INA826 as 10x instr. amplifier start deviating above 2V output

[dusan]

[ Specified attachment is not available ]I'm using INA826 (datasheet https://www.ti.com/lit/ds/symlink/ina826.pdf) instrumentation amplifier with 5k1 gain resistor (gain = 1 + 49400/5100 = 10.68x and 105.1mV on vref pin to get away from the bottom rail):



Up until 2V on the output everything is fine and output exactly matches calculated expected value. But above 2V it starts deviating. You can see it in this chart:



Notes:
- i tried 100k as a path to ground as datasheet suggest but it had no effect (the grounds of supplies are connected anyway so it doesn't matter)
- individual inputs are 525mV 275mV, difference is 250mV
- 100mV vref (actual value 105.1 remains constant during entire measurement), I separately measured it and it holds 100mV even at 47ohm load so it's not problem

So I can see 2 possible issues:
- Is something wrong with my schematic or am I making some wrong assumption?
  For example 0.525*10.6 = 5.56 which is more than rail 4.5V, is this a problem?
- my instrumentation amplifier is bad?

Update:
- I did another measurement  where I removed R7 so that voltages on inputs * gain never exceeds rail and it clips even sooner

 ina826_chart2.png (29.4 kB - downloaded 7 times.)

[Alex Nikitin]

You would need a negative supply for the INA826 in this configuration. If you look carefully at the internal circuit of the chip, you would see that one of the first stage opamps outputs will hit the ground half-way. The first stage is trying to produce a differential output to the second stage and if there is not enough headroom you'll get what you see in your circuit.

Cheers

Alex

[dusan]

Thanks.

Symmetrical version works: