The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Analog Design >> VCO question
https://designers-guide.org/forum/YaBB.pl?num=1172299520

Message started by YeeQ on Feb 23^rd, 2007, 10:45pm

Title: VCO question
Post by YeeQ on Feb 23^rd, 2007, 10:45pm

Hi,

I have a question about making VCO stop oscillating.

When doing 3-stage VCO design, I want to verify that - when low frequency gain of the delay cell is smaller than 2 (i.e. 6.02dB), the VCO will stop oscillating because the insufficient gain.

I run AC simulation on one delay cell (with parasitic loading), tuned the bias voltage to make it gain=5.45dB (<6.02dB). Then, I applied the bias voltage on VCO, however, the VCO doesn't stop oscillation.

The DC operating point is set at where the input common mode close to output common mode.

I don't know what I missed in the simulation, please help, thank you!

Title: Re: VCO question
Post by ywguo on Feb 28^th, 2007, 9:53pm

Hi,

Quote:

When doing 3-stage VCO design, I want to verify that - when low frequency gain of the delay cell is smaller than 2 (i.e. 6.02dB), the VCO will stop oscillating because the insufficient gain.

I don't think that is a correct condition that makes the VCO stop oscillate. Would you please explain it?

Thanks
Yawei

Title: Re: VCO question
Post by YeeQ on Mar 2^nd, 2007, 12:13pm

assume one-stage transfer function is A0/(1+S/w0), three stages would be H(s)=A0^3/(1+S/w0)^3

To achieve oscillation,
1) phase shift of each stage is 60deg. i.e. fosc=f0 * sqrt(3)
2) at oscillating freq., |H(s)|>=1. Applying condition 1), min. A0=2 (i.e. 6.02dB)

I wonder if my AC simulation setup is not correct. For the inputs to the diff. pair, vin is connected to DC common mode voltage, vip is connected to an AC source (with DC common mode voltage). or should I connect both inputs vin and vip to two seperated AC sources?

Thanks,
YeeQ

Title: Re: VCO question
Post by mg777 on Mar 2^nd, 2007, 12:41pm

The oscillation frequency needs to be determined by a resonant element. In a digital ring oscillator the oscillation frequency is determined by the loop delay, but an analog oscillator can build up the wiggles out of noise. By 'out of noise' I mean that the 'template' signal comes from the bandpass filter acting on noise, the energy for the oscillation has to come from the negative resistor.

In fact your circuit sounds like the Wien bridge oscillator:
http://en.wikipedia.org/wiki/Wien_bridge_oscillator

I didn't know that Hewlett (of HP fame) is credited with making Wien's idea work, neat stuff with the lightbulb.

M.G.Rajan
www.eecalc.com

Title: Re: VCO question
Post by fonseca.ha on Mar 6^th, 2007, 4:58am

Just for some cross checks:
Does the oscillator oscillate at the same frequency that resulted in 60deg phase margin when making the AC sim?

I think that analisys may not correlate because the AC is computed from the small signal analisys at a given DC point, and nomally ring oscillators have big swings and cannot be treated as operating in a linear region,

Regards,
Humberto

Title: Re: VCO question
Post by adesign on Mar 7^th, 2007, 10:27pm

fonseca.ha wrote on Mar 6^th, 2007, 4:58am:

You are right Humberto. Initially, for small swings, we can say that the gain required is 2. However, when the swing goes large enough, the small-signal analysis is no longer valid and in that case the output frequency is given by 1/(2nT) where n is the no. of inverting stages and T is the delay per stage.

Hope it helps.

Title: Re: VCO question
Post by nandy on Mar 21^st, 2007, 9:03am

Since the gain used here is only slightly lesser than 2, any disturbance anywhere (due to noise) would result in ringing for some period of time. Although in the small signal sense this oscillation isn't self-sustaining, the inverters pick up this disturbance and blow it up into a large-signal full-scale oscillation. So, I guess, your oscillator can potentially oscillate unless the gain of each stage is reduced well below 2.