PNP Matched Pairs

[Grant Richter]

Hi All,

There are a number of exponential converter designs available.
The designs based around the CA3046 work well and have the
advantage of convenient thermostating.

I personally like the "classic" Electronotes designs using the PNP
matched pairs. While this requires a Q81 type tempco, the low
power requirement and short warm-up time fit the live performance
aspect of designs.

One of the problems with this design is the PNP pairs themselves.
AD821s are obsoleted and matched 2N3906s, while they work fine,
are not optimized for this application.

I was pleasantly surprised to find that Analog Devices appears to be
packaging the MAT-03 fall outs as a different part number.
MAT-03s are the preferred choice, but with prices quoted at US$13.98
each in 100 pieces is hard to bear. The SSM2220P part number
appears to be a relaxed spec version of the MAT-03s. Quoted price
is US$3.19 ea in 100 quantity. Quite a savings.

The specifications are only slightly relaxed from the MAT-03s.
Hfe matching is speced at 3% max rather the 2% of the MAT-03.
Hfe in general in speced a few points lower (60 vs. 75 worst case),
and other parameters are similarly "wider". The important point
is the same 0.3 ohm Rbe for log conformance. Much better
than the 1.5 ohm Rbe of the LS358s that I have been using.

Thought this would be of general interest.

Grant