1

I designed a sine wave oscillator, but the frequency isn't correct and the wave shape is weird. I use 5V single supply. I create a 2V common point to be virtual ground. The opamp is acting as a non-inverting amplifier. There is a feedback diode in parallel to control the amplitude. The LC in series should act as a bandpass filter to select the frequency. What's wrong with the circuit?

I built the circuit in practice, and the result is almost the same as the simulation result.

simulation result

JRE
  • 67,678
  • 8
  • 104
  • 179
Willis Lin
  • 615
  • 1
  • 8
  • 1
    Does your oscillator circuit follow any known theory, or is it something of your own invention? Not that the latter is a problem. I learnt much of what I know by inventing a complicated something that didn't work, then discovering that there was a wealth of information on how other people had been solving issue for years with a simpler circuit. I'm glad simulation and practice produce the same result, it narrows down the things that need to be fixed. – Neil_UK Feb 01 '23 at 15:25
  • 1
    *I designed a sine wave oscillator* <-- where's the theory behind this design? What frequency did you expect it to oscillate at? Did you try using 1N4148 diodes instead of the 1N4007s – Andy aka Feb 01 '23 at 16:42
  • I remember seeing [this phase-shift oscillator answer](https://electronics.stackexchange.com/a/371292/330261) here and it appears to conform to prediction. – periblepsis Feb 01 '23 at 17:16
  • Regarding the theory behind this design: It is the same principle as the Wien oscillator - a passive bandpass in the pos. feedback loop and a fixed resistive negative feedback loop. For oscillation, the pos. feedback must be somewhat larger than the negative one (at w=wo). – LvW Feb 01 '23 at 17:23
  • thank you. seems like I figure out yhe – Willis Lin Feb 07 '23 at 21:26
  • thank you. I think I figure out the problem. I simulate the bode plot, realizing that the ability to filter out undesired frequency is weak. to improve it, the rei – Willis Lin Feb 07 '23 at 21:29

2 Answers2

1

I think, the design of the negative feedback path is not good. At the series resonance of the LC-bandpass the positive feedback is app. 50%. At the desired oscillation frequency, the negative feedback factor should be slightly smaller.

But why you have chosen (for small amplitudes) a negative factor as small as k(-)=22/69=0.32 ? I would propose to split the feedback R2 resistor in two parts:

20k plus 3k (with antiparallel diodes across 3k only). This ensures a "good" soft-limiting for the amplitudes.

LvW
  • 24,857
  • 2
  • 23
  • 52
1

The circuit operates at two different gains, one when the voltage across R2 is less than 0.6 V in either direction, and one when that voltage tries to be greater than 0.6 V and the diodes conduct.

To soften the transition between these two gain values, add a resistor in series with the diodes, between the right-side ends of the diodes and the opamp output.

AnalogKid
  • 16,865
  • 1
  • 13
  • 25