2
\$\begingroup\$

I'm designing a RIAA pre-amplifier based on the 2 op-amps from an NE5532 package. It seems that something is wrong, because the output of the 2nd stage is giving a 400V output which is way above the power rail voltage. Is it correct?

The 1st voltmeter, XM1, is measuring the input voltage; the 2nd voltmeter, XM2, is measuring the output voltage of the 1st stage; and the 3rd voltmeter, XM3, is measuring the output voltage of the second op-amp, across 100k R7.

Circuit without the voltmeters.

Circuit with the voltmeters.

\$\endgroup\$
5
\$\begingroup\$

If it is SPICE's AC analysis it's not surprising.

In AC analysis, SPICE figures out the small-signal linearized model of the system, then applies the input voltage to that model, assuming it is linear. There is no "resonableness" checking, or any attention paid to nonlinear effects of any sort -- and an op-amp hitting the rails is certainly a nonlinear effect.

It is your responsibility, when using AC analysis, to verify that your system behavior is consistent with a linearized model. If it is not, then you need to use transient (i.e., full-blown nonlinear) analysis.

Try a transient analysis with the same input -- you should see massive clipping on that second stage output.

\$\endgroup\$
7
  • \$\begingroup\$ Will try that, thank you, but I shouldn't expect anything near nor above 400v at the output of the 2nd stage upon implementing and measuring the real thing, right? \$\endgroup\$ Jul 17 '19 at 23:17
  • \$\begingroup\$ I think you are right but I would say it a little differently, blaming the model rather than the simulator. A SPICE simulator uses whatever model it is given for an op amp. There may be many different models available for a given op amp, and some may do a better job of modeling this kind of real-world behavior. Most simulators also provide an "ideal" op amp that is just a textbook kind of model with no concept of real limitations. \$\endgroup\$ Jul 17 '19 at 23:21
  • \$\begingroup\$ In Real Life, the output of an op-amp cannot go outside the supply voltages. If the gains are such that the output would theoretically be 400 V, the output signal will acutally be clipped at less than +/- 15 volts. \$\endgroup\$ Jul 17 '19 at 23:25
  • \$\begingroup\$ That's what I thought Mr. Bennet, it can't go over the rail voltage. \$\endgroup\$ Jul 17 '19 at 23:27
  • 2
    \$\begingroup\$ I think you are all misunderstanding the nature of SPICE's AC analysis. It's not about Real Life. It's about making a linear model, and then assuming that the model is accurate. Real life contains no linear systems, so any analysis involving linear systems needs to be verified after the fact for whether it is realistic. \$\endgroup\$
    – TimWescott
    Jul 17 '19 at 23:27
0
\$\begingroup\$

Well, I just figured out what was wrong. The gain of the 2nd stage was indeed about 40, which means 400V at the output, and as pointed out by TimWescott, simulators don't do a "resonableness" check making "possible" an over-the-rail output.

Also, another mistake was that RIAA pre-amps are used for MM cartridges which do provide no more than 5mV, nothing like ceramic ones, 300-500mV. I also found out the pre-amps for ceramic cartridges don't require RIAA equalization and generally have a unity gain, they just act as an impedance matcher.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.