From owner-stds-802-3-hssg@xxxxxxxx Mon Feb 28 20:03 GMT 2000
Received: from gatekeeper.pdd.3com.com (gatekeeper [188.8.131.52])
by isolan.pdd.3com.com (8.9.1b+Sun/8.9.3) with ESMTP id UAA09296;
Mon, 28 Feb 2000 20:03:26 GMT
Received: from ruebert.ieee.org ([184.108.40.206])
by gatekeeper.pdd.3com.com (Netscape Messaging Server 3.6)
with ESMTP id AAA719; Mon, 28 Feb 2000 20:01:44 +0000
Received: by ruebert.ieee.org (8.9.3/8.9.3) id OAA24631; Mon, 28 Feb 2000 14:14:21 -0500 (EST)
From: "Edward Chang" <edward.chang@xxxxxxxxxxxxxxxx>
To: <rtaborek@xxxxxxxxxxx>, <stds-802-3-hssg@xxxxxxxx>
Subject: RE: PAM-5 at 5 Gbaud
Date: Mon, 28 Feb 2000 14:20:42 -0500
X-Priority: 3 (Normal)
X-Mailer: Microsoft Outlook IMO, Build 9.0.2416 (9.0.2910.0)
X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2314.1300
X-Resent-To: Multiple Recipients <stds-802-3-hssg@xxxxxxxxxxxxxxxxxx>
X-Info: [Un]Subscribe requests to majordomo@xxxxxxxxxxxxxxxxxx
Content-Type: text/plain; charset="us-ascii"
The 5 GBaud, PAM-5 waveform at page 7 makes me very uncomfortable. For
example, the one in the middle has three transitions, of which the
peak-to-peak amplitudes are 2250 unit(mv?), 1300 unit (mv?), and 1000
unit(mv?) with estimated transition times, (0%-100%)65 ps, 60 ps, and 60 ps,
respectively. For a five level coding, they seems to match with the +2, +1,
0, -1, -2 level definition with equally spaced timing interval.
If they are not part of the coding, but a ringing or some other noise, they
will confuse the level and timing detecting circuits to cause undesirable
results. If it is supposed to have only one transition per a cell time, 200
ps, then those extra transitions will cause the PLL being out of
synchronization with the receiving data, or excessive clock-jitter to induce
high BER. With the near equal magnitude of amplitudes, and timing
intervals, no way a ordinary filter can differentiate the needed transition
from the unwanted transitions.
One way or the other, these issues should be resolved before July decision
Edward S. Chang
NetWorth Technologies, Inc.
[mailto:owner-stds-802-3-hssg@xxxxxxxx]On Behalf Of Rich Taborek
Sent: Sunday, February 27, 2000 2:14 AM
Subject: Re: PAM-5 at 5 Gbaud
My comments are interspersed below,
Edward Chang wrote:
> I am glad to discuss with you again. It is always enjoyable.
> Responses to your questions:
> 1. PAM-5 Symbol period.
> There are two clear PAM-5 waveforms on HSSG reflector.
> One of them is in page 7 (PAM-5 5 G Baud Modulated Laser Output), and page
> (Zoom Modulated Laser Output) of "10 Gb/s PMD Using PAM-5 Modulation by
> Oscar Agazzi in January.
> The zoomed modulated laser output picture has a very clear waveform of
> transitions per each 200 ps, which is 1/(5 G Baud).
The graphs on pages 7 and 8 are the output of a Laser model summarized on
prior pages. However, the signaling, as indicated by the title of the
presentation and on page 3 is PAM5 (5-level Pulse Amplitude Modulation) at 5
Gbaud. The signaling period is 200 ps with only one logic transition per
This signaling is the same as I've proposed earlier and described in the
'99 802 MAS tutorial.
> The another picture is in page 31 of your Tutorial presentation "MAS
> Techniques for 10 GbE".
> This picture is clear but in 60 M Baud rate, and the timing scale is not
> clearly denoted. I am not clear of this picture regarding its timing
This figure is meant to illustrate PAM5 signaling. I did not have the
to signal at 5 Gbaud at the time so the signaling rate is much slower.
