How to judge a ring osc osillate or not?I mean by simulation

[johnq_hu]

I used to use 3 differential stages ring osc in PLL designs. But now I am designning a 4 stages one. I know the basic structure. My question is: 4 stages osc may oscillate and it also may be deadlocked. Is there a method by which we can judge that the osc will never be deadlocked?

 

[khouly]

u need to check the phase and the gain of the oscillator open loop ,

check razavi book about optical communicatin IC , there is chapter about the oscillators very interesting

khouly

 

[neoflash]

what do u mean by dead locked?

 

[elbadry]

Re: How to judge a ring osc osillate or not?I mean by simula

I think you need to ensure that the gain at dc is low enough not to cause the oscillator to latch. I am not sure, but perhaps AC coupling can help

 

[xwcwc1234]

Re: How to judge a ring osc osillate or not?I mean by simula

For a ring osc. 3-stage is better than 4-stage in terms of jitter ? How to determine the stage number ?

 

[yibinhsieh]

Re: How to judge a ring osc osillate or not?I mean by simula

For a ring osc. 3-stage is better than 4-stage in terms of jitter ? How to determine the stage number ?

I'm sorry to say you are wrong.

The more stage of ring-osc has better jitter performance due to their Q is better.
But the power consumption is high for the same frequency.

I usually simulate the phase noise to judge the number of stage and power.

Therefore, you need to know the phase noise spec. derived from jitter spec.

Of course, you need to prepare some margin.

Yibin.

 

[husseinadel]

Re: How to judge a ring osc osillate or not?I mean by simula

first you should assure from loop gain and phase that you established a positive feedback system at the desired frequency ;this include of course as said above not to latch at zero frequency

2nd you should assure that you trigger the oscillator and that the build time had elapsed ; you can put initial condition to create a potential difference between the 2 nodes of oscillation , and estimate your start-up time needed ; if you use current sources & you put bias filter

 

[jfyan]

Re: How to judge a ring osc osillate or not?I mean by simula

For a ring osc. 3-stage is better than 4-stage in terms of jitter ? How to determine the stage number ?

hi
the num. osc stages maybe determined by the num. of the phases are needed.
and 4-stage is perefered in most cases, due to the stable oscilltion building.

jeff

Added after 11 minutes:

For a ring osc. 3-stage is better than 4-stage in terms of jitter ? How to determine the stage number ?

I'm sorry to say you are wrong.

The more stage of ring-osc has better jitter performance due to their Q is better.
But the power consumption is high for the same frequency.

I usually simulate the phase noise to judge the number of stage and power.

Therefore, you need to know the phase noise spec. derived from jitter spec.

Of course, you need to prepare some margin.

Yibin.

hi
i think you can not make this simple conclusion.
i suggest you to read hajimiri's book or papers about jitter and phase noise in ring oscillators.
jeff

 

[gamma1]

Re: How to judge a ring osc osillate or not?I mean by simula

Need to be an odd number of stages, can't be even

 

[dumbfrog]

Just make the gain >10 (ideal situation is >1, but the process might kick you into gain =0.1)

Razavi (Analog IC book and PLL book) has a good chapter on vco


PS: gamma1, 4 stage differential stage, not single ended

 

[chenjia]

make sure that the system loop is negtive feedback at the 0Hz,I think.

 

[chenjia]

make sure that the system loop is negtive feedback at the 0Hz,I think.