TINA/Spice: Tina THD analysis

Rotem,

There are two factors at work here that are affecting the sim results; one is the model and one is the simulator.
The model's text header gives a long list of parameters the model conforms to.
Unfortunately HD is not on the list, so any HD the model has may not agree with the data sheet.

The other factor is the simulator. Setting up the simulator takes a little care when you are simulating HD.
The default TINA max time step size is variable over the simulation. This is shown in the Simulation Analysis Parameters as
the TR Maximum Time Step with a default setting of '10G'.
This, in effect, limits what the Fourier Analysis can offer because it has to interpolate between the variable time points and this 
throws off the resulting values of the Fourier series coefficients.
The fix for this is to set the max step size to a fixed value so the simulation gives a series of evenly spaced time steps over one period
Note the Fourier analysis is applied to one cycle of the signal, in this case 1kHz.f the output signal.
To get the most accuracy out of the Fourier analysis, the time step should result in the same number of points over a single cycle of the signal are used in the Fourier Analysis. For your circuit, this was 4096 points, which you can get with a step size of 244.140625ns.

The attached TINA schematic used your circuit as a starting point.
The first step was to run a Fourier analysis using the existing settings without a signal to show the equivalent noise floor of the simulator.
This looks to be < -150dB below the signal, so there is enough dynamic range (the datasheet lists HD as -136dB).
Then a sim was run with the signal and the results show the HD to be greater then -120dB, which does not agree with the data sheet.
So the model with the default TINA analysis settings doesn't agree with the data sheet.
These are the left-most plots in the attached schematic. At this point, we don't know if the HD is due to the model or to the simulator settings.

The next step was to set the max step size to a fixed value of 244.140625ns (right-most graphic in the schematic) and re-run the sim using 4096 points in the Fourier Analysis.
Note the total sim time was increased to 10ms; the Fourier analysis was applied to the last cycle of the signal to minimize the impact of any start-up transients.
The dynamic range (w/o a signal) increased to more than 260dB, much more than we need to resolve any HD components the model may have.
This is the upper right-hand plot embedded in the schematic.
So these simulator settings look okay.
The sim with the 1kHz signal (lower right-hand plot) shows only odd-ordered harmonics with a max of -160dB relative to the desired signal.
So it appears the model has better HD than is given in the datasheet, which is consistent with the fact the model does not try to conform to the datasheet for HD.

I hope this helps. Its kind of a lot to go through, so please let me know if you have any questions.

Regards,
John

OPA1611 THD circuit_rev1.TSC