Hi

I'm building sine wave oscillator using 4 RC filter (Bubba oscillator)
Can someone explain to me that why don't you need non-linear feedback in phase shift oscillators to avoid distortion.

Also If someone can tell me which oscillator (only using op-amps) will be the best to generate closest to pure sine wave. Is bubba oscillator good?

My requirements are
DC input upto 20V and output AC 10V rms.

Thanks

 

I have never heard of a bubba oscillator, and neither has Wikipedia. The closest I've hear of is a phase shift oscillator, which uses 3 RC networks, each shifting the signal 60°. The op amp is in an inverting configuration, which gives the circuit it's positive feedback.

The other popular RC sine wave oscillator is the Wien bridge oscillator, which is relatively easy to build. A common feature of this oscillator is the use of a small light bulb to make a simple AGC (automatic gain control), which allows you to move the frequency around and still have a good quality sine wave. This circuit was invented by Bill Hewlett, and gave HP their start.

I googled "bubba oscillator", evidently it is pretty common. Here is the TI datasheet link on oscillators. I don't like the fact it takes so many resistors and capacitors, which is probably why it isn't that common in practice. Basically it uses 45° of feedback, and each op amp stage is a low pass filter, which explains it's purity. The output of the one amp with gain is going to be pretty distorted.

What frequency are you aiming for?

 

Hi

I'm building sine wave oscillator using 4 RC filter (Bubba oscillator)
Can someone explain to me that why don't you need non-linear feedback in phase shift oscillators to avoid distortion.

Also If someone can tell me which oscillator (only using op-amps) will be the best to generate closest to pure sine wave. Is bubba oscillator good?

My requirements are
DC input upto 20V and output AC 10V rms.

Thanks

The inverting amplifier gain is set slightly higher than the minimum required for oscillation, so it will saturate at one or both rails, generating a small amount of distortion. Each successive RC network filters out more of the distortion. The RC network immediately preceding the gain stage will have the least amount of harmonic distortion relative to the fundamental amplitude at that node.
The schematic erroneously identifies sine and cosine as coming from successive RC networks. Each RC network contributes 45° of phase shift, so the cosine will be on the second RC network after the sine output.
The Bubba oscillator looks good to me, unless you want it to be variable. I have never built one. I think an op amp with rail-to-rail outputs would be best, because it should inherently generate a lower level of 2nd harmonic than an op amp with non-symmetric limiting.

You won't be able to get more than 7V RMS with a 20V supply.

 

The principle of an sine oscillator an OP-amplifier and either a bandpassfilter between otuput and + input and a negative feedback to reduce amplification. This is i.e. the well known winebridge circuit. The feedback is critical for a good waveform and stabillity.

An other way is to use a notch filter between output and -input i.e. a twin T-filter. A weak positiv feedback will start the oscillation. The feedback is not so critical for the waveform and stabillity, it is more critical to design the in balance in components, R and R/2, C and 2C, so the notch is as deep as possibly. Use component from the same batch and connect them in parallell for R/2 and 2C.

 

The bubba oscilator isn't critical though, I suspect almost any quad op amp will work, and will produce hi purity sine waves.

 

I just checked out the bubba and the formula to calculate the frequency looks simpler than the one for the phase shift oscillator i am currently playing with, maybe i'll build the bubba instead

 

Because it has more stages, the bubba is less susceptible to component variations than the 3-stage PSO, making it a low sensitivity design. Both rely on non-linearities of the active element for amplitude stabilization, which is why neither is a good choice for "high purity" sine waves.

ak