Can you buffer a differential signal with two op-amps?

[Zipdox]

I want to use use an OPA1632 to amplify a differential signal and feed it into an ADC, but this draws current from the signal depending on the feedback resistors, which changes based on the gain configuration.

To deal with this, I want to use an op-amp configuration like this (though with two OPA1656 instead), to buffer the input before sending it to the OPA1632.

I'm worried about the input and output offset of the op-amps messing up my signal. Is this a valid concern? If so, what should I do instead? Should I pick a specific kind of op-amp? Or use a different approach altogether?

[MarginallyStable]

I think adding a matching 10K resistor in the feedback's of your voltage followers is a common way to help balance offset currents.

[TimFox]

Your two 10 k\$\Omega\$ resistors from the inputs to ground will draw current from the source:  what is their function?  Is the source truly galvanically isolated from ground?

[Zipdox]

Your two 10 k\$\Omega\$ resistors from the inputs to ground will draw current from the source:  what is their function?  Is the source truly galvanically isolated from ground?

The input will be capacitively coupled (because of phantom power). The resistors make sure the common voltage doesn't drift from ground. I think I should make them 100k instead. But this is negligible. I intend to control the gain of the differential op-amp (OPA1632) by making Rg and Rf a dual-gang potentiometer. This would mean that the current from Vin+ and Vin- would vary depending on the gain, and would thus vary the input impedence. This is why I want to use buffers.

I think adding a matching 10K resistor in the feedback's of your voltage followers is a common way to help balance offset currents.

Between the outputs and ground?

[TimFox]

If the input bias current through your input resistance is a substantial voltage error in your application, you can balance it partially by inserting an equal resistor from the output to the inverting input of the op amp.
This works if the "offset current" (difference between bias currents at the two inputs of the op amp) is much less than the bias current (typically true for bipolar op amps, not so good for fets).
However, that resistance will add thermal noise to your circuit:  presumably the two input resistors (10 or 100 k\$\Omega\$) will be "shorted out" at high frequencies by the actual impedance of your floating source.
(You never improve SNR by adding a resistor to the circuit.)

[Zipdox]

I'm considering just using an instrumentation amplifier instead. Though I need to figure out which ones are suitable for audio applications.

[srb1954]

The input will be capacitively coupled (because of phantom power). The resistors make sure the common voltage doesn't drift from ground. I think I should make them 100k instead. But this is negligible. I intend to control the gain of the differential op-amp (OPA1632) by making Rg and Rf a dual-gang potentiometer. This would mean that the current from Vin+ and Vin- would vary depending on the gain, and would thus vary the input impedence. This is why I want to use buffers.

Consider using the standard 3 op-amp differential amp circuit. This will provide the high impedance inputs you require and you can change the gain using a single pot rather than a dual gang pot.

Using a dual gang pot to change the gain of a single op-amp differential amplifier is not a good idea as you will never get good enough matching between the elements of a standard dual-gang pot to maintain good common-mode rejection over a wide gain range.

[John B]

I'm currently building a balanced mixer using INA1650 for the balanced inputs and OPA1632 to drive outputs.

[Zipdox]

Consider using the standard 3 op-amp differential amp circuit. This will provide the high impedance inputs you require and you can change the gain using a single pot rather than a dual gang pot.

Using a dual gang pot to change the gain of a single op-amp differential amplifier is not a good idea as you will never get good enough matching between the elements of a standard dual-gang pot to maintain good common-mode rejection over a wide gain range.

So basically an instrumentation amplifier? Because I'm currently looking at the INA851, which has a differential output, as well as a V_OCM pin (required for my ADC). This one IC can be the entire input stage for my application. It's a bit expensive, but it replaces multiple components so it might even pay for itself. The only concern I have is if I can find a potentiometer with a reasonable taper for my application. I need a gain from 0.5 to 600, which corresponds to a resistance between 4k and 2 ohms.
A linear pot would result in a very skewed scale. This chart is already logarithmic.

Using a reverse logarithmic pot helps, but it's still pretty inconsistent.

[David Hess]

I'm worried about the input and output offset of the op-amps messing up my signal. Is this a valid concern? If so, what should I do instead? Should I pick a specific kind of op-amp? Or use a different approach altogether?

That will work fine and it is commonly done in a 3 operational amplifier instrumentation amplifier.

Another approach uses two instrumentation or difference amplifiers which are cross coupled as shown below.  This is commonly used to convert single ended to differential however it also works as a differential amplifier.

[EPAIII]

If you are going to play with Op Amps, I strongly recommend getting a copy of the "Op Amp Cookbook" by Walter Jung. It answers so many basic questions about Op Amps and their circuits.

https://www.thriftbooks.com/w/ic-op-amp-cookbook_walter-g-jung/312633/item/614451/?utm_source=google&utm_medium=cpc&utm_campaign=pmax_high_vol_frontlist_%2410_%2450&utm_adgroup=&utm_term=&utm_content=&gclid=CjwKCAjwvpCkBhB4EiwAujULMtqSn9zjY65tFMwyhBLjrurwrUhRx9_0RXpW0FBpwBI5jZGtKa7diRoCTCEQAvD_BwE#idiq=614451&edition=5157816

and

https://www.google.com/search?client=firefox-b-1-d&q=Op+Amp+Cookbook

[Terry Bites]

You can save yoursef a fully diff amp and get some common mode rejection as an added bonus.
Use the front end from a traditional instrumentation amplifier. So simple it hurts. The ADC performs the same fucntion as the third difference amp stage would.
www.allaboutcircuits.com/technical-articles/learn-about-three-op-amp-instrumentation-amplifiers/

Its got a common mode gain of one and diff gain of what ever you like. Set with a single resistor or resistor + pot.

