Re: Cable characteristics [was: [10GBASE-CX4] WorkingPaperAvailable]
Chuck,
I agree with your concern about making sure we have a spec that
ensures interoperability. I believe that Howard does as well. During the
last meeting, we discussed the parameters that are
necessary to ensure an adequate specification. That discussion
was focused on an alternative method of measurement and specs
that the group agreed was correct.
Our approach was to define the step response, which if evaluated
properly, can render an accurate representation of the output
characteristics of a driver and how it implements pre-emphasis. Then we use
a definition of allowable random, and deterministic jitter that covers all
of the cases you are concerned about.
You see, the maximum jitter is even calculable, and so it is
possible to use a pre-defined data pattern with a calculable
amount of jitter, consider the effects of random jitter, and also the
variation about the compliant impulse response and you have
effectively done the same thing with less steps and comparable
accuracy.
Given our desire to get this spec done quickly as well, I believe
our current approach will yield a better solution to our challenge.
Regards,
Dan Dove
Dan
======================================
Principal Engineer, LAN PHY Technology
System Architect, Workgroup Switching
dan.dove@hp.com
hp ProCurve Networking Business
8000 Foothills Blvd. MS 5555
Roseville, CA 95747-5555
Phone: 916 785 4187
Fax : 916 785 1815
======================================
>Howard,
>
>"Howard A. Baumer" wrote:
>>
>[...]
>> [CH]> 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.
>>
>> <HAB>
>> If you could be more specific it would be helpfull. THanks
>> <HAB>
>
>Let's take your description of the spreadsheet (copied from your
>email this morning) and write it into the standard, something like...
>
>---
>
>The transmitter shall provide pre-emphasis to the NRZ data signal
>according to the normalized step response
> V(t) = (alpha) * H(t) - (alpha-1) * H(t-t0)
>where H(t) is the unit step function and t0 is one UI (320ps).
>
>The limits of compliant step response are defined in terms of
>a V(t) template based on the following parameters
>1) amplitude = first baud peak value
>2) deviation from amplitude = amount that the upper mask and lower
>mask
>are above and below the peak.
>3) alpha = amount of emphasis, where Vout = X(n)/(1+alpha) -
>X(n-1)*alpha/(1+alpha)
>4) devation from alpha = amount that alpha can vary, where the upper
>peak mask to lower plateau mask emphasis amount is above alpha by the
>deviation from alpha and likewise for the lower peak mask to upper
>plateau mask.
>5) delta in time that the lower mask starts it's first rising edge
>after
>the upper mask's first rising edge.
>6) max transistion time = time it takes the transition to go from 20%
>to
>80%, where the transition is the start of the upper mask's first
>rising
>edge to the end of the lower mask's first rising edge.
>7) min transition time = similar to the max except use the lower as
>the
>starting point and the upper as the ending point.
>8) Set the deviation in alpha and the peak such that the upper mask's
>peak to lower mask's peak is the same as the upper mask's plateau to
>the
>lower mask's plateau.
>
>[Insert table XX of min/max values for the parameters]
>
>The range of values constrained by the limits in table XX
>establish a family of permitted step responses. For any data
>sequence X(i) there exists a corresponding family of permitted
>pre-emphasized signals V(t). A compliant transmitter's output for
>any data sequence shall fall within the envelope of this
>sequence-dependent family of permitted V(t)'s.
>
>Retain Fig. 54-6 and Table 54-7 as a compliance test item.
>
>---
>
>Maybe too wordy, but the important point is to define "compliant
>behavior" for every possible data pattern, even if, in practice,
>we only test on 48A.2 LF.
>
>(I'm not sure alpha in my step-response formula matches alpha
>in your item 3), but that's just a detail to resolve.)
>
>Cheers,
> Chuck
>