Tube design software

[arvin]

I am about to start designing a SET headphone amp using tubes shortly.Can anyone recommend any software and where it can be obtained, that would help design the circuits and do the maths ,simulations etc?

Thanks

[arnoldc]

Try looking at TubeCAD if it fits your needs. For amp design they also have SE AmpCAD.

[grhughes]

Nothing beats real world bench building and testing. Software can get you in the ball park but the rest takes trial and error. Also if you enter historical precedent circuits in these software packages (circuits that have been tried and tested since the 50's) then red flags and warnings seem to go off! Some of these software packages don't cover all the bases. You also have to work on a stage at a time. None of them will simulate a whole amp with multiple feedback paths. It has been my experience, and I have several of these simulation and software packages, it takes probably a good year of replacing components and trying new things. You would be better off trying to find a design that has a precedent and has passed the acid test of simulations and bench testing and other people are building it too. Good luck!

[Sheldon]

Originally posted by grhughes
Nothing beats real world bench building and testing. Software can get you in the ball park but the rest takes trial and error. Also if you enter historical precedent circuits in these software packages (circuits that have been tried and tested since the 50's) then red flags and warnings seem to go off! Some of these software packages don't cover all the bases. You also have to work on a stage at a time. None of them will simulate a whole amp with multiple feedback paths. It has been my experience, and I have several of these simulation and software packages, it takes probably a good year of replacing components and trying new things. You would be better off trying to find a design that has a precedent and has passed the acid test of simulations and bench testing and other people are building it too. Good luck!

Can't argue your basis thesis. I do think that good software has more value than implied here. I bought both of the TubeCad packages and the filter designer and have used none for a bottom up design. But I did find them very useful to get a feel for the relative effect of various changes. This is one thing that interactive software can do. It allows you to change components and values at the touch of a key and quickly see the nature of the effects. No, it can't replace years of practical experience, but when combined with some experience it can speed up the learning process. For simple power supply design PSUD is a very accurate modeling tool, if part values are accurately designated. It's free too (and the TubeCad stuff is not free , but still very modestly priced).

Sheldon

[Salas]

You are right. PS designer is great. As for various spice style sims I have tried they are very nice for letting me see what kind of tubes are more suitable in gain stages etc, see about miller effect and the like. They are never spot on with currents and voltage drops +/- 25% I would say. Not even remotely I would sim a whole amp in them and go build it in final chassis with expensive components.

[grhughes]

Tubecad doesn't compute heater to cathode voltage limitations when you are arranging tubes in cascode or consider positive or negative or feedback. Feedback will change virtually every component in a design. It's a good package and I'm not trying to knock it. John Broskie deserves great KUDOS for his efforts. But it's a start, not the total answer.

[phn]

The answers here remind me of "macho" coders who insist on using NotePad or TextPad and call people who use development tools amateurs.

Thorsten Loesch swear by PSpice. And though I haven't built any tube gear, I would think LTSpice (which you can dowload for free) will take you pretty close to real-world measurement. You still need to do some sort of breadboarding and fine tuning. Triode models will take you closer to real-world measurements than pentodes.

And spice lets you get away with not being math wiz.

[tubelab.com]

Nothing beats real world bench building and testing. Software can get you in the ball park but the rest takes trial and error.

This is very true. If you are building an amplifier design that has already been done before, then you can build the basic amp, and tweak it until you get what you like. But what if you want something different. Simulation is very useful here. I use the TubeCad programs and LT spice, and PSUDII. Each have their advantages and limitations.

The TubeCad programs (SE amp cad, P-P calculator, and TubeCad) are all single stage amp simulations. Each has a few fixed circuit topologies and the tube choices are limited to the popular tubes in triode configuration only. The results (within these limitations) are surprisingly accurate. If you want to run through a "how much power can I get from a given tube, and what load and supply voltage do I need" scenario, then SE amp cad or the P-P calculator is excellent. They are useful for estimating the bias voltage and drive requirements. TubeCad is a single stage simulator for small signal tubes. It will tell you how much gain you will get and allow tweaking the bias and supply voltages to get the best operating point. I have used them all to get a new design "working in the simulation world" before actually building it.

PSUDII is a nice program again for determining a starting point for power supply design. There are a lot of factors that affect its absolute accuracy, but it is excellent for comparing different filter circuits with the same transformer.

LTspice is a full featured Spice simulator. I have simulated complete amplifier circuits (including feedback loops) in LT spice. I use this simulator to try out wild new ideas for feasibility before actually building any circuits. It allows mixing semiconductors and vacuum tubes in the same circuit. I have a simulation of the SimpleSE amplifier design (two stage SE amp) that works reasonably well in triode, UL and pentode mode using a 12AT7 and a 6L6GC. I can use it to test out experimental circuits in the simulated world before actually building anything. The accuracy is only as good as the tube models that you use. There are a lot of models on the web. Some work, some do not.

See post number 22 in this thread for an example using this simulation:
Adjustable distributed load discussion

If the simulator says something will work, it may have a chance of working. If the simulator says it won't work, the simulator is usually right. It is a useful tool for weeding out the dumb ideas cooked up in a blonde head! Many of my circuit ideas never make it past this stage, because it becomes obvious during simulation that I overlooked something important in the original idea. This saves breadboarding time for something that has a better chance of working, and improves the overall success ratio.

You can download LTspice here. It is called Switcher Cad, which was its original intent:
http://www.linear.com/index.jsp

I have used the models from Duncan Amps with good success. For triodes, use the "generic triode model".
http://www.duncanamps.com/spicemodels.html

There is a "how to page" here:
http://www.duncanamps.com/technical/ltspice.html

Some examples of simulator use are here:
http://www.tubelab.com/Simulations.htm
http://www.tubelab.com/PDcookbook.htm

[Salas]

Its all down to models accuracy and flexibilities. Sim as a tool is a great practice. I never got +/- 5% bench approximation with tubes though. Maybe bcs tubes (especially reissued-copied items that we actually use) vary from what their datasheets are about. +/- 15-25% is the norm for currents and voltages. For ballpark sims are great.