Even if its doesn't solve all your problems its worth keeping for another day.

[Zipdox]

You can save yoursef a fully diff amp and get some common mode rejection as an added bonus.
Use the front end from a traditional instrumentation amplifier. So simple it hurts. The ADC performs the same fucntion as the third difference amp stage would.
www.allaboutcircuits.com/technical-articles/learn-about-three-op-amp-instrumentation-amplifiers/

Its got a common mode gain of one and diff gain of what ever you like. Set with a single resistor or resistor + pot.

Even if its doesn't solve all your problems its worth keeping for another day.

So I've got three options

• Fully differential instrumentation amp IC (INA851)
• Discrete fully differential instrumentation amp using 2 regular op-amps and 1 fully differential op-amp for output
• Front end of an instrumentation amp (as you suggest)

Option 1 has the advantage of being a single solution with laser-trimmed resistors and output over-voltage clamping. But it's the most expensive, the INA851 isn't in stock on LCSC so I'd have to solder a QFN package manually.
Option 2 is cheaper, but no laser-trimmed resistors, and output clamping has to be added separately. But it can be assembled by JLC as the parts are in stock on LCSC.
Option 3 is really interesting. As I understand it, it relies entirely on the capacitive coupling and the input resistors for the common mode voltage offset. This is acceptable, as my input will be capacitively coupled. Of course I have to do output clamping myself. Are there any drawbacks? For example, does it work properly with a single-ended input signal?

[David Hess]

The 4th option is the last one I gave.  A pair of instrumentation amplifiers can be cross coupled to operate as a differential amplifier, and no matched resistors are required.

[donlisms]

I speculate a bit, but think you might find the data sheet for the SSM2019 differential preamp chip interesting as another perspective on similar issues.

[RogerThat]

Zipdox, look at reply #11 from Terry Bites, that's what you want to do assuming your ADC takes differential input. That circuit buffers and amplifies your differential signal. A dual op amp and three resistors. I use this to feed ADCs with differential signals from pressure sensors (wheatstone bridge). Gain can be adjustable by replacing Rb with a potentiometer.

Instrument amplifiers is mainly used when you want to convert diff to single ended.

[Zipdox]

Zipdox, look at reply #11 from Terry Bites, that's what you want to do assuming your ADC takes differential input. That circuit buffers and amplifies your differential signal. A dual op amp and three resistors. I use this to feed ADCs with differential signals from pressure sensors (wheatstone bridge). Gain can be adjustable by replacing Rb with a potentiometer.

Instrument amplifiers is mainly used when you want to convert diff to single ended.

I think I'll do that. Now I just have to pick the op-amps. Also, wouldn't electrolytic capacitors be problematic in this case? The negative pins go towards the op-amps because the other end can be at 48V (phantom power), but if phantom power is disabled, the capacitors will be at -2.5V.

[RogerThat]

electrolytic capacitor? Yes, they should not have reverse polarity. Ignore everything to the left of the op amps in Terrys's picture.

[Zipdox]

Ignore everything to the left of the op amps in Terrys's picture.

I need a 2.5V Vocm for the ADC though.

[RogerThat]

Ok. Terry's picture is then lacking two capacitors to AC couple the input signal.

[Terry Bites]

This twin opamp thing will only work for balanced inputs, if you grounded one input the outputs are no longer balanced.
Maybe not such a useful thing for your project.

Note that fully diff amps can be used to create a diff output with correct Vcm with one inpur grounded. So the same (but less DC accurate) function as a line driver, SSM2142, DRV134 etc
When you run them from the same rail as the ADC you need no further input protection for it.
You could go down the route attached- full flexibility and gain distribution options.
You might even use a single supply inst amp and ac couple it to the diff amp. Go all 5V.
If you can, get Vcm from your micro system. It lets you do an easy DC cal. Maybe well filtered PWM?

[Zipdox]

This twin opamp thing will only work for balanced inputs, if you grounded one input the outputs are no longer balanced.
Maybe not such a useful thing for your project.

Note that fully diff amps can be used to create a diff output with correct Vcm with one inpur grounded. So the same (but less DC accurate) function as a line driver, SSM2142, DRV134 etc
When you run them from the same rail as the ADC you need no further input protection for it.
You could go down the route attached- full flexibility and gain distribution options.
You might even use a single supply inst amp and ac couple it to the diff amp. Go all 5V.
If you can, get Vcm from your micro system. It lets you do an easy DC cal. Maybe well filtered PWM?

I'm leaning towards option 2: building a fully differential instrumentation amplifier with two discrete op-amps and one fully differential op-amp. Basically copy the internals of the INA851.