[sdiy] Sanity check: equal power panner and VCA law

[Steve Lenham]

On 02/01/2019 11:54, Rutger Vlek wrote:

> I'm working on an equal power (or -4.5-dB BBC law) panning and 
> cross-fader circuit, and have been looking at this implementation using 
> THAT2180 VCAs http://www.thatcorp.com/datashts/dn120.pdf. I've adapted 
> the principle of operation from this circuit to work with the SSI2164, 
> and in Spice that works relatively well (my breadboard version is still 
> buggy). However, the error from the ideal equal-power curve is a bit 
> higher than I was expecting, about twice the error the document reports. 
> It made me think about possible improvements, and then I 
> realized.....why use an exponential response VCA (THAT2180 or SSI2164) 
> while a linear response is so much closer to the equal power curve? Out 
> of curiosity I simulated a similar circuit with a linear VCA, using the 
> same four diodes circuitry to obtain an approximated equal power curve 
> and..... the result is a much better fit!
> 
> This makes me wonder who invented the approach with the diodes, and 
> weather it was originally intended to be used with a linear VCA and 
> slavishly copied by THAT corp disregarding the gain law. Does anyone 
> know? Or do I miss something, and is there a good reasons for using this 
> with an expo VCA, even though the fit to the ideal curve is worse 
> (perhaps tolerance issues with circuit imply it's difficult to fully 
> close linear VCAs)?

The good reason may simply be that the THAT Corp VCAs have genuine 
pro-audio levels of performance, but happen to be exponential, whereas 
most of the linear OTA/VCAs still available are, performance-wise, poo.