LTSpice model for output transformer

[jazbo8]

I tested different tube models a while back.

What's the reference shown in the graph?

[überfuzz]

It's just a sine wave used as in-signal with big enough amplitude to make the models clip.

I seem to remember that the linear behaviour among tube models is very close. I.e. he, or we, won't see any significant difference in frequency behaviour between the different models in our simulations. In addition to this, I could not really detect any audible difference in clipping behaviour between tube models in audio simulations of my test circuits. So, this boils down to, no it doesn't matter what tube model you use.

Edit
Disclaimer: The lets say, poor, clipping simulation might not be due to the tube models, rather an under the hood feature of LTspice.

[jazbo8]

It's just a sine wave used as in-signal with big enough amplitude to make the models clip.

I seem to remember that the linear behaviour among tube models is very close. I.e. he, or we, won't see any significant difference in frequency behaviour between the different models in our simulations. In addition to this, I could not really detect any audible difference in clipping behaviour between tube models in audio simulations of my test circuits. So, this boils down to, no it doesn't matter what tube model you use.

I see, for small signal analysis such as AC sweep (frequency response), that is indeed the case, but when you push it to the point when the grid begins to conduct, then the SPICE model that you use does make a difference. In any case, this is usually not a too much of a concern for small signal tubes, but for the output tubes, using models with more accurate control grid and screen grid modeling are generally preferred.

[überfuzz]

Hmm... I just had a brief look at my latest build, or rather the spice file, to see how I did set up the OT (normally I use connected coils). Seems I didn't even bother to set the series resistance in the coils..?

[überfuzz]

I see, for small signal analysis such as AC sweep (frequency response), that is indeed the case, but when you push it to the point when the grid begins to conduct, then the SPICE model that you use does make a difference. In any case, this is usually not a too much of a concern for small signal tubes, but for the output tubes, using models with more accurate control grid and screen grid modeling are generally preferred.

Ok... Well, if you're able to set up a simulation where you, actually, hear or see the difference I'd be mighty surprised. I mean, lets not forget that we simulate amplifiers of signals going to a speaker, eh, in the real world.

[Tooboob]

Great thread! Glad I didn't miss it, and it's great to find out that there are others here using LTSpice on tube amp circuits.

[hkc]

Has anyone simulated a tube rectified linear power supply?
I haven't been able to find much guidance on the transformer models but I see that Nakabayashi has models for tube rectifiers.
Seems that most of the discussions about transformers and LTSpice focus on the design of switching supplies.
I would like to try to compare tube rectification against semiconductor rectification.
Blencoe emphasises the importance of power supply sag for pentode screen compression and I'd like to try to simulate that in Spice.

Unfortunately the Fender Bandmaster simulation on the Yahoo LTSpice group page doesn't simulate the rectification of AC it just uses a fixed DC voltage source so no power supply sag.

Hank

[jazbo8]

Has anyone simulated a tube rectified linear power supply?
Hank

You can try Ducan's PSDII, which was written exactly for such purpose, easier than duplicating the effort in LTSpice.

[überfuzz]

Has anyone simulated a tube rectified linear power supply?
I haven't been able to find much guidance on the transformer models but I see that Nakabayashi has models for tube rectifiers.
Seems that most of the discussions about transformers and LTSpice focus on the design of switching supplies.
I would like to try to compare tube rectification against semiconductor rectification.
Blencoe emphasises the importance of power supply sag for pentode screen compression and I'd like to try to simulate that in Spice.

Unfortunately the Fender Bandmaster simulation on the Yahoo LTSpice group page doesn't simulate the rectification of AC it just uses a fixed DC voltage source so no power supply sag.

Hank

I do LTspice simulations of SS rectifiers in my builds. However, if you do a transient simulation you'll just sit there and wait for the levels to adjust. Well, lo and behold, they often level out at the exact figures you calculate. The sag, voltage drop when pushed, is however harder to estimate, at least for me. You could set B+ as a voltage source and it will be good enough. If you want to now the sag you'll be able to so by simulating the rectifier+filter, but it wont really simulate how it effects the amplifier.

[jazbo8]

If you want to now the sag you'll be able to so by simulating the rectifier+filter, but it wont really simulate how it effects the amplifier.

Why not? You can use a step waveform on the input and observe the sag's effect across the whole amplifier, since all the operating conditions of the tubes would be changed, assuming of course that you took the time to incorporate the whole PSU into the netlist. But I think it's easier to just use PSUII to figure out the lowest voltage of the sag, and plug those numbers back in the voltage sources in LTSpice to see the "worst case" scenario. But either way will work...

[überfuzz]

Why not? You can use a step waveform on the input and observe the sag's effect across the whole amplifier, ...

Did you read my post..? :-)

[Dave H]

Unfortunately the Fender Bandmaster simulation on the Yahoo LTSpice group page doesn't simulate the rectification of AC it just uses a fixed DC voltage source so no power supply sag.

Hank

Why not 'right click' on the voltage sources and add series resistance to simulate power supply sag? I'd just use one voltage source (V5) and add the normal filter Rs and Cs you would use in a real amp circuit.

[jazbo8]

May be I don't understand what you are saying - not sure why you want to wait for things to settle down - that's steady state...

[überfuzz]

May be I don't understand what you are saying - not sure why you want to wait for things to settle down - that's steady state...

If you model a power section it will take about a minute or more for it to settle. Any readings before this will not represent the live amplifier very good. With tube that models the heaters the warm up is even longer. You know, just like in an amplifier. Well, this is basically why I mostly opt for a plain voltage source for the B+.

[jazbo8]

If you model a power section it will take about a minute or more for it to settle. Any readings before this will not represent the live amplifier very good. With tube that models the heaters the warm up is even longer.

I actually agree with you, I also use plain voltage sources for the B+ - there is really no need to add more complexity to the simulation. So the following is just for academic discussion... you make a good point WRT the time it takes for the circuit to settle down, but with SPICE, you can take a snap shot using timed or stepped waveform, which is done everyday for power supply design, etc., so I don't see why the same procedure could not also be used to analyse audio power amps, whether it is tube-based or not is besides the point. But whether it is worth the bother is another thing...