I'm trying to do noise simulations for the AD8220 in LTspice and my noise numbers seem to be a bit off compared to the datasheet.
For example, if you load the LTspice test fixture for the AD8220 and run a noise analyis (make sure to set AC amplitude to 1 for Vin), the noise appears to be higher than the datasheet would indicate. I've even set the gain and load resistors to be noiseless, and that didn't make a significant difference.
For example, the test fixture is set to a gain of 10. According to formula and specifications in the datasheet (page 3), the input referred noise voltage should be SQRT(eni^2+(eno/GAIN)^2). This comes out with a value of SQRT(14^2+(90/10)^2)=16.6 nV/rtHz.If you look at Figure 10 on page 10, the RTI voltage noise looks to be just under 20 nV/rtHz.
If you simulate the RTI noise (plot output noise and divide trace by "gain"), you get 25.2 nV/rtHz at 1 kHz, and if you make the gain resistor noiseless you get 23.7 nv/rtHz. I'd be surprised if the plots in the datasheet assumed a noiseless gain resistor, but even then, the noise in the simulation seems to be off by quite a bit.
Is there a reason that the noise is higher in the simulation? Is the simulation or the datasheet correct?
And just for comparison, when I do the calculations for the AD8421 INAMP, the hand calculations come within 4 % of the spice simulation. With the AD84220 they are off by 42%!