New tri-VCO design

[jhaible]

> Does anyone have a handy way to translate transistor matching specs (hfe,
> hfe matching, Vbe matching, whatever) into current-mirror errors? Don't
> remember ever seeing this.

In a 2-transistor mirror, the main error related to hfe would be from
1/hfe rather than hfe1/hfe2.

And Vbe - I guess the approx. 20mV/Oct equation would be valid here
as well, i.e 20mV difference = 1:2 current ratio, +100% or -50% error.

JH.