[music-dsp] Non-linear processing/Volterra series?

[robert bristow-johnson]

> ----- Original Message -----
> From: antti at smartelectronix.com
> To: "music-dsp" <music-dsp at ceait.calarts.edu>
> Subject: Re: [music-dsp] Non-linear processing/Volterra series?
> Date: Thu, 2 Mar 2006 18:28:57 -0500 (EST)
> 
...
> AFAIK the emitter current equation holds over a very large range, so Ie 
> should be close to real device while not in saturation.

it's because the emitter-base junction is forward biased in normal operation (meaning the current resulting from that bias is large) and the collector-base is reverse biased in normal operation (meaning the current resulting from that bias is small but there is lots of current "collected" from the emitter junction).  there is a smalled effect of changing collector voltage on the emitter current, but it's small (h_oe and h_re are small).

> 
> > so you have several little black boxes to analyze and model.  the
> 
> The choice of what is included in a black box is where I disagree with you. 
> IMO only the tubes should be considered black boxes and not rest of the 
> elements (which do have good analytical models).

but (in my proposition) as long as we're measuring the grid to anode transfer characteristics for each stage, why not, since the signals are already recorded, look at (experimentally) the transfer charachteristics from the plate of stage N to the grid of stage N+1?

> For measuring tubes you're much better off by measuring the V-I curves 
> directly, instead of based on signal within the amp (tubes are seated in 
> sockets so you won't need to dissect the amp to measure them).

well, as you said, building a "jig" and getting the right (high) voltages applied to measure those curves in an environment different from where the tubes are used might be difficult in practice.  i think far more difficult than opening the amp box and attaching multiple probes to key points and recording the signals simultaneously.

> Modeling each stage and some limited inter-stage interaction.

that, we agree on

> Again, I disagree that you have to measure anything in the actual circuit
> for building the model

that, we disagree on.

> - you have tube data (measured previously), you have models for all 
> the passive components, rest is just mathematics.

no, there is also experimental error because the parts don't behave exactly as they are expected to.

> In-circuit measurements are IMO best left for the final comparison and 
> possible minor tweaking.

the (multiple simultaneous) in-circuit measurements are for breaking down a large complex problem into many, much smaller and hopefully less complex problems and accuarately analyzing and modeling them.

--

r b-j                  rbj at audioimagination.com

"Imagination is more important than knowledge."