Hi all, I'v seen this ring oscillator several times , but I didnot know its performance. Does anyone give me some paper about it ? thank you in advance!
It looks common, obviously the behaviour depends on the used technology. High performance VCOs are often using differential buffers to reduce susceptibility to supply interferences. I would expect rather 10 to 20 than 3 gates with a typical ring oscillator.
I've seen this a lot in papers and theses that discuss PLL design, since this is a basic topology for the PLL's VCO. This topology is attractive because it's voltage vs. frequency response is fairly linear.
I'll try to search the documents I have and get back to you.
Thank you, Diemilio!
I found its frequency to voltage ( to inverter supply voltage ) is
(vc -vt)^2/( N*vc*C*(1/kn+1/kp)^(-1) ), if vc >>vt, then it is linear, but in actual case vc is about
2~3 vt, so this pullzes me. It is not so linear but many people use this topology.
I havent seen this before (Using source follower for the ring..) but this ring will not have linear KVCO as you said...
I hope you are not misled by the wrong polarity of the M5 transistor (A PMOS there is much more common and has linear KVCO)
I don't see a wrong polarity in the circuit. Also, a voltage follower ring doesn't oscillate, cause it has a gain < 1. If not using differential stages for better performance, it can work as shown (typically using more stages).
The book is called CMOS Circuit Design, Layout, and Simulation (By Baker) and you can find it in chapter 19 (Digital PLLs) section 19.2 (Voltage controlled oscillators). There you can see that the frequency is directly proportional to the current across the stages, which can be linearly controlled with the bias stage also explained there.
Also Razavi Design of Analog CMOS Integrated Circuits has a chapter on ring oscillators and another on VCOs, discussing e. g. current steering. He's particularly focussing on differential pair ring oscillators.
FvM., All these links point to a current source pumping current in to the supply node of the ring oscillator., which is very common. But in the first diagram., michaeljackson has shown something different., Vctrl is buffered by a source follower..This could also work as a VCO...but has a lot of diadvantages IMO... But are you sure this is discussed in that book?
I raised my suspicion about this but you reasoned something which I dont understand
I see your point. For simplicity, we can designate the said circuit a voltage-steered oscillator. I agree, that some disadvantages should be expected, particularly temperature sensitivity, larger exemplar variation and a basic non-linearity. But the linearity may be still sufficient for some restricted frequency range applications.