[ibis-macro] Re: Revision to the Jitter BIRD (123.2)
- From: "Muranyi, Arpad" <Arpad_Muranyi@xxxxxxxxxx>
- To: "ibis-macro@xxxxxxxxxxxxx" <ibis-macro@xxxxxxxxxxxxx>
- Date: Tue, 1 May 2012 01:04:12 +0000
Walter,
Here are a few comments from Vladimir on the proposed
BIRD 123:
1. There are several places where Walter introduced "half the peak to peak
variation" instead of "peak to peak variation".
These are:
Page 3, for Tx_Dj
Page 5, for Rx_Clock_Recovery_Dj
Page 7, for Rx_Dj
In all such cases, the proposed formulas use rand(), still defined as "a
function that returns floating point numbers between -.5 and +.5 with white
uniform distribution".
However, the formulas were not changed accordingly, and remain respectively:
Time(n)= n*bit_time+Tx_Dj*rand()
actual_time = ideal_time + Rx_Clock_Recovery_Dj*rand()
actual_time = time + Rx_Dj*rand()
We cannot change only the definition for rand() since it is used in the
expressions like "actual_time = ideal_time +
Rx_Clock_Recovery_Sj*sin(Pi*rand())"
Therefore, we should either
(a) add a factor 2 in above formulas, to make them:
Time(n)= n*bit_time+2*Tx_Dj*rand()
actual_time = ideal_time + 2*Rx_Clock_Recovery_Dj*rand()
actual_time = time + 2*Rx_Dj*rand()
Or,
(b) we can change the definition of rand() to "a function that returns floating
point numbers between -1.0 and +1.0 with white uniform distribution", but also
modify all its occurrences under "sin" or "cos" by adding a factor 1/2:
sin(Pi/2*rand()) instead of sin(Pi*rand())
2. In a couple of places we need to change (-1)n into (-1)^n, these are on
pages 6 and 7:
Page 6:
actual_time=ideal_time+Rx_Clock_Recovery_DCD*(-1)n
Page 7:
actual_time = time + Rx_DCD*(-1)n
Thanks,
Arpad
=========================================
From: ibis-macro-bounce@xxxxxxxxxxxxx [mailto:ibis-macro-bounce@xxxxxxxxxxxxx]
On Behalf Of fangyi_rao@xxxxxxxxxxx
Sent: Thursday, April 26, 2012 7:49 PM
To: wkatz@xxxxxxxxxx; ibis-macro@xxxxxxxxxxxxx
Subject: [ibis-macro] Re: Revision to the Jitter BIRD (123.2)
Hi, Walter;
Since the Tx_Dj definition is changed to "half the peak to peak", shouldn't the
formula for Tx_Dj be changed accordingly to
Time(n)= n*bit_time+2*Tx_Dj*rand()
rand()is a function that returns floating point numbers between -.5 and
+.5 with white uniform distribution.
Thanks,
Fangyi
From: ibis-macro-bounce@xxxxxxxxxxxxx<mailto:ibis-macro-bounce@xxxxxxxxxxxxx>
[mailto:ibis-macro-bounce@xxxxxxxxxxxxx]<mailto:[mailto:ibis-macro-bounce@xxxxxxxxxxxxx]>
On Behalf Of Walter Katz
Sent: Friday, April 20, 2012 9:49 AM
To: IBIS-ATM
Subject: [ibis-macro] Revision to the Jitter BIRD (123.2)
All,
Based on comments from James Zhou, and discussions in today's Open Forum
meeting, I have made changes to BIRD 123.2 (enclosed).
The two classes of changes relate to:
1. Making the new Dj jitter parameters "half the peak to peak" instead of
"peak to peak" to be consistent with all of the other jitter parameters.
2. Clarifying the "time" in statistical equations for
a. Rx_Clock_Recovery_Sj
b. Rx_Rj
c. Rx_Sj
d. Rx_DCD
Please review these changes, and comment by e-mail. Discussing this will be on
the agenda for the next IBIS-ATM meeting.
Walter
Walter Katz
wkatz@xxxxxxxxxx<mailto:wkatz@xxxxxxxxxx>
Phone 303.449-2308
Mobile 303.335-6156
Other related posts: