ina214 gain error questions

Hi Frank,

Thanks for the questions on INA214...with regards to question number 1:

Using the equation I calculate a gain of 52.63.  I suspect there actually may be an error of 2x in the denominator error term (due to an approximation in teh equation simplification)...if I factor that in I would get a gain of 35.7...still nowhere near your reported gain of 5...it might be good to know more about how you measured your gain exactly.

With regards to question number 2: you have inserted 1k Ohms in series with a circuit network.  Please refer to Figure 24 in the data sheet....in that figure there is a block called "bias"...in the equations that follow Figure 24 that help estimate the error due to inserting a small series resistance the "bias" block is assumed to have a simple impednace of 1250 Ohms...this is a simplification as the "bias" block is formed from the sources of CMOS transistors that make up the device input stage.  Typically 28uA of current flow into both ends of that bias block with zero differential input voltage is applied, so any mismatch in your 1K resistors will react with that bias current and creat an offset voltage....in this case with 1% resistors I guess it could be as much as 280uV additional input offset voltage that will have a large drift component.  Also the internal resistors ratio match very well to set the gain = 100V/V for the INA214, but the absolute tolerance of the internal resistors is not guaranateed.  Given all these factors it is not too much of a surprise that you would see some amount of variability.  I have to recommend not to use such large value resistors in series due to these issues, so can you reduce the resistor values and increase the capacitor values (I am assuming you are using a filter here)?

Dear Ed,

First of all, thanks for the reply.

It is quite interesting to work on the INA214 and we have found the fundamental problem with our application. Due the small value of the shunt resistors, the PCB track and layout become quite essential. Now, we have come to realise that the input resistors need to be placed as close as possible to the shunt, well, very close! the tracks on the board will affect the shunt resistor voltage, which is generally in less than 10 mV with our application. Anyway, we modified the board and it did output around 100 gain mark ( little variation). The input resistor 1K also did affect the differential voltage across V+ and V- and dropped the gain down to around 55. We also tested that with input resistor 1K, INA214 will still output a gain of 100 across V+ and V- pins.However, the reduction in gain relative to voltage across shunt is caused by the voltage drop on the 1K resistor.

I am not sure if our tests are valid in general, but it works for us.

Best regards

Frnakie