Directly couple singly supply op amp buffer

[Stretto]

When directly coupling a single supply non-inverting op amp buffer there are
issues that seem to be impossible to fix.

If you simply hook up the source to the op amp without any coupling then any
signal swings below V- of the op amp will be clipped. One can remedy this by
"lifting" the source by biasing it's ground. This requires a resistor in
series, at least in some cases(not sure if all). But doing this attenuates
the signal.

Is there any way to direct couple an AC signal to a single supply op amp
without having these issues(either clipping or attenuation)? I need to
maximize headroom(hence bias it half way between the rails) and have no
attenuation(as to reduce noise reduction. Simply amplifying the signal after
the attenuation won't work).

Maybe there is some configuration of negative feedback that can compensate
for the attenuation resistor in a way that doesn't increase noise? Or maybe
there is another way to directly couple the source that avoids these issues?

 

[John S]

The usual way is to use a summer amplifier:

So, a winter amplifier won't work?

 

[Jamie]

When directly coupling a single supply non-inverting op amp buffer there
are issues that seem to be impossible to fix.

If you simply hook up the source to the op amp without any coupling then
any signal swings below V- of the op amp will be clipped. One can remedy
this by "lifting" the source by biasing it's ground. This requires a
resistor in series, at least in some cases(not sure if all). But doing
this attenuates the signal.

Is there any way to direct couple an AC signal to a single supply op amp
without having these issues(either clipping or attenuation)? I need to
maximize headroom(hence bias it half way between the rails) and have no
attenuation(as to reduce noise reduction. Simply amplifying the signal
after the attenuation won't work).

Maybe there is some configuration of negative feedback that can
compensate for the attenuation resistor in a way that doesn't increase
noise? Or maybe there is another way to directly couple the source that
avoids these issues?

lift your common with a Divider for the AC input? This will work fine
if your circuit has some kind of isolated supply.

Jamie

 

[Stretto]

"Fred Bloggs" wrote in message

When directly coupling a single supply non-inverting op amp buffer there
are
issues that seem to be impossible to fix.

If you simply hook up the source to the op amp without any coupling then
any
signal swings below V- of the op amp will be clipped. One can remedy this
by
"lifting" the source by biasing it's ground. This requires a resistor in
series, at least in some cases(not sure if all). But doing this attenuates
the signal.

Is there any way to direct couple an AC signal to a single supply op amp
without having these issues(either clipping or attenuation)? I need to
maximize headroom(hence bias it half way between the rails) and have no
attenuation(as to reduce noise reduction. Simply amplifying the signal
after
the attenuation won't work).

Maybe there is some configuration of negative feedback that can compensate
for the attenuation resistor in a way that doesn't increase noise? Or
maybe
there is another way to directly couple the source that avoids these
issues?

The usual way is to use a summer amplifier:
Please view in a fixed-width font such as Courier.


..
..
..
..
.. V+
.. |
.. --[2R]-+-----
.. | |
.. | |
.. | |
.. | | \
.. IN>-----[R]--+-----------|+ \
.. | | >--+-> OUT=IN + V+/2
.. | ---|- / |
.. [2R] | | / |
.. | | | |
.. | | --- |
.. | | |
.. +--[R]--+---[R]----
.. |
.. |
.. ---
..
..

This would work. The only thing I'm worried about is the noise/nonlinearity
that now exists. The gain's depend on the source impedance. Using a large R
will reduce it but then will add noise. My source impedance can vary a great
deal from 10k to < 10M. I guess it's a trade off that is potentially better
than the other methods but I'm not sure how much.

 

[Stretto]

"Dave Platt" wrote in message

When directly coupling a single supply non-inverting op amp buffer there
are
issues that seem to be impossible to fix.

If you simply hook up the source to the op amp without any coupling then
any
signal swings below V- of the op amp will be clipped.

Or worse. Some op amps (mostly older ones) suffer from an abrupt
"jump all the way to the opposite rail!" problem if you push the
inputs more than a short distance outside of the voltage rails... this
can lead to a nasty latchup.

One can remedy this by
"lifting" the source by biasing it's ground. This requires a resistor in
series, at least in some cases(not sure if all). But doing this attenuates
the signal.

Is there any way to direct couple an AC signal to a single supply op amp
without having these issues(either clipping or attenuation)?

One way is to work back through the circuit and force the previous
stage's common-mode reference point to be near the middle of the op
amp's common-mode range. However, doing so simply pushes the whole
problem/question back one stage in the circuit... it doesn't eliminate
it.

You could stick a stack of diodes between the previous stage output
(cathode end) and the op amp input (anode end) and use a pull-up
resistor to V+ at this point. This will probably introduce some
significant distortion, though.

I need to
maximize headroom(hence bias it half way between the rails) and have no
attenuation(as to reduce noise reduction. Simply amplifying the signal
after
the attenuation won't work).

So, just how much noise are you concerned about, and why? Can you
quantify it for us?

If you stick (e.g.) a 1k resistor between the previous stage and the
op amp input, and hang a 1k resistor from this input to V+ to bias it,
you'd suffer a 50% (6 dB) loss of gain due to this biasing
arrangement. Simply double the gain in the op amp stage to compensate.
With modern low- noise low-distortion audio amplifiers, I really doubt
that the added thermal noise from these resistors is going to be even
vaguely significant... your incoming signal will have been through
*far* worse, earlier in its chain-of-processing.

