I encountered some problems when simulating the input noise (RTI) of an op amp.
(The simulation model of AD8428 was downloaded from the ADI official website. )
QSPICE gave a strange simulation result, compared with LTspice.
In addition, the result calculated by using the command “.meas noise RtiRms Integ V(inoise) from 0.1 to 10” in QSPICE is inconsistent with the result calculated in Waveform Viewer.
I like the operation of QSPICE very much, which can greatly improve efficiency, but these two problems are very troubling to me.
I want to know where the problem is. Is there something wrong with my settings?
#The output noise looks the same, but the result calculated by using the .meas command is very different from the result obtained by operating in Waveform Viewer.
After a long time, I finally find out the mistake I have made on this issue. I had forgive to uncheck the ‘Fast (less accurate) math’ box in the Preferences setting. Just like the name says, the low-precision mode made the results wildly off. Don’t be like me and miss that tiny checkbox…
Initially, I failed to recognize the discrepancy between the results generated by .meas integ and the integral calculation in the waveform viewer. After reviewing KSKelvin’s simulation results and conducting further research, I discovered that I had used an incorrect method to compute noise.
As illustrated in the figure above (excerpted from Analysis and Measurement of Intrinsic Noise in Op Amp Circuits Part II: Introduction to Op Amp Noise by Art Kay, Texas Instruments (TI)), I revised the calculation expression accordingly. This adjustment successfully aligned the simulation results with those displayed in the waveform viewer.
I just took a look at the LTspice help, which specifies that AVG, RMS, and INTEG in .noise analysis for .meas are defined differently as to other analysis types. It appears that Qspice does not currently adhere to this definition, which causing the discrepancy between the two with same .meas statement for .noise analysis.
I just found a post where Mike provided an example of calculating RMS noise using two separate .meas statements: the first one squares and integrates, and the second one takes the square root. As what you shown.
