"new" dual transistors

[Martin Czech]

The recent discussion made me (finally) aware of "new" dual transistors,
may it be japanese or other stuff. In fact this was discussed back in 97,
but I forgot about it.

Maybe it is not such a bad idea to review my quiet, but drifting
multiplier, the use of such dual transistors should give lower drift
for at least one input (two inputs would mean quadruple npn, which we
have not).

Btw.: would it be suitable to use a quadruple MOSFET array for the
multiplier core? I think so. AFAIK the core operates in the "linear"
region anyway, the exponential bipolar properties are avoided. It is
harder to use MOSFETS and bipolar on one chip, but discrete things are
no problem.

Another variant could be to take the usual differential loging input
stage (discrete) for the 1496 multiplier. The 1496 has a brother 1495,
which has such a stage for better linearity. For some reason I don't
understand the 1495 is not availble (is this so?).

If you don't understand what I mean:  look into the schematics of old
AD multipliers, or the 1495/96 data sheet or Tietze/Schenk, 8th Edition,
p. 337 or http://www.geocities.com/SoHo/Museum/4459/diy/ringmod.html.

One can develop back such a four quadrant multiplier to a two quadrant
type, we call this vca. Example: http://www.uni-bonn.de/~uzs159/vca.gif.

Wouldn't it be interesting to use a loging input stage, or a MOSFET pair
for/in the transconductance stage? Rene?

As Juergen Haible said, OTAs are nice, but you have up to three (not
very good) current mirrors in the path. (O well, after so many years I
have finally realized this, isn't it fantastic ;->).

Same applies for OTA usage in triangle vco.

m.c.