Download as pdf or txt
Download as pdf or txt
You are on page 1of 1

  Close


Test & Tools

+
Blog  More

OpAmp basics: unbalanced to


balanced signal with dual OpAmp

Jan Cumps 28 Aug 2022

A little OpAmp circuit to turn an


unbalanced signal into a balanced one.
An unbalanced signal is the typical signal
coming out of a signal generator,
consumer audio equipment, headphone
jacks.
A balanced signal uses two separate
wires, with the same signal 180° inverted.
It's used in RF, and professional audio
equipment. It has some advantages, e.g.
allows for less noise. But that's not the
topic of this post.
What I show here is a common circuit to
convert from unbal to bal.

lmant14 project14

I did this to test the little breakout board I


made for a TI OPA2170. It's a dual
OpAmp, and I wanted a neat circuit with
two OpAmps to test it.

The circuit

I took an example from tina.com: 10.


Amplifiers with balanced Inputs or Outputs.

source: tina.com Amplifiers with balanced Inputs or


Outputs

It's a very simple circuit, where the output of


the first OpAmp is identical to the input
signal - but buffered. The second OpAmp is
an inverting buffer and delivers the input
signal, 180° phase shifted.
Together, these two signal form a buffered
rendition of the input signal, with two times
the amplitude.

I used a TI OPA2170 dual OpAmp. It's a low


power device, with decent noise specs.
Bandwidth up to 1.2 MHz. On the breakout
board that I made, I soldered two 1 µF
decoupling capacitors.

I used 5 10K resistors. One for the input,


two in parallel for the 5K resistor, and the
remaining ones for the two 10K resistors
of the circuit.
This design is high impedance input. The
input resistor can be made much higher
than the 10K I used. In my case it wasn't
critical. The unbalanced input comes from
a function generator. So I just took the same
value as all other resistors.

I built the circuit on a breadboard. Good


enough in this case, where I use a 5 KHz
signal.
в06

08
§Z
02

I used a balanced lab supply, set it to +9V /


-9V. The function generator (the unbalanced
source signal) is connected between ground
and input, set to 5 KHz, 250 mVRMS, 0 V
offset.

Probing a balanced signal

It's easy to probe an unbalanced signal.


Connect the probe ground to ground, probe
pin to the signal, and you have it.
A balanced signal runs over two wires, and
none of these are ground. So you can't put
your probe ground on one part, and probe
pin on the other*.
A solution is to use two oscilloscope
channels, and probe each of the two signals
with one signal, as usual. Then use the
oscilloscope's MATH function to show the
difference of the two.

You see the result in the scope capture


above. Channel 1 (yellow) is the unbalanced
input - the signal coming out of the
generator.
Channel 2 (cyan) the output of the left
OpAmp. It's the bufffered input signal, and
forms one part of the balanced output.
Channel 3 (magenta) is the output of the
right OpAmp, and is the inverted component
of the balanced output.
The difference between the two is the
balanced signal. I used the Math (purple)
difference (A - B: channel2 - channel 3)
function to show it. Because both balanced
halves have the same amplitude as the
source - and are 180° inverted, the total
balanced signal  has double the amplitude
of the input.

One of the properties of a balanced signal is


that the inversion removes noise. External
noise will impact both components of the
balances signal almost the same. When you
then take the difference, identical noise on
the lines cancels itself out, while the signal
keeps its form. 

*you can. I leave it to fellow community members to put


possibilities in the comments

victorjoseteixeira
Offline  3 members are here

Leave a comment...

Paragraph    

Comment

Oldest Best Newest

genebren 3 months ago


Online
Another informative blog! I really like
the way that break down and clearly
explain your experiments.  That looks
like a handy protoboard design.

Jan Cumps 3 months ago in reply to


Online
genebren

That looks like a handy


protoboard design.
I got the little board from shabaz.
He had hidden it, as a little
surprise, in a packet of goodies he
sent me.

shabaz 3 months ago in reply to Jan


Cumps

I didn't recognise it initially! I


should have mentioned, the
holes are offset, but if you
solder on an inner row on one
side, then it fits the
breadboard perfectly (I did it
that way to save some
space). I think you've
probably already figured that
out, I couldn't tell from the
birds-eye view.
genebren the proto board files
are here if you need any (I
would send you some but I'm
low on them currently too, I
need to order more). The
export-soic-only.zip file are
the most useful ones, but the
export-tssop-only.zip can be
handy too.

Jan Cumps 3 months ago in reply to


Online
shabaz

if you solder on an inner


row on one side, then it fits
the breadboard perfectly (I
did it that way to save some
space). I think you've
probably already figured
that out
Yes. While soldering, I used a
breadboard to hold the
connector, then put them in
the PCB rows where the
connector was a good fit,
perfect vertical.

shabaz 3 months ago


Nice tower of test equipment!
My sig-gen doesn't do balanced either,
but I've been using a transformer,
although it's then very frequency-
dependent : ( 
There's a single-BJT option which is ok-
ish, but again is frequency dependent
and level dependent. It needs
capacitors on inputs/outputs:

Jan Cumps 3 months ago in reply to


Online
shabaz

In theory, the OpAmp design


starts at DC. 
I'm away for work next week (first
trip in 3 years - Warsaw). When
back, I may try and build a
LabVIEW design to sweep and
measure ...

