DISTORTION ANALYSIS

Overview

Distortion analysis determines the level of distortion harmonics produced by the circuit under test. These can be either the 2nd and 3rd harmonics of a single fundamental or the intermodulation products of two test signals.

Distortion is created by non-linearities in a circuit's transfer function. A circuit comprising linear components only (resistors, capacitors, inductors, linear controlled sources) will not produce any distortion. SPICE distortion analysis models distortion for diodes, bipolar transistors, JFETs and MOSFETs.

Correct modelling of distortion for circuits involving IC macro models is not guaranteed in any way, because these models may well use linear controlled sources to model basic device behaviour only.

Similar information can be established using a Fourier analysis, but the Distortion analysis is also able to show how the distortion varies as the fundamental frequency is swept.

Method of Computation

This analysis is based on the small signal (AC) models for the devices in the circuit so the first step is to compute the operating point. Each non-linear device model then contributes a complex distortion value for the appropriate harmonics, dependent on how much the device is seeing of the input fundamental. The extent to which these harmonics appear at the output determines the values plotted. The process is repeated across the specified range of input frequencies. In fact, the mathematics of this analysis is extremely complicated and involves the construction and manipulation of Taylor series to represent the device non-linearities.

Note that complex values are used, so the analysis yields information about both the magnitude and phase of each harmonic.

Choosing the Mode

Which of the two analyses is performed is decided by the IM ratio field on the Edit Distortion Graph dialogue form, and by nothing else. Leave it at zero and the analysis is harmonic distortion of a single fundamental. Enter a non-zero value and it becomes an intermodulation analysis, with the value taken as the ratio of the second frequency (F2) to the fundamental (F1).

For single frequency harmonic distortion, two curves are produced for each trace on the graph, one for the 2nd harmonic and one for the 3rd.

For intermodulation distortion, three curves per trace are displayed, showing the intermodulation artefacts at F1+F2, F1-F2 and 2F1-F2.

Some care is needed in choosing the ratio F2/F1, since mathematical oddities can otherwise occur. For example, if F2/F1 is 0.5, the value of F1-F2 is F2 and the value of the F1-F2 plot will then be meaningless as it coincides with the second fundamental. In general you should try to choose irrational numbers for this ratio. F2/F1=49/100 would be a much better choice.

F2 is intended to be the lower of the two frequencies, so the ratio should be less than 1. The value is not checked by the simulator, so a ratio of 1 or more will simulate without complaint but will not give meaningful results.

Using Distortion Graphs

Distortion graphs show the magnitude of harmonics on the left axis and the phase (normally of less interest) on the right axis. The test frequency (F1) is plotted on the x-axis.

The swept range of the test frequency is set by Start frequency and Stop frequency. As with a frequency graph, Interval chooses whether the range is divided into DECADES, OCTAVES or LINEAR steps, and No. Steps/Interval sets how many points are computed within each interval. A reference generator must also be nominated, in the Reference field, to tell the simulator where the test signal is injected.

For Harmonic Distortion analysis (one input frequency F1) two curves are plotted per trace, one each for the artefacts at 2F1 and 3F1.

For Intermodulation Distortion analysis (two input frequencies F1 and F2) three curves are plotted per trace showing artefacts at F1+F2, F1-F2 and 2F1-F2.

In either case, which curve is which may be established by placing a cursor on the graph: the curve you are pointing at will be identified on the right hand side of the status bar.

See Also:

Fourier Analysis

Noise Analysis

Frequency Analysis

Cursors and Measurements