[BP] FW: proposed channel model change
Forwarded on behalf of Charles Moore...
-----Original Message-----
folks,
Last week i proposed basing the channel spec on fitting to Amax, as
is done for the interference tolerance test channel. Since then i have
looked at the fit for the 23 channels in the task force data base. I
did the fits over the range F1 to F2 as shown in table 69B-1 and also
over a range which i think would be more appropriate:
1000BASE-KX 62.5MHz-750MHz (.05*Fs-.6*Fs)
10GBASE-KX4 156MHz-1.875GHz (.05*Fs-.6*Fs)
10GBASE-KR 50MHz-6.1875GHz (Fmin-.6*Fs)
I chose the values because i do not think that it is important to go all
the way up to Fs, where there is a null in the data spectrum, and for
8B10B encoded data, disparity correction starts to drop the spectrum
around Fs/20. For scrambled data the data spectrum goes all the way to
DC but Fmin is a practical limit to the available channel data.
Included are the results of my fitting.
charles
Charles Moore wrote:
> guys,
>
> Rich Mellitz and i have been discussing changing the channel model,
> to use a fit to Amax, as we are now doing for the interference
> tolerance test channel (see moore_03_1105.pdf), instead of a fit to a
> straight line.
>
> This would have several advantages:
>
> 1. It would use a common method for both the test channel and the
> working channel
> 2. It would allow us to replace three graphical tests:
> i. That the linear fit is better than Amax
> ii. That the insertion loss is always better than Amax +
something
> iii. That the ripple is better than something
> with one numeric test, that mc < 1.0, and the graphical ripple
> test. We may need an additional limit on bc.
> 3. Since the fit to Amax will be substantially better than the fit
> to a straight line, the ripple will be smaller and more
meaningful.
> This will reduce the number of false fails. We should be able
> to shrink the ripple bounds (Rich: will you figure out by how
> much?) which may also reduce false passes.
> 4. Since the fit will be especially improved at low frequency, we
> will be able to reduce f1, especially for 10GBASE-KR, which
> actually uses low frequencies more than either 1000BASE-KX or
> 10GBASE-KX4.
>
> Does anyone see any problems with this proposal?
>
> charles
>
--
|--------------------------------------------------------------------|
| Charles Moore
| Avago Technologies
| Image Solutions Division
| charles.moore@avagotech.com
| (970) 288-4561
|--------------------------------------------------------------------|
Results of applying 2 possible revised specifications to channels for 1000BASE-KX
channels F1=125MHz F2=1.25GHz F1=62.5MHz =750MHz
Mc Bc |IUD/spec|max Mc Bc |IUD/spec|max
peters_01_0605_B12_thru.s4p 0.551 -0.41dB 0.057 0.529 -0.31dB 0.054
peters_01_0605_B1_thru.s4p 0.368 -0.29dB 0.052 0.348 -0.19dB 0.027
peters_01_0605_B20_thru.s4p 0.687 -0.46dB 0.069 0.668 -0.37dB 0.066
peters_01_0605_M1_thru.s4p 0.372 -0.18dB 0.020 0.368 -0.16dB 0.016
peters_01_0605_M20_thru.s4p 0.719 -0.49dB 0.118 0.692 -0.37dB 0.064
peters_01_0605_T12_thru.s4p 0.585 -0.54dB 0.153 0.563 -0.42dB 0.123
peters_01_0605_T1_thru.s4p 0.358 -0.21dB 0.076 0.367 -0.24dB 0.062
peters_01_0605_T20_thru.s4p 0.723 -0.61dB 0.136 0.694 -0.47dB 0.099
1m_INBOUND_TRU/sj2k2g2h2_SPARS.s4p 0.946 -0.89dB 0.149 0.874 -0.56dB 0.230
1m_INBOUND_TRU/sj3k3g3h3_SPARS.s4p 0.951 -0.91dB 0.184 0.876 -0.56dB 0.226
1m_INBOUND_TRU/sj4k4g4h4_SPARS.s4p 0.953 -0.92dB 0.212 0.874 -0.55dB 0.223
1m_INBOUND_TRU/sj5k5g5h5_SPARS.s4p 0.956 -0.94dB 0.238 0.876 -0.56dB 0.223
1m_OUTBOUND_TRU/sj2k2g2h2_SPARS.s4p 0.933 -0.86dB 0.151 0.871 -0.57dB 0.236
1m_OUTBOUND_TRU/sj3k3g3h3_SPARS.s4p 0.941 -0.92dB 0.154 0.874 -0.61dB 0.233
1m_OUTBOUND_TRU/sj4k4g4h4_SPARS.s4p 0.937 -0.90dB 0.147 0.875 -0.61dB 0.235
1m_OUTBOUND_TRU/sj5k5g5h5_SPARS.s4p 0.937 -0.91dB 0.153 0.876 -0.62dB 0.236
Case1 FM 13SI 20 T D13SI L10.s4p 0.943 -0.52dB 0.151 0.923 -0.42dB 0.167
Case2 FM 13SI 20 T D13 L10.s4p 0.975 -0.55dB 0.179 0.949 -0.43dB 0.187
Case3 FM 13SI 20 T D6 L10.s4p 1.092 -0.67dB 0.227 1.042 -0.43dB 0.265
Case4 FM 13SI 20 T D13 L6.s4p 0.794 -0.43dB 0.191 0.792 -0.40dB 0.179
Case5 DS 13 10 T D13 L6.s4p 0.536 -0.27dB 0.114 0.541 -0.29dB 0.102
Case6 DS 13 10 T D13 L6.s4p 0.541 -0.34dB 0.140 0.541 -0.33dB 0.132
Case7 FM 13SI 1 T D13SI L6.s4p 0.327 -0.18dB 0.041 0.327 -0.18dB 0.036
Where an RMS fit has been made to the channel loss by the curve: Mc*Amax+Bc over the
range F1-F2
the specs will be:
Mc < 1.0
Bc no spec yet
|IUD/spce|max < 1.0
Only Case #3/Case3 FM 13SI 20 T D6 L10.s4p fails spec, gain is too low in both set of specs.
