The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> RF Design >> PLL phase diognosis?
https://designers-guide.org/forum/YaBB.pl?num=1241452951

Message started by liletian on May 4^th, 2009, 9:02am

Title: PLL phase diognosis?
Post by liletian on May 4^th, 2009, 9:02am

Hi guys
The attached pic is a ring oscillator PLL, the spectrum has some lines which I assume which comes from ring oscillator? Any suggestion on origination and how to purified it? Any suggestion will be appreciated!
Thanks

Title: Re: PLL phase diognosis?
Post by pancho_hideboo on May 4^th, 2009, 9:09am

Why do you set RBW=VBW=8MHz ? It is very wide.

liletian wrote on May 4^th, 2009, 9:02am:

The attached pic is a ring oscillator PLL, the spectrum has some lines which I assume which comes from ring oscillator?

I can't observe remarkable spurious spectrum lines.
Floor spectrums are very dirty.
Maybe this is due to your measurement conditions, such as voltage power supply.
Or this might be improper loop filter design.

Title: Re: PLL phase diognosis?
Post by liletian on May 4^th, 2009, 9:23am

pancho_hideboo wrote on May 4^th, 2009, 9:09am:

Why do you set RBW=VBW=8MHz ? It is very wide.

liletian wrote on May 4^th, 2009, 9:02am:

The attached pic is a ring oscillator PLL, the spectrum has some lines which I assume which comes from ring oscillator?

I can't observe remarkable spectrum lines.
Floor spectrums are very dirty.
Maybe this is due to your measurement conditions, such as voltage power supply.
Or this might be inproper loop filter design.

the loop filter is a simple RC filter, is that the problem? I did not use active filter in here because I did not see much advantages.
Thanks

Title: Re: PLL phase diognosis?
Post by pancho_hideboo on May 4^th, 2009, 9:29am

Why do you set RBW=VBW=8MHz ? It is very wide.

liletian wrote on May 4^th, 2009, 9:23am:

the loop filter is a simple RC filter, is that the problem?

Do you determine bandwidth of loop filter by enoughly considering followings.
- Reference frequency(Comparison Frequency)
- Corner frequency of VCO phase noise

liletian wrote on May 4^th, 2009, 9:23am:

I did not use active filter in here because I did not see much advantages.

Again see surely http://www.designers-guide.org/Forum/YaBB.pl?num=1241145270/4#4

Anyway floors of your spectrum are very high and dirty regardlerss of RBW=VBW=8MHz which are very wide.

You should consult someone who has much experience of actual RF measurements about your measurement setups and usages of instruments.

Title: Re: PLL phase diognosis?
Post by liletian on May 5^th, 2009, 7:59pm

pancho_hideboo wrote on May 4^th, 2009, 9:29am:

Why do you set RBW=VBW=8MHz ? It is very wide.

liletian wrote on May 4^th, 2009, 9:23am:

the loop filter is a simple RC filter, is that the problem?

Do you determine bandwidth of loop filter by enoughly considering followings.

- Reference frequency(Comparison Frequency)
I use signal generator to generator signal at 2Ghz
- Corner frequency of VCO phase noise
do you mean 1/f^3 noise? How could loop filter will affect affect Corner frequency and Reference frequency?
Thank you

Title: Re: PLL phase diognosis?
Post by Berti on May 6^th, 2009, 3:33am

Hi,

I am just wondering: What's the loop filter bandwidth?

Title: Re: PLL phase diognosis?
Post by liletian on May 6^th, 2009, 7:45am

Berti wrote on May 6^th, 2009, 3:33am:

Hi,

I am just wondering: What's the loop filter bandwidth?

500MHz
loop filter bandwidth is approximately equal to pll lock range, right?

Title: Re: PLL phase diognosis?
Post by pancho_hideboo on May 6^th, 2009, 8:45am

Berti wrote on May 6^th, 2009, 3:33am:

I am just wondering: What's the loop filter bandwidth?

Expression of "the loop filter bandwidth" is not proper.

Correctly it should be loop bandwidth of PLL.

Title: Re: PLL phase diognosis?
Post by pancho_hideboo on May 6^th, 2009, 8:45am

I think default RBW is 3MHz for your Spectrum Span.
You dared change RBW from 3MHz to 8MHz.

Why do you set RBW=VBW=8MHz ? It is very wide.

liletian wrote on May 5^th, 2009, 7:59pm:

I use signal generator to generator signal at 2Ghz

Can your phase comparator work in 2GHz directly ?
Do you understand "Comparison Frequency" ?

liletian wrote on May 5^th, 2009, 7:59pm:

do you mean 1/f^3 noise? How could loop filter will affect affect Corner frequency and Reference frequency?

See text books on basic of PLL.

See figure-20 of http://www.designers-guide.org/Analysis/PLLnoise+jitter.pdf

You should consult someone who has much experience of actual RF measurements about your measurement setups and usages of instruments.

Title: Re: PLL phase diognosis?
Post by Berti on May 7^th, 2009, 12:00am

Hi again,

I am confused. From your spectrum, the 'loop bandwidth of the PLL' (thanks for the proper terminology! ;) ) doesn't look like 500MHz but more like 50MHz. Furthermore the comparison frequency of 2GHz seems to be pretty high and somehow doesn't sound correct for a loop bandwidth of 500MHz.

I think it is difficult to help you with the measurements when there are even confusions on the PLL itself.

Cheers

Title: Re: PLL phase diognosis?
Post by liletian on May 9^th, 2009, 7:12pm

Berti wrote on May 7^th, 2009, 12:00am:

Hi Berti
when I mentioned it, it means the RC filter's pole location. 1/(1+w/w0). I guess you guys definition is "The
loop bandwidth is defined as the
integrated magnitude of the PLL
frequency transfer function over
the entire frequency spectrum." please let me know. Also, how can you see my PLL's loop bandwidth is less than 500MHz.
Thank you

Title: Re: PLL phase diognosis?
Post by rfmems on May 10^th, 2009, 3:06am

liletian wrote on May 9^th, 2009, 7:12pm:

when I mentioned it, it means the RC filter's pole location. 1/(1+w/w0).

First question, you are really using a RC filter (one R and one C)? If that is true, what kind of phase comparison circuit are you using?

liletian wrote on May 9^th, 2009, 7:12pm:

I guess you guys definition is "The loop bandwidth is defined as the
integrated magnitude of the PLL frequency transfer function over
the entire frequency spectrum." please let me know.

Yes, I think we are all talking about pll bandwidth, not the loop filter bandwidth.

liletian wrote on May 9^th, 2009, 7:12pm:

Also, how can you see my PLL's loop bandwidth is less than 500MHz.

I did not check you spectrum, but first of all, if you vco is free running, no information can be seen from your meansurment. And loop bandwidth of 500MHz is very big, and if that is true, what is your comparison frequency? If you can tell your loop filter R and C values, PFD/PD structure, kvco, maybe I can help you to figure out what might be wrong.

cheers
rfmems