TL071/TL081

[Martin Czech]

:::I thought the 8x had a higher impedance on the non-inverting input than the
:::7x. Or did someone already mention that? Or am I just confused?

Ok, now I was really curious and I looked again into all the spec sheets:

1. The top metal mask of the 051 is much different from the 081.
   So we can say the 051 is a new design. It also has zener zapping
   for offset trim (one can see the additional pads for that).
   This explains the "high" cost.
   
2. The 071 has almost the same typical spec parameters as the 081.
   The typical noise specs are the same , as are power consumption
   and bandwidth. So I was wrong!
   The 071 data sheet has no top metal mask picture.
   The only apparent difference is input bias current.
   My guess: the 071 differs slightly from the 081, but it is a minor change
   in the input stage (geometry, implant or biasing). The difference
   is so small that it won't be observed in a -say- 10x microscope,
   so no additional top metal mask photo is needed for people who want
   to get dice.
   
   Since the change was made for lower noise, it is a bit strange that
   typical noise figures did not change.
   This indicates to me that noise is not tested.
   Normally you might get lower noise, but it is not guaranteed.
   That is the way I read it.
   It could also be an error in the spec, that happens, too.
   Sometimes reading a spec sheet is like "reading between the lines"
   as we say here.
   
Anyway, the possibly improved noise spec of the 071 will only be
visible with low impedance source (and also low Johnson noise
resistors). This is the region which is really a problem for FETs
due to their high input noise voltage. OTOH their input noise current
is good, so high impedance sources might benefit from that.
The "breakp point" is about 100k to 1M, depending on noise specs.
1/f noise is not mentioned (or I did miss it). FETs usually have
difficulties there as well. No problem for audio, but for high gain dc.


m.c.