> 2. Bit Error Rate.
> The system bit error rate involves every components including receiver and
> others, and it is also affected by the data recovery technique. PAM-5
> requires not only the timing information of each bit, but it also requires
> the voltage level information of each bit, together, to correctly recover
> the data. The "double requirements" will make it harder to recover
> correctly, than a straight forward "timing information" only requirement.
> However, if a link has a very reliable recovery technique and circuit, it
> can achieve the target BER without any problem. All we need is keep
> on it to come out with a right solution.
> The BER issue is not a absolute issue, rather it is depending on how much
> like to pay to achieve the target BER.
...and when paying, with what currency. Consider coding to be a "cheaper
> Edward S. Chang
> NetWorth Technologies, Inc.
> Tel: (610)292-2870
> Fax: (610)292-2872
> -----Original Message-----
> From: owner-stds-802-3-hssg@xxxxxxxx
> [mailto:owner-stds-802-3-hssg@xxxxxxxx]On Behalf Of Rich Taborek
> Sent: Saturday, February 26, 2000 2:41 AM
> To: stds-802-3-hssg@xxxxxxxx
> Subject: Re: PAM-5 at 5 Gbaud
> I heartily agree that proof of the general principles of MultiLevel
> need to be proven to an extent to remove any reasonable doubt of technical
> feasibility by the July Plenary meeting. To this end, I would like all
> proponents of MultiLevel signaling to put their best resources to work.
> I'll try to respond to some of the questions recently posed to this
> regarding PAM5 below. I don't mean to imply that all MultiLevel problems
> completely solved, only that solutions have been proposed, many of which
> been applied to other signaling technologies.
> Edward Chang wrote:
> 1) PAM5 Symbol period
> > For the timing part, the PAM-5 coding has maximum of three transitions
> > period defined by its baud rate. For example, at the 1.25 G Baud data
> > (800 ps period), within an 800 period, there are maximum transitions of
> > every 266 ps, which is equivalent of 3.7 G Baud of a NRZ coding.
> I may have missed something between you and Jaime here, but a baud
> generally denotes a single transition (ignoring undershoot, overshoot,
> etc.). The value of the symbol encoded during a baud period may have a
> value for a binary encoded signal, or a quinary value, in the case of
> Therefore, at 1.25 Gbaud, the symbol period is 800 ps, not 266 ps. If you
> believe this not to be true, please explain.
> 2) Bit Error Rate
> > As it was mentioned before, to recover the bit timing and level
> > simultaneously from the 1.25 G baud receiving signals to meet 10^-12 BER
> > not a easy job, which has not been proved in the field yet. To assume a
> > fiber will automatically achieve 10^-12 is a desirable objective, but
> > assurance.
> Coding may be employed in MultiLevel signaling schemes to increase the
> BER to a value required to meet link BER objectives. This has been
> several HSSG presentations included those noted immediately below. It is
> generally more cost effective to deploy coding to enhance the diminished
> of a MultiLevel link than to than trying to attempt increase the raw link
> BER to
> match the system BER via traditional means (e.g. Receiver Sensitivity
> improvements, higher RIN lasers, etc.)
> http://grouper.ieee.org/groups/802/3/tutorial/july99/mastut.pdf (See
> to FEC)
> 3) Laser Linearity
> Wenbin Jiang wrote:
> > Laser linearity is one concern. What about its slope efficiency
> > with the temperature?
> A Link Calibration procedure has been proposed which initializes the link
> transmitter and receiver levels using a closed-loop feedback system
> the existence of a full-duplex link. Relatively slow changes in slope
> such as those associated with temperature changes can be readily and
> tracked with the same feedback system operation with a low loop bandwidth.
> Calibration principles were introduced in the MAS tutorial (link below)
> 42-56, 84.
Richard Taborek Sr. Phone: 408-845-6102
Chief Technology Officer Cell: 408-832-3957
nSerial Corporation Fax: 408-845-6114
2500-5 Augustine Dr. rtaborek@xxxxxxxxxxx <= NEW!
Santa Clara, CA 9 5008 http://www.nSerial.com <= NEW!