[sdiy] SSM2220 dual PNP clones?

[Ian Fritz]

Unfortunately, something has been missing in this discussion, which is why 
you even need matching in the first place.  The purpose of using a matched 
pair is so that the exponential prefactors -- i.e., the reverse leakage 
currents --  of the two devices (including their temperature dependences) 
cancel.  Matching of the Vbe's is not necessarily required for this, 
although it is a fair indicator.  I measured a large batch of transistors 
once and found that matched betas usually correlated fairly well with 
matched Vbe offsets.

The idea that matching is required for an accurate expo response (good 
tracking) is absolutely wrong, although we see this myth repeated over and 
over, even by people who should know better.  Good log conformance comes 
from high-quality material and low Rbe.

For inexpensive PNP pairs, the good old 2SA798 does remarkably well, 
provided Rbe compensation is included in the converter.

   Ian