resistor with a 1k would do it.

[katopan]

This is driving past the smaller effect of bias point. I did simulations in LTspice of centre 750R and cold biased 1K vs the warm biased 470R. Made a difference at clipping but not at saturation. With all three resistors the duty cycle shift was the same direction!

[martin manning]

That would be worth an experiment to see if the sim is getting it right.

[katopan]

Defininately!