Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: [802.3BA] CL 86 TP1a Over emphasis

Hi Ali

I'm not sure that TWDP is relevant to 802.3ba

 

and are the graphs of  dual dirac jitter or data dependant jitter please?

Could you please spell out the jitter terms so that the topic is accessible to those who haven't been part of (or can't remember) the 802.3aq or SFP+8431 projects ?  Thanks !

 

 

Here's a rough plot overlaying your data.  It looks like TDP is the same for your DDJ or DDPWS apart from a lateral offset of DDJ and DDPWS which is possibly a definitions or simulation artefact.  I don't see a case for an additional spec.

 

 

 

Maybe I'm confused, but I thought LRM cared about DDPWS because it's not equalizable, whereas other forms of data dependent jitter are somewhat equalizable.  In a limiting system none of its equalizeable, so who cares ?

 

 

best wishes

jonathan

Jonathan King

Finisar Corp

1389 Moffet Park Drive

Sunnyvale, CA 94089

 

ph: 1 408 400 1057

cell: 1 408 368 3071

e-mail: jonathan.king@xxxxxxxxxxx

cube C127

From: Ali Ghiasi [mailto:aghiasi@xxxxxxxxxxxx]
Sent: Tuesday, September 01, 2009 5:06 PM
To: STDS-802-3-HSSG@xxxxxxxxxxxxxxxxx
Subject: Re: [802.3BA] CL 86 TP1a Over emphasis

 

Brian

Your points about and why everyone believe LRM modules require DDJ compliance but somehow limiting link
don't care puzzles me too!  Attach 802.3ae and 802.3aq link simulation shows both DDJ and DDPWS impact the
TDP/TWDP, my observation are DDPWS has more linear relationship to the penalty but in the case of TDP
it is pretty resilient to DDJ to some point then you start seeing hockey stick penalty.  May the reason DDJ is
discarded for limiting link is it's insensitivity at low DDJ values?

If the output eye has significant closure and jitter it will impact both DDJ and TWDP.  I have also added Norm
as he was involved heavily with LRM and might be able to shed some more light on this subject.

(Just remember TDP results are based on 300 m per definition of 802.3ae with 55 ps FIR filter).

Thanks,
Ali

Brian Misek wrote:

Ali,
You bring up the comparison of SFF-8431 which is for a link that must support a linear (LRM) interface and host Rx. That type of link, I have been told, is more sensitive to this phenomena. Is this a true understanding or am I miss informed. If we are going to add additional requirements at this date to this interface I would like to be VERY sure they apply to the LIMITING nature of the nPPI interface and J2 and J9 and the eye mask do not sufficiently bound the problem. I would appriciate the insight from module suppliers as to the way this signal would cause them to fail TP2 tests.

Cheers,
Brian Misek

On Tue, Sep 1, 2009 at 3:09 PM, Ali Ghiasi <aghiasi@xxxxxxxxxxxx> wrote:

Hi
 
John and I had side discussion if the current TP1a specifications of J2 and DDPWS
are sufficient for TP2 compliance regardless of input de-emphasis or not.  
 
One condition that I am concerned is the case of a host with low RJ and with 
significant over emphasis 3-5 dB, the DDJ in these case could be in the range of 
0.15 to 0.17 UI.  An example TP1a eye diagram with about 5 dB is shown below.
 
I have seen with SFP+ some module much more sensitive to over emphasis than others,
in SFF-8431 max DDJ does not allow this type of setting and is controlled.  If all 
of the module suppliers building 40GBase-SR4/SR10 have no issue with over-emphasis 
at TP1 by as much as 5 dB then the current specification is sufficient.  On the other 
if other have the same concern we could pile on the unsatisfied comments 216 and 218 
against on D2.1.
 
Thanks,
Ali
 
 
 
 
 
Ghiasi, Ali BroadcomCl 86A SC 86A.4.1 P 408 L 30 # 216
Comment Type TR
If the transmitter has very low RJ~0 then DDJ will approch J2 or 0.18 UI 
due to over
emphasis. Over emphasis can result in sever eye degradation depending on 
the laser
driver gain, etc.
SuggestedRemedy
To protect aginst these over-emphasis scinearios DDJ must be added with 
propose value
of 0.12 UI.
REJECT.
The combination of other specifications, including the eye mask may 
protect against this.
There was no consensus to add this parameter.
Further study of this issue is invited.
Comment Status R
Response Status U
Ghiasi, Ali Broadcom
Response
Cl 86A SC 86A.4.1 P 410 L 19 # 218
Comment Type TR
If the transmitter has very low RJ~0 then DDJ will approch J2 or 0.18 UI 
due to over
emphasis. Over emphasis can result in sever eye degradation depending on 
the laser
driver gain, etc.
SuggestedRemedy
To protect aginst these over-emphasis scinearios DDJ must be added with 
propose value
of 0.12 UI.
REJECT. [Reclassified from 86A.4.2]
See response to comment 216.
Response