Electronics > Repair

Replacement for old Intech or Fairchild op-amp.

<< < (8/8)

Kleinstein:

--- Quote from: magic on July 24, 2021, 11:20:12 pm ---You need to learn how to read TI marketing. When they say "FET input" without "J", they mean "it frankly is CMOS, but we don't want to scare the audiophools away".

That kinda explains the flicker noise ;)

--- End quote ---
The OPA140/145/827/1641 are labeled JFET and they are very low flocker noise. The low flicker noise is an exceptional point of these OPs.

The OPA810 is a bit exotic, as a hybrid rail to rail with a JFET and a CMOS input stage:  CMOS when the input is very close to the positive supply.

magic:
TL072 is noisy but also labeled JFET. All truly JFET parts from TI are labeled JFET because they are JFET ;)

You are right about OPA810, it's slightly unusual. It's the first time I see something like that, so when I read "FET" I assumed 100% MOS input because it's a common euphemism with those guys.


--- Quote from: David Hess on July 24, 2021, 06:44:42 pm ---The OPA810 datasheet shows a completely different multiple-path topology

--- End quote ---
That's just your ordinary RRI configuration; two complementary diff pairs and a comparator which switches (gradually transitions, really) their tail currents depending on input common mode. More elaborate multi-path configurations are used in some of their CMOS devices where an auxiliary input stage provides feedfowrard straight to the output stage, bypassing the phase shift of multiple gain stages used in the normal path for high gain at DC or audio. I don't know what is the supposed advantage of this architecture over the capacitive feedforward more common in older designs, but the block diagrams are very unambiguous that this is how they do it.

The subsequent integrator connected right to the output is typical for RRO and they don't tell you what's inside. It could be a single stage identical to the outputs of OPA14x/209/211 or a two-stage configuration with Miller compensation around the whole lot and internal feedforward - also like 14x/2xx.

You may want to read this; I'm almost sure it's about National LMC660 which appears to have become a kind of "classic" RRO design. Those TI opamp block diagrams look very similar.
http://home.engineering.iastate.edu/~hmeng/EE501lab/TAHelp/MonticelliRailToRailOutSwing.pdf

David Hess:
I misunderstood what they were doing.  I am well acquainted with the dual complementary input stage for rail-to-rail input operation.  I do not remember anybody ever using different devices types though, although it would not surprise me if someone tried it with JFETs and PNPs in the distant past.

I did the math and the 0.1 to 10 Hz noise for the OPA810 comes out to 2800 nV pp so despite relatively low broadband noise, the high corner frequency precludes low frequency precision applications.  The positive common mode 0.1 to 10 Hz noise comes out to an incredible 70,800 nV pp, as some would expect from a fast CMOS process.

Navigation

[0] Message Index

[*] Previous page