divider values for diff out for inamp

I found this circuit to provide a differential output to an in-amp in a datasheet for one of Analogue's in-amps. I'm actually using Microchip's MCP6N11 in-amp, which uses 2 gain setting resistors. I think the choice of opamp in the circuit isn't very important.

Microchip don't publish a simulation for their inamp, so I just used any old one I found in LTSpice.

The gain setting resistors for the MCP6N11 are connected from output to Vfg, and Vfg to Ref, so in this circuit they will also appear to be in parallel with R3 and R2.

The gain setting resistors will be 4k32 and 7k68 (if I can get these values) to give a gain of 1.5625 (so 1.6v in will give 2.5v out), the Vfg and Ref inputs of the MCP6N11 are high impedance, so the load on the in-amp will be 12K in parallel with R2 and R3. The in-amp itself has an output impedance of 900R.

So my question is, R2 and R3 should be equal, but what value should they be? Highest I can get away with? Or is there some optimisation I can do here?

Look at LTC6362. Diff in and diff out. Linear has many parts like that one. Some slower, Some faster.

I need high impedance inputs and low bias current, so a diff i/o amp is out of the question, since it looks like two inverting amplifiers to the source circuit. The MCP6N11 has 10pA bias, lower than most inamps I've seen from other mfr's. OTOH, it has terrible offset voltage at low gain, but OTOOH it has an auto-calibration feature. (And I can only look at it's datasheet in small doses because it makes my eyes go funny)

The thing that's most annoying about this is I'll be switching out an existing diff i/o amp (THS4130) to connect this one in. The only alternative I can see is use one of these: https://www.linear.com/product/LTC1043 on the input of the existing diff i/o amp to get the high input impedance, but then I'd still need to change the gain of the diff i/o amp, and it seems more messy to switch it's gain setting resistors.

Both amps have a power down feature, so still trying to decide if I can use this instead of actual switching - if I can then the extra opamp will need it too.

So anyway, I guess I have to choose the highest values I can based on the opamp's bias current, and the output impedance of the inamp? (since the load will be an adc and hence very high impedance)

I could just tie one leg of the adc's input to ground, but the extra opamp is needed to set the Ref pin voltage anyway, so it's trivial to make this modification.

Of course, better suggestions are as always most welcome!

The MicroChip instrumentation amplifier IC is Cmos so its input reasistance is extremely high.
Its has TWO REF inputs, a (-) and a (+).
You copied the circuit of a DIFFERENT instrumentation amplifier IC with only one REF input that will not work with this one.
Most of the circuits in the MicroChip datasheet do not even have an opamp that you have.

audioguru said:

You copied the circuit of a DIFFERENT instrumentation amplifier IC

Click to expand...

Did I? Gosh, I never noticed!

If you can look past the different style of gain setting for a minute, and the fact that it's a completely different device, you will see that the inverting input to the output stage of the MCP6N11, which you call an inverting Ref input, is only used for negative feedback - if I did have the correct component in the simulation it would not be shown connected to the additional opamp at all. Microchip call it Vfg. As far as I can see, for most external purposes it can be ignored unless you want to use it for something else. This is why I think I can use the circuit shown.

Sorry it's confusing asking people to make the mental substitution of one for the other when looking at the drawing - I'm happier with pencil and paper, but it won't do simulations for me!

Beyond it's gain setting function, the Ref input is used in exactly the same way as any other inamp. The gain is set by the ratio of the two resistors, Rf and Rg, plus 1. So although Ref will be relatively low impedance due to the gain setting resistor being connected to it, which I do need to bear in mind, this shouldn't matter since the extra opamp will have low output impedance.

So anyway, without the means to test the real circuit, which I won't have for a few months at least, my original question still stands, unless someone can show me a better way to keep the very low input bias current, and also get a differential output. (Unless I abandon the differential output, but where's the fun in that?)

Sure it is confusing when you post the wrong schematic. You could have pressed "Print Screen" from the datasheet then saved it as a schematic in Microsoft Paint program.
Most instrumentation amplifier ICs use a single resistor to set the gain like an AN620.