Simulation trouble: Bode plotting an oscillator

Article By : John Dunn

Using my Multisim SPICE tool to analyze an oscillator, I ran into a little bit of simulation trouble.

Using my Multisim SPICE tool to analyze an oscillator, I ran into a little bit of simulation trouble.

My circuit was a simple Colpitts oscillator and the simulation showed that it would indeed oscillate. However, I wanted to look at the transfer function of the passive components just to make sure that I had everything right, and when doing that, I got into some trouble trying to get the Bode Plotter to function (Figure 1).

Figure 1 An unsuccessful analysis attempt.

Using a Bode Plotter on the passive component network alone, I saw what I was hoping to see: a sharp peak in the transfer function at the correct frequency—in this case at approximately 1.6MHz. The signal source V1 provided the circuit excitation required by the Bode Plotter tool to get that instrument to function.

However, when I tried to use the Bode Plotter on the oscillator circuit itself, there was no response, even though the oscillator was working and delivering signal. The Bode Plotter would not accept the oscillator’s own oscillation as the circuit excitation. I had to do something more (Figure 2).

Figure 2 A successful analysis attempt.

I had to add V2 and R4 as shown in Figure 2 to serve as an external stimulus to convince the Bode Plotter to do its thing. (It’s fun when you can anthropomorphize an inanimate thing, isn’t it?)

It sure didn’t take much. I used one microvolt peak for V2 and fed that to the top of R3 via one teraohm. To be sure, the circuit disruptive effect of V2 and R4 on the oscillator was negligible, but by including those two parts in the simulation, the Bode Plotter came to life and gave me a usable result.

Maybe I could have put my one microvolt signal in series with R3 or something like that but I had what I needed so I didn’t look any further into it.

To be sure, the Bode plot of the passive components alone and the Bode plot of those components in the oscillator differ slightly, but I chose to ascribe that difference to a slight loading effect, perhaps by the 2N4392 gate.

Good enough!

This article was originally published on EDN.

John Dunn is an electronics consultant, and a graduate of The Polytechnic Institute of Brooklyn (BSEE) and of New York University (MSEE).


Leave a comment