Cable characteristics [was: [10GBASE-CX4] Working Paper Available]
Hi, all.
Sec. 54.8.2 'Cable Assembly Insertion Loss' sends a mixed message.
Eqn. 54.3 is an inequality, and suggests that *any* loss curve
which falls above the line in Fig. 54-8 is compliant. However
the associated text places a 10% deviation limit on frequency
response characteristic.
This is the tip of the iceberg on the variable-cable-length
equalization issue. While the text of the standard does not
need to say (and should not say!) how receivers are implemented,
it needs to make it possible to implement reliable receivers
using available technology.
The most common thing for high-rate serial NRZI receivers that
deal with variable cable lengths is to implement adaptive EQ
at the receiver end. (E.g. SMPTE 292M Hi-definition TV at
1.485 Gb/s). The channel is modelled as a variable length L
of "typical" cable, so the receive EQ only needs to estimate
*one* parameter (L) in order to apply an appropriate EQ curve.
For coax a typical model is
H(f, L) = exp(-(K1 * L * (1+j) * sqrt(f) + K2 * L * f) ).
Note that this expression defines *both* loss and phase
response, which can be important.
I suggest that sec. 54.8.2 should be written to include the
following features:
(1) Both attenuation and delay dispersion should be controlled.
(2) The cable-response formula should include a parameter L
corresponding to "equivalent cable length"
(3) The response of a compliant cable should be specified to
be within a tolerance window (+/- 10%?) of the formula,
for *some* value of L (not necessarily equal to the
actual cable length).
The resulting specification basically calls for any compliant
cable to "look like" a length of the nominal cable, within
some tolerance. This is a very reasonable requirement for
cable vendors to meet, and a very tractable behavior for
receiver designers to work with.
Here is a useful reference from the IEEE ISSC conference a few
years ago:
http://iroi.seu.edu.cn/isscc99/digests/1999/DATA/23_3.pdf ,
WP 23.3 "A 2.5Gb/s Adaptive Cable Equalizer".
Sony Broadcast also published a white paper on the development
of the 292M standard ca. 1998.
On the transmit side, 54.7.3.6, the preemphasis is defined only
by an example of behavior on a particular test pattern. While
the intended behavior on other signals may be intuitively clear,
it needs to be spelled out more clearly in order to properly
ensure interoperability of compliant equipment.
Great work so far!
Cheers,
Chuck Harrison
Far Field Associates, LLC
+1 360 863 8340 (voice) PST = GMT-0800
ddprocurve@antelecom.net wrote:
>
> 10GBASE-CX4 Study Group Members,
>
> In anticipation of the editorial review we will want to perform
> at the upcoming meeting in Vancouver, Howard Baumer has provided
> a copy of his working paper proposal.
>
> I have placed it on the 10GBASE-CX4 website at the following URL.
>
> http://grouper.ieee.org/groups/802/3/10GBCX4/wp2p01.pdf
>
> Please open and review that document and provide feedback to Howard as soon
> as possible. You can reach him at hbaumer@broadcom.com.
>
> Regards,
>
> Dan Dove
> Chair, 10GBASE-CX4 Study Group