[sdiy] CA3046 vs. CA3083 and matched pair madness

[Aaron Lanterman]

I've got a whole bunch of CA3046s and CA3083s to pass out tomorrow. When 
is one better than the other?

For VCO design, Ian favors the CA3083.

I have these notes on my site:

CA3046 five NPN, Q1 and Q2 have emitters tied, substrate is tied to one of 
the other emitters - used in many classic Electronotes VCO designs, and 
the Serge PTO ($0.45/each/25 Jameco)

CA3083 five independent NPN with a separate substrate pin; Q1 and Q2 are 
"matched at low current" (a favorite of Ian Fritz; he says they have "low 
Re and very good matching (way better than spec.)... still my favorite for 
NPN expo converters (use transistor #3 for HF tracking)") ($0.89/each)

And... what about a cheap array of matched PNPs? There seems to be a gap 
there. A lot of my students are using OTA designs; to drive them 
exponentially I've generally been handing out 2SA798s, since I happened to 
come across a bunch of them at a reasonable price a while back, but that 
doesn't seem like a permament solution. Also, PNP pairs like the MAT03 and 
SSM2220 seem like severe overkill for something like a filter. Of course, 
not that that would stop Paul S. from doing such a thing. ;)

There's various bits of wisdom in the archives, but I thought it would be 
good to get the current thinking of the group.

- Aaron