6 Feb
2012
6 Feb
'12
9:33 p.m.
On Feb 6, 2012, at 3:53 PM, Pete Turnbull wrote:
Right. You can see the effect if you look up 2N3904/3906 datasheets and
compare; the 3906 has similar characteristics to the 3904, but "worse" in
several ways because they've resized the device to have similar gain
characteristics to the 3904. For example, comparing Fairchild's datasheets
side by side I see the 3906 has slightly higher typical gains by about 10%,
but the Ft is 250 MHz instead of 300, probably mostly due to the increased
capacitance from the larger junction area, while Vce_max is lower on the
3906 presumably due to the shorter base region length (or heavier doping,
it's hard to say).
- Dave
Nor am I an
expert, but that's pretty much it. In CMOS ICs, the typical
practice is to double the channel width of the P channel to compensate for
the difference. This roughly equalizes the current of the two halves, but
the capacitance for the P transistor's gate is doubled because the gate
area is doubled. I don't recall how it affects bipolar devices, though I
imagine it's the same.
The end effect is reduced gain, sometimes accompanied by lower fmax. Nowadays a factor of
about two in hfe, but in years gone by it wasn't unusual to see
apparently-complimentary transistors with more than a factor of ten difference.