DAB 3 months ago


Offline
Nice back to basics post.

jc2048 3 months ago


Offline
Hello Jan. Did you notice the voltage
noise plot in the datasheet?

Above 10kHz, the noise starts going up


(that's an unusual feature). That will
add quite a lot of high-frequency noise
to the outputs. If you assumed the
noise stopped at the 1MHz limit of the
graph (it doesn't), that still looks like at
least 70uV or 80uV of noise into the
load with the two op-amps (in practice it
could be a couple of hundred uV). Not
really a problem for you experimenting
with the circuit, but it might be a bit of a
concern in some applications,
depending on what you were doing with
it.
BTW be cautious of the spice model.
Although they've characterised the
noise, it looks like the high frequency
end, above 10kHz, is somewhat
different to the datasheet. In a
simulation of your circuit, I get a total of
2mV of noise into the load. Let's hope
it's the datasheet that's accurate and
not the model.
A second thing to note is that these op
amps can only drive 300pF, so watch
out if you are intent on attaching long
cables to the outputs.

Jan Cumps 3 months ago


Online
I leave it to fellow community
members to put possibilities in the
comments
Alight. They are asleep.
One alternative is to use a differential
probe. This probe is isolated from
ground.
It can measure the signal
between two test points that aren't
ground.

I've attached the two inputs of the


differential probe to the output of
OpAmp A and B. Those are the points
that are also probed by
oscilloscope CH 2 and 3.
In the capture you can see that the dark
blue differential probe signal
(oscilloscope CH 4) is virtually identical
to the MATH signal (calculated
difference between CH 2 and 3). But
cleaner:

In the capture below, the original signal


and the differential output are
compared. The output has twice the
amplitude of the input. In the capture
they show with the same hight,
because I set CH 1 to 500 mV/div and
CH4 to 1 V/div.

jc2048 3 months ago in reply to Jan Cumps


Offline
Who's making all that noise? You
woke me up.
A third possibility, for low
frequencies and voltages, is to do
your own differential input
amplifier with an op amp.
Something like this would get you
started

That's from Analog Signal


Processing by Ramon Pallas-
Areny and John G. Webster. It's a
nice book. I'm sure all the material
in it could be found on-line, as
application notes etc, if you knew
where to look, but it's good to
have it all collected together in a
well-structured way. I wouldn't be
able to justify the cost of the new
book, now, but I got this one
secondhand for a few pounds and
for that it was an absolute bargain.
The beautiful seed case is from a
Hollyhock (Alcea).

Jan Cumps 3 months ago in reply to


Online
jc2048

The Art of Electronics runs


this one too. With little
explanation though (that's
typical for that book).

tangent: the image above is


taken from the book with a
tablet camera, then filtered by
Microsoft Lens. Almost as
good as a flatbed scan.

jc2048 3 months ago in reply to Jan


Offline
Cumps

Here's why parts are offered


with closely matched resistors
and why The Art Of... shows
them to you.

This is the circuit twice, using


the sim model for the op amp
you chose. U1 has all the
resistors exactly the same,
U2 has one of the resistors,
R8, out by 1%. Voltage
sources VS3 and VS4 are
each 0.5V and give 1V
between the inputs at all
times. I've got a voltage
generator then sweeping the
common point between them
from -10V to +10V over 20
seconds (slow enough that
there won't be any
complications from dynamic
effects). Ideally, the outputs
will sit on 1V the whole time.
Here is the result.

With the resistors perfectly


matched, the output is out by
500uV, which is an offset
error from the op amp, and
there's a slight slope, which is
the common-mode error
contributed by the op amp. (If
you needed the offset to be
better, you could either test
your op amps, to see if one
was particularly good, or you
could pay more for a superior
one with better specs.)
With the second circuit, you
can see that there's a
considerable problem with the
common-mode error as it
moves away from zero.
(Although I didn't investigate
it, I think that there will also
be a problem if it's
temperature cycled -
essentially it's a form of
bridge, but once the
symmetry goes, the
temperature variation in the
resistances would now no
longer cancel.)
How much it's a problem
depends on what you're doing
with it. In your case, your
driver outputs are individually
referenced to ground, so it
would work better for you
than for situations where the
common-mode voltage can
move away from zero.

Jan Cumps 3 months ago in reply to


Online
jc2048

I did a naïve attempt - just


build the single difference
amp circuit and compare
output to input.

Engagement

Author: Jan Cumps

Date Created: 28 Aug 2022 4:32 PM

4097 views

15 likes

13 comments

Blog Post Actions

 opamp

Like  15

Group Actions

Members 〉

Learn 〉

Technologies 〉

Challenges & Projects 〉

Products 〉

Store 〉

About Us Feedback & Support FAQs Terms of Use

Privacy Policy Legal and Copyright Notices Sitemap

Definições de cookies

An Avnet Company © 2022 Premier Farnell Limited. All Rights


Reserved.
Premier Farnell Ltd, registered in England and Wales (no 00876412),
registered office: Farnell House, Forge Lane, Leeds LS12 2NE.

Test & Tools requires membership for participation - click


to join

You might also like