The other (obvious) approach is simple - buy one high-quality
audiophile-grade coupling capacitor, and AC couple the signal. For
not a lot of money you should be able to buy a cap which doesn't
introduce distortion anywhere near the threshold of audibility... once
again, the signal you're reproducing will have suffered from much
greater degradation before it ever saw your circuit!

----------------

I am trying to get away from using a capacitor. This is a VLF app(around
1Hz). The goal is to add as little distortion and noise as possible within
reason and constraints. Since this is the first amplifier in the chain it is
much more important to reduce the noise than further down the line. The
source impedance has significant variation which can cause significant
problems in biasing. Of course all this is expected to run on a battery with
significant lifetime(not minutes or days but weeks to months).

The only problem I am having is the input biasing issue. Fred's idea of a
summing amplifier works but I'm a bit concerned about some of the
noise/non-linear distortion introduced which are inversely related(I can
trade one off for the other). The Johnson noise will almost surely not be a
problem but I've not done any noise analysis on the circuit to know for
sure.

 

[Jamie]

"Dave Platt" wrote in message



Or worse. Some op amps (mostly older ones) suffer from an abrupt
"jump all the way to the opposite rail!" problem if you push the
inputs more than a short distance outside of the voltage rails... this
can lead to a nasty latchup.



One way is to work back through the circuit and force the previous
stage's common-mode reference point to be near the middle of the op
amp's common-mode range. However, doing so simply pushes the whole
problem/question back one stage in the circuit... it doesn't eliminate
it.

You could stick a stack of diodes between the previous stage output
(cathode end) and the op amp input (anode end) and use a pull-up
resistor to V+ at this point. This will probably introduce some
significant distortion, though.



So, just how much noise are you concerned about, and why? Can you
quantify it for us?

If you stick (e.g.) a 1k resistor between the previous stage and the
op amp input, and hang a 1k resistor from this input to V+ to bias it,
you'd suffer a 50% (6 dB) loss of gain due to this biasing
arrangement. Simply double the gain in the op amp stage to compensate.
With modern low- noise low-distortion audio amplifiers, I really doubt
that the added thermal noise from these resistors is going to be even
vaguely significant... your incoming signal will have been through
*far* worse, earlier in its chain-of-processing.

The other (obvious) approach is simple - buy one high-quality
audiophile-grade coupling capacitor, and AC couple the signal. For
not a lot of money you should be able to buy a cap which doesn't
introduce distortion anywhere near the threshold of audibility... once
again, the signal you're reproducing will have suffered from much
greater degradation before it ever saw your circuit!

----------------

I am trying to get away from using a capacitor. This is a VLF app(around
1Hz). The goal is to add as little distortion and noise as possible
within reason and constraints. Since this is the first amplifier in the
chain it is much more important to reduce the noise than further down
the line. The source impedance has significant variation which can cause
significant problems in biasing. Of course all this is expected to run
on a battery with significant lifetime(not minutes or days but weeks to
months).

The only problem I am having is the input biasing issue. Fred's idea of
a summing amplifier works but I'm a bit concerned about some of the
noise/non-linear distortion introduced which are inversely related(I can
trade one off for the other). The Johnson noise will almost surely not
be a problem but I've not done any noise analysis on the circuit to know
for sure.

if you're using a battery for this device, why can't you use a virtual
ground ? it's simple and would give you a basic +/- rail from a single
battery, this is assuming that this battery is only operating your
device here?

http://www.newark.com/texas-instruments/tle2426clp/ic-precision-virtual-ground-to/dp/88K1249

Just an example.

Jamie

 

[Stretto]

"Jamie" wrote in message

if you're using a battery for this device, why can't you use a virtual
ground ? it's simple and would give you a basic +/- rail from a single
battery, this is assuming that this battery is only operating your
device here?

http://www.newark.com/texas-instruments/tle2426clp/ic-precision-virtual-ground-to/dp/88K1249

Just an example.

-----------------------

Battery or not it doesn't seem to work. I've tried using an op amp version
to split the rails but it doesn't work. When connecting the source it draws
a ton of current. I've also tried using a resistive divider to split and it
doesn't work.

I don't know why I can just split the rails and treat the virtual ground as
ground and not bias anything but it seems it's more complex than that? That
is, V - 0 - -V is not the same as 2V - VV - 0 where VV is virtual ground?

Then again I may have had something hooked up wrong. If simply splitting
the rails would let me treat the circuit as if I had a dual supply then life
would be so much easier. I'll try again and see if I can get it to work. I
may have left something connected as if it were single supply.

 

[Stretto]

..
..
.. V+
.. |
.. .-------+-----.
.. | |
.. [220K] |
.. | |
.. 1.5u | | \
.. IN>--||------+-------------------|+ \
.. - + | | | >--+-> OUT=IN + V+/2
.. [100K] | ---|- / |
.. | | | | / |
.. | | | | |
.. | | | --- |
.. | | | |
.. | | '--[200K]--+
.. | | |
.. | | |
.. | | 0.33u |
.. '-------+----------||------'
.. |
.. |
.. [220K]
.. |
.. |
.. ---
..
..