Results of applying 2 possible revised specifications to channels for 10GBASE-KX4
channels F1=312.5MHz F2=3.125GHz F1=156.25MHz F2=1.875GHz
Mc Bc |IUD/spec|max Mc Bc |IUD/spec|max
peters_01_0605_B12_thru.s4p 0.613 -0.91dB 0.510 0.547 -0.39dB 0.064
peters_01_0605_B1_thru.s4p 0.418 -0.62dB 0.249 0.379 -0.35dB 0.069
peters_01_0605_B20_thru.s4p 0.752 -0.97dB 0.320 0.685 -0.46dB 0.060
peters_01_0605_M1_thru.s4p 0.494 -0.92dB 0.547 0.401 -0.34dB 0.254
peters_01_0605_M20_thru.s4p 0.815 -1.19dB 0.551 0.732 -0.57dB 0.196
peters_01_0605_T12_thru.s4p 0.794 -2.13dB 1.169 0.605 -0.65dB 0.302
peters_01_0605_T1_thru.s4p 0.593 -1.77dB 0.646 0.430 -0.61dB 0.186
peters_01_0605_T20_thru.s4p 0.937 -2.23dB 0.946 0.745 -0.73dB 0.236
1m_INBOUND_TRU/sj2k2g2h2_SPARS.s4p 0.959 -0.83dB 0.476 1.014 -1.26dB 0.345
1m_INBOUND_TRU/sj3k3g3h3_SPARS.s4p 0.964 -0.85dB 0.596 1.013 -1.26dB 0.332
1m_INBOUND_TRU/sj4k4g4h4_SPARS.s4p 0.964 -0.86dB 0.485 1.016 -1.27dB 0.334
1m_INBOUND_TRU/sj5k5g5h5_SPARS.s4p 0.961 -0.86dB 0.368 1.002 -1.19dB 0.259
1m_OUTBOUND_TRU/sj2k2g2h2_SPARS.s4p 0.962 -0.98dB 0.277 0.974 -1.09dB 0.243
1m_OUTBOUND_TRU/sj3k3g3h3_SPARS.s4p 0.965 -1.00dB 0.358 0.974 -1.11dB 0.206
1m_OUTBOUND_TRU/sj4k4g4h4_SPARS.s4p 0.956 -0.96dB 0.270 0.971 -1.09dB 0.221
1m_OUTBOUND_TRU/sj5k5g5h5_SPARS.s4p 0.979 -1.16dB 0.213 0.964 -1.05dB 0.163
Case1 FM 13SI 20 T D13SI L10.s4p 0.992 -0.94dB 0.359 0.934 -0.48dB 0.158
Case2 FM 13SI 20 T D13 L10.s4p 0.999 -0.80dB 0.441 0.956 -0.45dB 0.208
Case3 FM 13SI 20 T D6 L10.s4p 1.117 -0.96dB 0.427 1.062 -0.51dB 0.275
Case4 FM 13SI 20 T D13 L6.s4p 0.814 -0.65dB 0.297 0.768 -0.29dB 0.209
Case5 DS 13 10 T D13 L6.s4p 0.563 -0.52dB 0.385 0.517 -0.17dB 0.119
Case6 DS 13 10 T D13 L6.s4p 0.653 -1.25dB 0.649 0.521 -0.23dB 0.149
Case7 FM 13SI 1 T D13SI L6.s4p 0.541 -1.82dB 0.676 0.337 -0.24dB 0.067
Where an RMS fit has been made to the channel loss by the curve: Mc*Amax+Bc over the
range F1-F2
the specs will be:
Mc < 1.0
Bc no spec yet
|IUD/spce|max < 1.0
The following channels fail due to low gain:
1m_INBOUND_TRU/sj2k2g2h2_SPARS.s4p F1=156.25MHz F2=1.875GHz
1m_INBOUND_TRU/sj3k3g3h3_SPARS.s4p F1=156.25MHz F2=1.875GHz
1m_INBOUND_TRU/sj4k4g4h4_SPARS.s4p F1=156.25MHz F2=1.875GHz
1m_INBOUND_TRU/sj5k5g5h5_SPARS.s4p F1=156.25MHz F2=1.875GHz
Case #3/Case3 FM 13SI 20 T D6 L10.s4p both ranges
One channel fails one spec for IUD:
peters_01_0605_T12_thru.s4p F1=312MHz F2=3.125GHz
