The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> High-Speed I/O Design >> Transition density estimation
https://designers-guide.org/forum/YaBB.pl?num=1353991372

Message started by BackerShu on Nov 26^th, 2012, 8:42pm

Title: Transition density estimation
Post by BackerShu on Nov 26^th, 2012, 8:42pm

May be it's a trivial question, but I'd really appreciate your comments.

Transition Density is defined as the ratio of transitions (or edges) to the number of unit intervals in a serial data stream. For instance, clock signal has a transition density of 1. For many serial communications test signals, this ratio approaches 0.5, or half as many transitions as bit intervals, but it's not always the case.

If we want to estimate the transition density of any serial data stream, or how much it is offset from 0.5, how is it normally done?
If there is no reference clock signal available for estimation, can we do it?

Welcome any comments or please point me to any reference which you think is related to this.

Thanks,

Title: Re: Transition density estimation
Post by raja.cedt on Nov 28^th, 2012, 8:49am

Hello,
In real time you can't guarantee. In measurement it depends on the type pattern you are applying. But to make sure for min transition encoding will be used (8b/10b or 64b), otherwise CDR would be problematic.

I heard few very old standards have RZ coding, just to have .5 transition density.

Thanks,
raj.

Title: Re: Transition density estimation
Post by ywguo on Dec 10^th, 2012, 9:47pm

Hi BackerShu,

Sure you can estimate transition density if you have recovered clock and data in the receiver. Do you want to measure the transition density on the fly?

Best Regards,
Yawei

Title: Re: Transition density estimation
Post by raja.cedt on Dec 11^th, 2012, 12:49am

Hello Yawei,
can you clearly, how do you estimate density from recovered data and clock?, you mean by checking BBPD output or any thing else?

Thanks,
Raj.

Title: Re: Transition density estimation
Post by ywguo on Dec 11^th, 2012, 1:42am

Hi Raj,

I think the transition density can be count if the data and clock are recovered. Some counters. Probably that is not what BackerShu needs.

Best Regards,
Yawei

Title: Re: Transition density estimation
Post by raja.cedt on Dec 11^th, 2012, 2:25am

hello Yawei,
do you have reference on transition density, out of curiosity i am asking how people guarantee min transition for CDR in encoding?

Thanks,
Raj.

Title: Re: Transition density estimation
Post by BackerShu on Dec 29^th, 2012, 5:47pm

Thanks Yawei and raja!

ywguo wrote on Dec 10^th, 2012, 9:47pm:

Sure you can estimate transition density if you have recovered clock and data in the receiver. Do you want to measure the transition density on the fly?

Yes, if both recovered clock and data are available, we can do transition density estimation by counting.
What I really want to know whether there is a possible a way to estimate the transition density to help with frequency acquisition in referenceless CDR, meaning before the CDR locks; thus recovered clock and data are not available. That's what I mean by "no reference"
The reason why I am asking this is there is a interesting paper in which a frequency acquisition is presented but its inaccuracy is limited by transition density. If you guys are interested, here is the paper.
[1] R. Inti, S. Member, W. Yin, and A. Elshazly, “CDR With Unlimited Frequency Acquisition Range and Improved Input Duty-Cycle Error Tolerance,” Solid-State Circuits, IEEE Journal of, vol. 46, no. 12, pp. 3150–3162, 2011.
If transition density can be estimated without a reference clock, or even with a clock which has unknown relationship with the data rate, the error can be compensated.

Title: Re: Transition density estimation
Post by BackerShu on Dec 29^th, 2012, 5:52pm

raja.cedt wrote on Dec 11^th, 2012, 2:25am:

hello Yawei,
do you have reference on transition density, out of curiosity i am asking how people guarantee min transition for CDR in encoding?

Thanks,
Raj.

Hello Raja,

I think transition density is guaranteed in Tx side with some special encoding. As I know, the way it usually is implemented is some scrambler.
Is this what you are asking? Maybe I am missing your question~