# What is the source of distortion in this simple common-em. Colpitts oscillator waveform?

I took the standard common emitter NPN (3904) amplifier with an LC tank on the base:

I experimented with R1/R2 thinking it was a gain issue, but it is still there with lower gain to the point where my scope can't trigger in all the noise. I also swept the frequency from 70kHz to 500Khz by changing C1/C2 and L1. Still there.

Here's an image.

Both channels are in DC coupling and I'm probing the output (yellow) and the input prior to the coupling/bypass cap (green) at the base.

It is also there with a 2N2222.

EDIT: See Andy and Enrico's replies below. Reason? Non-linearity of slamming a 2n3904 rail-to-rail. I replaced the the transistor with a 1458 op amp (using a 2426C rail splitter to generate +/-2.5V) and Rf/Ri of 10k/1k .... and presto, a nice clean sinusoid that is more stable over a larger range of CL values.

• It’s a standard problem with a BJT colpitts oscillator. The only thing that controls amplitude is the collector voltage crashing into the rails. Try adding a low value emitter resistor, maybe a hundred ohms and see if it improves else, you need a better method if distortion is still unacceptable. There’s nothing elegant or subtle about a basic bargain basement sinewave oscillator Commented Mar 14, 2021 at 21:47
• If you need a less distorted sine wave, use the "input" side as the output, and amplify it if needed. Commented Mar 14, 2021 at 22:05

The source of distortion is the non linearity of the 2N3904.

Usually, oscillators are filtered at the output using selective LC circuits which help remove unwanted frequencies and narrow the spectrum.

This is a poor way to make a Colpitt’s. The Q is too low on the output as the current source collector turns into a switch from high to low R.

The better configuration re-arranges the parts so that the inductor is on the collector to Vcc and there are 3 caps from supply to collector to emitter to ground that ought to be around 3C,C,3C respectively for gain.

Emitter voltage determines the quiescent current from the base to prevent saturating the collector with the appropriate <50 mV AC across Vbe to pump the current. This gives almost 2x Vcc output swing and approaching saturation dampens the Q but is much easier to avoid distortion than your circuit. I have a simulation in other answers if you wish to search me.

Your way (more power hungry to get higher Q)

My way. More signal, more efficient, higher Q across collector.