Good evening.

Last week at work I stumled upon a quite old circuit used for testing active inrush-circuits (some engineer made it back in the early 90's), but noone knows anything about it, nor do they know where it came from, and curious as I was, I had to at least do a simulation of the circuit in question, since it's really quite unlike something I've seen before. It's a 3-phase sinusoidal oscillator, but done in a way I can't quite figure out.

A picture of the circuit has been attached to this thread.

Are there anyone here who can provide a rough explanation of the rough mechanism of action, which components are frequency-determining, or even point me in a direction where there is a similar circuit (maybe even with related equations)?

Thanks in advance.

 

U1A,B,C form a crude phase shift oscillator with notional equal phase shifts of 120° through each of the three stages.

If you scoped the waveforms at the 3 individual amp [U1A,B,C] outputs you would note they aren't sinusoidal - they will be very square since the amps are operating in comparator mode. Hence each output is "cleaned up" with a low pass Sallen-Key filter [U2A,B,C]. The filters reduce the higher order harmonics in an attempt to make the outputs at U2,A,B,C more sinusoidal and thereby represent a balanced 3-phase sinusoidal system. Even with the filtering applied the outputs aren't nicely sinusoidal. A second stage of filtering in each case would enhance the sinusoidal nature of the outputs.

Attached is an alternative 3-phase oscillator format.

 

Thanks for the response - much appreciated!

Sallen-Key was something I've never heard about before, so there's something to read up on.

Thanks again for pointing me in the right direction.

Now how does one determine the frequency of the oscillator? IE, how does 1.5μF and 2.7kΩ yield ~53Hz?

 

In my simulation of your oscillator schematic I get a frequency of ~81Hz with 2.7k and 1.5uF.

I therefore make the frequency of oscillation to be

f=1/(3*R*C) Hz.

With R=4.7k and C=1.5uF I get a simulated value of 48Hz which is consistent with the aforementioned formula.

Unfortunately this apparently doesn't agree with your simulation results.

 

The main difference is that I didn't include the low pass filters. I was simply interested in the oscillator performance in isolation. The filters may be having some effect but I'm not sure why this would be the case. It could be that the filter input impedance is loading the op-amps in the oscillator loop. You could try increasing the filter R values by a factor of 10 and reducing the capacitor values by the same factor. Then observe if the frequency changes.

 

An 'exact' estimate for the unloaded oscillator steady-state operating frequency using ideal op-amps would be ...

f=1/(2.887*R*C) Hz

I can provide a formal proof if it is of any interest.

 

An 'exact' estimate for the unloaded oscillator steady-state operating frequency using ideal op-amps would be ...

f=1/(2.887*R*C) Hz

I can provide a formal proof if it is of any interest.

Thanks again for shedding some light on the subject. I would really like a proof if it isn't of too much trouble, as it'd probably help me understand how you actually came up with that equation in the first place.

The initial equation, "f=1/(3RC)", was that just an educated 'guess'?

 

Sorry this has been slow in coming. I've had problems printing into pdf and I've been distracted by other matters.

 

Wow, that is... I don't even have the words. You just blew my mind.

I'm half a year away from finishing my education as an electronics engineer, which I'll follow up with a BEng in electronics.

Right now, I wouldn't have ever been able to come up with that solution you just provided but perhaps in some years

Anywho, thanks a ton for that absolutely awesome explanation. That's being printed for my archives without a doubt!