How to you guess the open-loop output impedance?

I'd like to do some feedback analysis of op-amps using Mathematica, i.e.
doing those bode plots of open loop gain and beta, etc. This is to
check for stability and other stuff. Problem is that most spec sheets
don't give the open loop output impedance. What can I do?

 

Why? You are usually far better using an electronic simulation program
for this sort of job. Some actually have hooks to do loop gain plots
directly, i.e. no fiddling about

Use a good spice model, guess, measure, or calculate from the known
circuit.

Typically open loop output impedances are 1-200 ohms.

Kevin Aylward

http://www.anasoft.co.uk
SuperSpice, a very affordable Mixed-Mode
Windows Simulator with Schematic Capture,
Waveform Display, FFT's and Filter Design.

 

Typically the open-loop output impedance would be something like
looking into the emitter of a Darlington with its base having 10-30pF
on it.

I've never tried modeling it that exactly, but you certainly have
broached an interesting question!

...Jim Thompson

 

Hello all,
this has been true for the non rail to rail amplifiers.
The RRIO and RRO amplifiers have the drains or collectors
connected to the output. That's the reason why they have
an open loop output impedance of a few hundred ohms to one kOhm.

Best Regards,
Helmut


RRIO: Rail to Rail Input and Output
RRO: Rail to Rail Output

 

Jim Thompson wrote...

Or often a single-transistor emitter follower. Following your
suggestion, that's something like Zo = beta * Xc = beta / 2pi f C,
but at the higher frequencies beta = f / f_T, so that in the end
Zo looks more or less resistive. E.g., we get 40 ohms for 20pF
and an output transistor f_T = 200MHz. And even tho we're being
rather crude, we should still add on r_e, which can be another
20 to 100 ohms, depending on the class-AB quiescent current.

It's when we put this resistive Zo inside the feedback loop that
the opamp's overall Zo looks like an inductor, as Alan will see.

 

Open drains or collectors would give open-loop output impedances well
in excess of 1K, UNLESS there was a local loop around the output
stage.

Does anyone own a network analyzer? It would be interesting data to
see open-loop output impedance versus frequency for various popular
OpAmps... under no load, sinking current and sourcing current
conditions.

...Jim Thompson

 

I think the open-loop output impedance of 4558 type amplifiers at
moderate frequencies is surprisingly high. Had trouble with this years
ago with low-level amplfiers in extremely noisy situations


Best regards,
Spehro Pefhany

 

Are we defining "open loop" as having the compensation circuit disabled or
just breaking the external feedback loop.

With compensation enabled:

The output devices are typically either driven by the last point that the
compensation circuit encloses or are them selves enclosed in the loop so
i'd expect:

With an op-amp like an LT1498, I'd expect the output impedance to look
somewhat capacitive at lowish frequencies. At, lets say about 100Hz, the
impedance will start to look almost purely resistive and somewhere in the
50 to 100 ohms range. As the frequency gets near the top of the range of
the op-amp, I'd expect to see a slight swing into the inductive side of
the phase curve.

With the very high speed op-amps, I'd expect the output to not start
looking resistive at a higher frequency. The swing into the inductive
side to be either missing or very slight.

 

Jim Thompson wrote...

My HP Agilent network analyzer starts at 300kHz. I have an old HP
network analyzer that goes from analog frequencies to 13MHz, but its
digital interface is rather obsolete. So now I'm looking for eBay
to present me with an HP 4192A 5Hz-to-13MHz vector impedance analyzer
at an affordable price sometime soon. If not, I may have to consider
the $3.5k to $8.5k asking price in the surplus-instrument markets.

In the meantime, I may be able to employ my 10kHz to 10MHz HP 4275A
LCR meter for some of these tasks, with an appropriate test fixture.

 

Paul Kiciak, N2PK, has an interesting VNA design using a DDS and a
novel method of measuring phase. Here is some info

"This is a homebrew VNA capable of both transmission and
reflection measurements from 0.05 to 60 MHz, with about 0.035 Hz
frequency resolution and over 110 dB of dynamic range. Its
transmission measurement capabilities include gain/loss magnitude,
phase, and group delay."

"Its reflection measurement capabilities include complex impedance
& admittance, complex reflection coefficient, VSWR, and return
loss."

http://users.adelphia.net/~n2pk/index.html

The main interest for rf work is extending the high frequency
capability, but it might be possible to go lower in frequency.

Mike Monett