# voltage follower question

Discussion in 'Electronic Basics' started by tempus fugit, Apr 17, 2004.

1. ### tempus fugitGuest

Hey all;

I don't know if there's a general answer for this question, but here goes.
I'm reading up on voltage followers (noninverting opamp), and it says that
the gain of a voltage follower is approximately 1. Is the gain a little less
than 1 or a little more than 1, or is there no straight answer for this
question?

The reason I ask is that in the same textbook it says that for instability
to occur, the gain must be greater than 1. I'm looking at using a high speed
opamp to buffer the output of a signal generator, and I don't know if it
will oscillate or become otherwise unstable if I configure it as a voltage
follower.

Thanks

2. ### John LarkinGuest

Simplistically, the gain is G = g / (1+g)

where g is the open-loop opamp gain, typically 20,000 to several
million at low frequencies, rolling off as F goes up. So G is slightly
less than 1. At very high frequencies, more complex stuff can happen,
and G can possibly get above 1.
Most opamps are unity-gain stable, so work as followers, but some
minority are undercompensated; the datasheet has details. An external
capacitive load - even a healthy hunk of cable - will make most opamps
oscillate; a series 50 ohm or so resistor between the opamp output and
the load will generally fix this. National has some cool "C-load" amps

John

3. ### tempus fugitGuest

A buffer provides a very high impedance input and a low impedance output.
This way, the opamp is not loaded by the incoming signal, and the outgoing
signal doesn't load the next stage. In my application, I want to use a pot
to control the output level, but using one large enough to do the job would

4. ### electrickedGuest

I'm a newbie and this is a question. What's the purpose of a buffer? I know
it can delay the output for a while but other than that how is it used? What
applications does a buffer have?

Thanks!

--Viktor

5. ### tempus fugitGuest

My recollection of the specifics of this stuff is a little weak, but I'll
try my best.

Generally speaking, you want to have a low impedance output going into a
high impedance input. I don't really remember the math, but if you have the
reverse, you end up with a voltage divider situation and your signal becomes
weakened. This effect is also frequency dependent, in that lower frequencies
(I think?) will be affected more than higher frequencies. This effect is
known as loading. If you have a matched output to input Z, then you have
maximum power transfer. The long and short of it is that if you are feeding
a high impedance signal to a low impedance input, the signal will be

I also recall something about wanting to have a signal feeding into an
input that is 10x higher in impedance for audio, but, again, I don't remeber
exactly why. I think it is because of the fact that impedance itself carries
a frequency component, so if you have a range of frequencies (say 20 to
20kHz, as in audio), the impedance will be different at each of the
different frequencies, being higher at lower frequencies.

6. ### tempus fugitGuest

At very high frequencies, more complex stuff can happen,
and G can possibly get above 1.

Thanks John. By very high frequencies, do you mean 20MHz by any chance? This
is the upper limit of my function generator. If I might end up with
problems, I could always just go with a buffer amp, but they are so much
more expensive than a normal high speed opamp.

Thanks again

7. ### electrickedGuest

So how is the impedance important for input and output? What happens if you
have low impedance? What about high impedance?

If it's high impedance output, does that mean that you can drain a lot of
current from the output? I see how high impedance input would help, being
that the device requires less current and therefore less power dissipation
and I would think with high impedance the device would work much easier with
higher frequencies. Does that make sense?

Any help is strongly appreciated!

Thank you!

--Viktor

8. ### Rich GriseGuest

You need to go to whoever or wherever you heard the word "impedance" and
find out what it means.

Good Luck!
Rich

9. ### electrickedGuest

I know ac impedance is the equivalent of dc resistance due to components
such as caps and inductors and I know that the impedance changes with
frequency. What I don't know is how this is used in everyday practical
circuits. Please shed some light on my understanding.

Thanks!

--Viktor

10. ### John LarkinGuest

Near the opamp's Ft (frequency where gain falls to 1) some opamps
start to have additional phase shift, more than the ideal 90 degree
lag. Around there, any small additional phase shift in the circuit can
make trouble. Phase donations may also come from capacitive loads, or
from the use of a resistor in the negative feedback path (don't do it,
except for current-mode amps) or from non-ideal power rail impedances.

A medium-fast, say 100-200 MHz, opamp in a tight layout with good
power bypassing should be a good 20 MHz follower and not be too
twitchey. Opamps like this are cheap nowadays. *Really* fast opamps
(1-10 GHz GBW) are harder to work with.

John

11. ### MantraGuest

This is basically correct.

The key thing is that everything has an "incident" impedance, that is,
"looking" into a terminal of XYZ, you (or rather a connected circuit)
will "see" an impedance and the resulting voltages and current will
respond accordingly (via Ohms Law, Kirchhoff's Voltage and Current
laws). This means impedances and impedance matches affect how your
overall circuit behaves.

Things like digital rise times, power efficiency and all depend on
how the impedance of one circuit interacts with the impedance of
another circuit.

Typically an engineer reduces a bulk circuit down to a simple voltage
or current source and an impedance also known as Thevenin or Norton
sources, respectively (for example your stereo amplifer could be
modeled thus). Similarly a complex circuit load can be reduced to a
you do this source model and load model are combined to form a circuit
with just a source and two resistors: very easy to understand and to
manipulate with just Ohms and Kirchoff's laws (in their AC forms).

There are all sorts of things you can calculate with just this simple
circuit. For example if your speaker if a subwoofer, then the
impedance will "conjugate match" your source impedance at low
frequencies so you'll get maximum power tranferred, which is why
subwoofer do what they do. If you knew the impedances of the source
and load you could predict both the frequency response of each
component and the total response of your amplifier and speakers when
connected.

12. ### MantraGuest

The thing with voltage followers is that:

1) They are low output impedance - damping of oscillations requires
resistance which can imply high impedance, so there can be times when
low impedance could pose a problem. This is the first clue.

2) You do need *total loop* gain to be greater than one *and* a
multiple of 360 degrees of phase shift to get oscillation. The
voltage follower's "unity gain", which is actually <1, is only a
*forward gain* but it pretty close. Further it nearly meets the other
criteria due to *forward* phase by being 0 degrees shift. All you
need is a little *reverse* feedback.

3) They can have non-ideal, non-trivial output impedance. Not just a
simple Thevenin source with a resistor, but an impedance. For
example, emitter followers (bipolar transistors operating as voltage
followers) can have net inductive output impedances at high frequency,
which can cancel Miller capacitances that normally band-limit voltage
impedance current feedback path. These two feedback paths could give
enough total loop gain for oscillation.

Op Amps, however, are a bit easier to deal with. To determine if your
Op Amp will be unstable you need to look at its Bode Plot and see what
the gain margin and phase margin are. That's the quick, cheap
analytic method. The other is the empirical method - pick something,
try it, if it oscillates add compensation to tweak the phase back to
stability as per the manufacturer's app note.

13. ### tempus fugitGuest

Thanks again John. I'll order up a couple of 150MHz opamps and see how it
goes.