Hi Dana, thanks for the hint.Think in terms of phase angle, to get osc loop phase must be n x 360, inverter elements
contribute 180, so in this example they contribute 540, so you need how much phase out
of the RC's ?
Then back into what freq this occurs at.
Regards, Dana.
The propagation delay is very small relative to the typical frequency of a common phase shift oscillator, and so while it will certainly have some effect it would not be very large at 1 Kilohertz. The variation in switch point voltage will have a larger effect, and with cheap CMOS that will matter. And in the really cheap consumer products, using a 4001 quad nor as a set of inverters is fairly common.Of course the prop delay affects frequency. In fact in semiconductor processing,
using 3 inverters in series feedback, on wafer test site, the freq is measured (and
largely dependent on Tpd, Vdd) to evaluate the Fmax of a process.
When you look at transfer curves of the inverter there is a substantial "linear"
region over which small signal behavior exists to create an osc.
Is there no relation to switching levels, no, Vdd very much impacts Vth, and G of
the inverter. But the osc can and will operate around the bias point which Vth is
very much involved. Main contribution to osc freq however is R and C, available
current and capacitance. Tpd directly impacted, Vth not so much.
The CMOS inverter structure is largely a linear G element. one with a RC lowpass
behavior. Switching speed characterized by available current to charge C, usually
expressed as prop delay.
http://bwrcs.eecs.berkeley.edu/Classes/icdesign/ee141_f01/Notes/chapter5.pdf
https://www.egr.msu.edu/classes/ece410/mason/files/Ch7.pdf
Regards, Dana.
I concur.The propagation delay is very small relative to the typical frequency of a common phase shift oscillator, and so while it will certainly have some effect it would not be very large at 1 Kilohertz.
This formula only applies to the circuit of Fig. 2 of the referenced paper where the CR phase shifting pairs are separated from one another by unity gain buffers.I think View attachment 155434
eg. 60 degrees /section.
https://www.inictel-uni.edu.pe/site...blicaciones/12/rc_phase_shift_oscillators.pdf
Regards, Dana.
Assuming F is << Tpd, probably formula still pretty good. Why, because the otherWithout the buffers, as in Fig. 3, the formula no longer applies.
Yes, but when we characterized parts for process standards we often found fairlyWhat we find in the real world is that the inverters are not perfect and the components are seldom identical.
Jony, I am wondering about where in the circuit the waveforms shown are taken from. Or is it a simulation? I am especially puzzled by the lack of squareness of what should be a square wave. Please let us know which circuit, or variation, this is from. AND where the measurement points are located.I don't think so.
View attachment 155523
As you can see for R = 10kΩ and C = 1nF; Vdd = 5V the frequnct of oscilation is arond 28Khz.
So the formula for RC Phase-Shift Oscillator won't work in this case.