CFB amplifiers at high gain

[Winfield Hill]

CFB op-amps are famous for not loosing their
wide -3dB amplifier bandwidth as you increase
gain, as do VFB op-amps. Many of us know this
aspect degrades at high gains, but not as well
as we should. There's a deadly parameter, Rb,
which is the input resistance of the "-" input.
This ranges from 8 ohms to 500 ohms. For low
Rb, transistors have to run at high currents.

Rb competes with low Rg for feedback current,
forcing one to use high Rg and Rf values to
get high gain. But high Rf means poor slewing
and bandwidth. Most CFB op-amp formulas don't
mention or include this critical Rb parameter.
If you force Rg too far below Rb, oscillation
or severe peaking and ringing can occur.

This can mean that if one needs a stage gain
of 20 to 100x, a VFB op-amp can compete with
a CFB type for the job. Today I updated my
op-amp spreadsheets, taking Rb into account.
Only a fraction of manufacturers give us the
information we need; most content themselves
describing their op-amps for use at G=1 or 2.
But much can be surmised by studying curves.


--
Thanks,
- Win

 

[Winfield Hill]

John Miles, KE5FX wrote...

On September 8, 2019, Winfield Hill wrote:

Only a fraction of manufacturers give us the
information we need; most content themselves
describing their op-amps for use at G=1 or 2.
But much can be surmised by studying curves.

Are you making a distinction between inverting and
non-inverting configurations?

The issue is pretty much the same for inverting or
non-inverting gain, but for high gains with low Rg,
it's a pain to drive a super-low Rg inverting input,
so non-inverting is usually the way to go.

It's true that the available gain starts to roll off at higher
frequencies for whatever (parasitic?) reasons, but I've never
run into a case where reducing Rg by itself causes peaking or
instability. (I've also never built an inverting amp with
these parts, or tried to go beyond Av=20 in a single stage.)

Some informal measurements: http://www.ke5fx.com/ths3491.htm

Other parameters like the output swing capability get much
worse at high frequencies, but I'd assume that has more to do
with the output stage than anything happening at the + or -
inputs.

Your THS3491 is the fastest high-voltage part in my table,
with Rb = 15 ohms, so you're less likely to run into the
issue. You'd probably need Rg below 10 ohms to see it.


--
Thanks,
- Win

 

[John Miles, KE5FX]

On Sunday, September 8, 2019 at 7:07:02 PM UTC-7, Winfield Hill wrote:

Only a fraction of manufacturers give us the
information we need; most content themselves
describing their op-amps for use at G=1 or 2.
But much can be surmised by studying curves.

Are you making a distinction between inverting and non-
inverting configurations?

It's true that the available gain starts to roll off at higher
frequencies for whatever (parasitic?) reasons, but I've never
run into a case where reducing Rg by itself causes peaking or
instability. (I've also never built an inverting amp with
these parts, or tried to go beyond Av=20 in a single stage.)

Some informal measurements: http://www.ke5fx.com/ths3491.htm

Other parameters like the output swing capability get much
worse at high frequencies, but I'd assume that has more to do
with the output stage than anything happening at the + or -
inputs.

-- john, KE5FX