Real time SPICE simulation

[dsharlet]

Hi All,

I've been working on a project that you might find useful: LiveSPICE

It is a real time SPICE simulation tool for audio signals. It allows you to design circuits and simulate them in real time with actual audio signals into/out of your computer.

I'd be curious to hear your feedback if you try it out! It's completely free and the source code is available on GitHub.

 

[spookydd]

Sounds very intriguing. I will take a look at this and I am looking forward to attempt some sims myself.

 

[fireworks]

That's interesting. What SPICE engine did you use, an existing one or your own ?

 

[spookydd]

After taking a look, I find that I will automatically be out of it, since it's strictly windows based. Too bad!

 

[dsharlet]

It is a custom SPICE simulation engine. I had to do a lot of crazy things to make it fast enough to run in real time. Also, to be able to do it with low latency would make it very hard to re-use an existing simulation engine.

Sorry, yes, it only runs on Windows.

 

[jcx]

simulation in the loop for motion control is a active field

I suppose you could sim pedals/effects inline, live during a performance


but for recorded music playback I'm not so sure of the usefulness - you can already feed .wav in, get .wav written out of a sim in free LTspice

 

[fireworks]

I've seen that the code is open source, under the MIT license.
I haven't tested it, but it seems to have a nice user interface.

All I can say is Congratulations !
The open source community needs a fast SPICE with a nice interface. Even if the engine is not as feature rich as others, other developers could improve it.
Really good work !

 

[Zero D]

I've only had a very quick peek, but it seems quite Brilliant

Thanx & all the best with it.

 

[dsharlet]

but for recorded music playback I'm not so sure of the usefulness - you can already feed .wav in, get .wav written out of a sim in free LTspice

I found that when I'm playing with a circuit, I want to adjust how I'm playing in response to what I hear. If I have to record and simulate each time, that is quite limiting/annoying. With this, you can just play as if you had breadboarded a circuit.

The open source community needs a fast SPICE with a nice interface. Even if the engine is not as feature rich as others, other developers could improve it.

Thank you for your kind words, but, I suspect that this simulator is too highly specialized for performance to serve a more general purpose. For example, it would be very difficult to support mixed mode devices in my engine. But, maybe someone will surprise me!

 

[jcx]

if you really want to hear a circuit change you need DBT anyway for any but the grossest change or for thoroughly understood signal changes that need to be controlled for in the listening test - like level, frequency response matching

Sy's article is a start, available free at Linear Audio http://www.linearaudio.net/images/LA Vol 2 Yaniger(1).pdf

 

[jazbo8]

Looks like a very interesting way to test pre-amp circuits for guitar amps. Are you planning to add more tube models, especially the power pentodes? Also, what are the triode models based on?

 

[dsharlet]

More tube types and models is definitely on my todo list. Right now, they are just using the Child-Langmuir model. My next step was Norman Koren's model (Improved vacuum tube models for SPICE, Part 1) but before I can use any more advanced tube models, I need to do some major work on my computer algebra. Right now the only non-linear functions it can really deal with well are the exponentials found in diode and transistor models.

 

[jazbo8]

That's great! For the pentodes, please consider other models besides the Koren model, for more details, please see this sticky thread.

 

[dirkwright]

I haven't looked at this software yet, so pardon my ignorance, but can you change part values (resistors, capacitors, etc.) live, while the software is measuring an FFT, noise or a Bode plot or something like that?

 

[dsharlet]

Currently, transient (time domain) simulations are the only type of simulations supported. During simulation, you can change some resistors (if they are pots or variable resistors), but you can't change most components or the circuit topology. The hardest problem in doing this really isn't any simulation/analysis issue, it's just a matter of how to build the user interface to allow it.

 

[96khzorg]

Very Cool. I wonder how complex the ciruit can be until real time is not possible anymore.