1. Home
  2. Electronics Design
  3. Spice simulation

Spice simulation

I am using LTSpice III to simulate a filter using an LT1739 opamp. The filter is a two pole Multiple Feedback type set up for 4 kHz corner freq and a gain of 10. When I run the simulation it oscillates (in spite of the fact that the opamp is supposed to be stable at gains of 10 or more, but I guess the gain is not 10 at the higher freqs). So I have added a resistor and capacitor across the + and - input as indicated in the data sheet for the LT1739. That takes out the oscillation just fine. It appears that the cap value can be anything above about 2 pF to work and I will likely use 15 or 20 pF.

When I use 4 kHz, 1 Vp to drive the input, it simulates ok. When I use 8 kHz, I get other oscillations. Playing with the component values I find I can stop these oscillations, but I still see a few perturbations. At a few time points, I see an abrupt spike in the output signal which then rings and damps out. I don't see any rhyme or reason to when these spikes occur and when it was oscillating, it would start at different times, as much as 80 mS into the simulation.

What the heck is going on? The spikes don't seem to come from the input, an ideal sinusoidal voltage source. I only see the spike and oscillation in the output and feedback points. If I graph the output minus the inverting input to the opamp, it shows that the output is leading the input, so that seems to be the source. How can the opamp model produce this spikes and why would it be related to the input frequency?

Should I be concerned about this in a real circuit? The original component values stopped the oscillations due to a lack of compensation. But I don't know if I need to worry about these "spike" related oscillations and the component values that stop them seem to be chaotic and not predictable. That makes me very concerned as any parasitics could easily make the "spike" oscillations reappear in a real circuit.

Reply to
rickman
Loading thread data ...

Try reducing the maximum time step in the simulation. Some of these things may not be real.

You can make a parallel R-C oscillate if you set the sim parameters right.

John

Reply to
John Larkin

The minimum open loop gain should be at the order of G x Q^2. For the normal operation, you need about x10 more gain then minimum. Do you have that with LT1739?

[...]

[...]

This could be a Spice issue. Try reducing the timestep. This could be a model issue. Try different opamp. You may run out of the open loop gain. Modify the circuit. However all of those problems may also indicate a general lack of stability of the circuit.

Vladimir Vassilevsky DSP and Mixed Signal Design Consultant

formatting link

Reply to
Vladimir Vassilevsky

Did you try using VCVS for opamp amps (i.e. finite gain but no BW limitations) in the circuit to insure the network is behaving as you expected?

It's a good idea to run the filter with VCVS and the macromodels of the opamp just to see the Q-enhancement effects.

Make sure you use all 4 parameters on the .tran line. The tmax parameter is the important one.

Reply to
miso

Many active filters, which have even moderately high "Q", will clip internally and do all kinds of weirdness at certain signal levels.

...Jim Thompson

--
|  James E.Thompson, P.E.                           |    mens     |
|  Analog Innovations, Inc.                         |     et      |
|  Analog/Mixed-Signal ASIC\'s and Discrete Systems  |    manus    |
|  Phoenix, Arizona            Voice:(480)460-2350  |             |
|  E-mail Address at Website     Fax:(480)460-2142  |  Brass Rat  |
|       http://www.analog-innovations.com           |    1962     |
             
         America: Land of the Free, Because of the Brave
Reply to
Jim Thompson

From the description, this sounds like a one amp two pole filter. The "internal clipping" you mentioned is a dynamic range adjustment issue. This isn't always included in text books for some reason. Basically, you sweep the filter, finding the peaks of every op amp node. Then you scale each node in a manner where you only adjust the gain, not the time constants. If you do this, the noise is minimized and nothing should clip internally.

The proper frame of mind is to think of every op amp output is a filter output, even if only one op amp output has the desired response.

Reply to
miso

Thanks to everyone for their comments. I never found out why the simulation has the spikes in the filter output, but I have to assume that is an artifact of the simulation. I did finally get a handle on the compensation required and have a circuit working well in simulation. We will see what happens when I construct it, but I am confident that it will work as expected.

Reply to
rickman

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.