Group Actions
Engagement
  • Author Author: Jan Cumps
  • Date Created: Date Created
  • Views 20398 views
  • Likes 11 likes
  • Comments 17 comments
Related
Recommended

OpAmp basics: balanced to unbalanced signal with difference amplifier

Jan Cumps

A little OpAmp circuit to turn a balanced signal into an unbalanced one. This is the circuit that  shows in my previous article.
It's a very simple circuit, discussed in many books (Analog Signal Processing by Pallas-Areny and Webster , The Art of Electronics by Horowitz and Hill). Jon proposed it as a possible solution to probe balanced input on an oscilloscope - for low voltage, low frequency use. I'm going to test this, with one of the OpAmps in the same OPA2170 package I used in the previous blog.
image
A very good application note by Harry Holt of Analog Devices puts some reality to this seemingly simple design. Why doesn't it perform as well as the textbooks say. How well-matched do the resistors have to be? And how does (wanted or unwanted) capacity work with this?
I'm ignoring all of that, and use 5% resistors and a breadboard with uncontrolled stray capacity. The goal is to just get the circuit running. But also to learn the limitations of the naïve approach.

The circuit

The concept is very simple. An OpAmp that can accept a balanced signal on its + and - inputs. At the output, it serves the same signal unbalanced, referenced to ground.
I've used 33 kΩ resistors. 5%, but from the same batch. At this point, I haven't checked how well matched they are, but given that they are 5%, that isn't relevant.

As balance input, I used an isolated function generator, 520 mVRMS, 1 kHz sinus. The generator has to be galvanically isolated from power supply and oscilloscope. A battery or power brick powered one usually works. If the generator output is grounded, it will not be able to generate a balanced signal.

Probing

Probing the balanced input signal

There's a chicken-and-egg situation at the input - for this blog only, not in real life. This is a circuit to measure balanced signals, and render its unbalanced representation. But if you put a normal probe at the input, the probe ground will disturb the balanced input. As suggested in the previous blogs, there are several solutions to measure that (one of them being this circuit). I'll use a differential probe to measure that input signal.
image

Probing the unbalanced output

That's easy. The output is unbalanced (ground referenced). So you can just use a common oscilloscope probe, with the probe ground connected to the circuit ground.

image

Channel 1 is the input, probed with the differential probe. Channel 2 the output with a common x10 probe.
The purple math channel is the difference between output and input. 

Parents

  • Hi Jan,

    Your blog post reminds me of a technique popular with ham radio enthusiasts, who often need to convert from unbalanced to balanced (and vice-versa), but at frequencies from tens of kHz to tens or hundreds of MHz (or higher). They use 'baluns' which get built in different ways depending on the optimal frequency range and power capability. I wrote a draft blog post on them a while back, mainly a lot of theory, but never published it, It had no construction or testing of anything as such since I didn't get around to that. I don't know if it's worth still publishing, although I will not get around to writing up more on that topic for a while due to other stuff getting in the way : (

  • in reply to shabaz

    At 5kHz, an audio transformer should work ok as a balun. For sine waves, even repurposing a mains transformer might do.

    Here's a very old mains transformer (240V in, 12-0-12 out) - it's so old that the windings are wound on top of each other rather than on separate bobbins - and that passes a sinewave fine at 5kHz.

    image

    I'm driving it 50 ohms from a cheap function generator. The output centre tap I've offset by 1.5V relative to the 'scope ground with a battery, to show how it floats. (The red trace is the calculated difference between the two channels.)

    image

    There's about 1% difference in amplitude between the two outputs. Hardly precision stuff, but ok for experimenting, or functional testing, or fault-finding where you just want to see waveforms present.

    It doesn't do so well with other waveforms, though, because of all the inter- and intra-winding capacitance. Here's a triangle and square wave. Yuck!

    image

    image

  • in reply to jc2048

    Wrong! If I then add a load resistor to each output (this is with 680R), I get nicer waveforms without all the ringing.

    image

    image

  • in reply to jc2048

    That's decent. I have a few higher frequency transformers here, from a design challenge and from scavenging PCBs. I'll do some tests too....

Comment Children
No Data