Spice is a powerful tool for circuit analysis. But it often surprises users with
. holes in its abilities and strange definitions
. lack of industry-wide standards
. the need for the user to model the physics of their circuit, not just draw the schematic

I hope this blog will educate users and promote discussion in these areas.

Distortion analysis Limitations

I am just going to skim this large topic.

Spice is basically very good at calculating a circuit's distortion.
see the "Large Signal versus Small Signal simulation" post since both the calculation method and the meaning of the distortion numbers varies among Spice programs.

BUT  it usually doesn't work out

1. Traditionally Distortion in Spice is a "small signal" analysis. However for most real world nonlinear circuits, small signal results understate the distortion at medium to large waveform amplitudes.

2. Whether your Spice program uses a "small signal" or a "large signal" method to calculate distortion, the following limitations apply.

Many circuits use op amps and other IC's. These are modeled using subcircuits. While the IC manufacturer has a complete Spice subcircuit for their device, they don't give that to the public. Instead we get a dumbed down subcircuit. Typically an op amp that has 60-100 transistors is represented by subcircuit with 5 transistors and some of Spice's "behavioral modeling" parts. These subcircuits do not try to model distortion and several other important device behaviors.

Further, the differential transistor input pair is matched as only Spice can do - a perfect match (both transistors are identical).

All this means that simplified subcircuits are not going model the real part's distortion. We will get incredibly low distortion numbers.

Even models of specific transistors, power MOSFETs and tubes/valves may not be precise enough to show the correct distortion, though I believe they can be, if fully modeled. After all, distortion is in the small deviations of the part from ideal. The model must be super accurate to capture that correctly.

Wish it was otherwise.

If you have one or two critical parts, you can try asking the IC manufacturer for a copy of the full model. I knew an engineer who claimed to have gotten access to one. he and his company signed a non-disclosure agreement.

There are a few exceptions
perhaps a dozen part models from TI and Analog Devices do say they model distortion - this is usually noted directly in the subcircuit listing file. Open the file to check.

No comments:

Post a Comment