[sdiy] low rbe tranies?

[ASSI]

On Monday 15 September 2003 01:16, Paul Perry wrote:
> The 'classic' exponetial circuits are based on the assumption that
> matched low rbe pairs are available, an assumtion that while still
> true, is not as true as it was 40 years ago.
> Is it possible (at the price of some added trimmers) to make an
> 'unmatched' version?

The "standard" way to deal with transistor mismatch (from looking at 
some introductory textbooks) is to add degeneration resistors, which 
makes the matching properties of the circuit depending on the resistor 
matching rather than the transistor matching (this includes thermal 
matching as well). Matched resistor pairs are available (for instance 
from Vishay), but are just as expensive as matched transistor pairs, so 
most discrete circuits I've seen just add the occasional trimmer here 
and there (this would be a place where you don't want any low quality 
part). With modern metal film resistors it should be possible to get to 
the <500ppm/K range for temperature drift, at least if you compensate 
the transistors first-order with a diode drop (just like the linearized 
input of an OTA). The downside of degeneration is that the linear range 
may be noticeably smaller and there is some added noise.

Sorry, I don't have any ready-made circuit for you, but I'd guess that 
some of the earlier expo circuits from Moog, EMS and the like should 
use degeneration in conjunction with hand-matched transistor pairs. 
Early transistors probably had enough series resistance on all their 
terminals that some degeneration was included for "free".


Achim.

-- +<[_Q+_Matrix-12_WAVE#46_microQkb/Omega_XTk/30_sonic_heaven_]>+ --

Factory and User Sound Singles for Waldorf Q, microQ and rackAttack:
http://homepages.compuserve.de/Stromeko#WaldorfSounds