Results of applying 2 possible revised specifications to channels for 10GBASE-KR
channels F1=1GHz F2=6GHz F1=50MHz F2=6.1875GHz
Mc Bc |IUD/spec|max Mc Bc |IUD/spec|max
peters_01_0605_B12_thru.s4p 0.590 -0.51dB 0.655 0.589 -0.53dB 0.666
peters_01_0605_B1_thru.s4p 0.424 -0.50dB 1.081 0.428 -0.56dB 1.076
peters_01_0605_B20_thru.s4p 0.720 -0.44dB 0.656 0.721 -0.51dB 0.670
peters_01_0605_M1_thru.s4p 0.509 -1.02dB 1.018 0.509 -0.97dB 1.003
peters_01_0605_M20_thru.s4p 0.835 -1.49dB 0.544 0.815 -1.11dB 0.608
peters_01_0605_T12_thru.s4p 1.331 -9.00dB 2.015 1.236 -6.44dB 4.918
peters_01_0605_T1_thru.s4p 1.107 -8.21dB 2.613 1.064 -6.41dB 4.956
peters_01_0605_T20_thru.s4p 1.443 -8.70dB 1.887 1.347 -6.20dB 4.679
1m_INBOUND_TRU/sj2k2g2h2_SPARS.s4p 0.931 -0.44dB 0.449 0.943 -0.69dB 0.501
1m_INBOUND_TRU/sj3k3g3h3_SPARS.s4p 0.944 -0.56dB 0.567 0.951 -0.73dB 0.610
1m_INBOUND_TRU/sj4k4g4h4_SPARS.s4p 0.957 -0.67dB 0.429 0.964 -0.82dB 0.466
1m_INBOUND_TRU/sj5k5g5h5_SPARS.s4p 0.925 -0.36dB 0.333 0.935 -0.60dB 0.393
1m_OUTBOUND_TRU/sj2k2g2h2_SPARS.s4p 0.865 0.21dB 0.394 0.887 -0.27dB 0.420
1m_OUTBOUND_TRU/sj3k3g3h3_SPARS.s4p 0.873 0.15dB 0.394 0.897 -0.34dB 0.458
1m_OUTBOUND_TRU/sj4k4g4h4_SPARS.s4p 0.878 0.03dB 0.333 0.899 -0.41dB 0.357
1m_OUTBOUND_TRU/sj5k5g5h5_SPARS.s4p 0.866 0.20dB 0.408 0.890 -0.30dB 0.432
Case1 FM 13SI 20 T D13SI L10.s4p 0.927 0.02dB 0.680 0.941 -0.28dB 0.683
Case2 FM 13SI 20 T D13 L10.s4p 0.977 -0.33dB 0.504 0.986 -0.50dB 0.503
Case3 FM 13SI 20 T D6 L10.s4p 1.090 -0.35dB 0.766 1.101 -0.56dB 0.770
Case4 FM 13SI 20 T D13 L6.s4p 0.791 -0.14dB 0.632 0.804 -0.37dB 0.626
Case5 DS 13 10 T D13 L6.s4p 0.590 -0.62dB 0.854 0.594 -0.63dB 0.836
Case6 DS 13 10 T D13 L6.s4p 0.884 -3.94dB 1.140 0.851 -2.96dB 2.203
Case7 FM 13SI 1 T D13SI L6.s4p 0.593 -2.82dB 0.812 0.547 -1.86dB 1.317
Where an RMS fit has been made to the channel loss by the curve: Mc*Amax+Bc over the
range F1-F2
the specs will be:
Mc < 1.0
Bc no spec yet
|IUD/spce|max < 1.0
Both sets of spec show the following channels hav too much loss:
peters_01_0605_T12_thru.s4p
peters_01_0605_T1_thru.s4p
peters_01_0605_T20_thru.s4p
Case #3/Case3 FM 13SI 20 T D6 L10.s4p
IUD specs fail both sets of specs for:
peters_01_0605_B1_thru.s4p
peters_01_0605_M1_thru.s4p
peters_01_0605_T12_thru.s4p
peters_01_0605_T1_thru.s4p
peters_01_0605_T20_thru.s4p
Case #6/Case6 DS 13 10 T D13 L6.s4p
and the wider frequency range spec shows one more channel failing IUD
Case #7/Case7 FM 13SI 1 T D13SI L6.s4p