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

Re: [10GBT] "EXTERNAL NOISE" Immunity



A few questions regarding measurement method, data, and HP sampling scope,

1. I assume the noise is sampled at the speed of 1GHz. According to a quick check of HP54720D (HP sampling scope you used), bandwidth is 2G with 8bit accuracy. Just want to confirm that the sampling speed.

2.  Are the numbers along time axis (for example in page 11), from 1, 244, 3889 sampling points?

3. Common to differential. How good is the setup including the transformer in terms of common to differential? I assume you found 1000BASE-T operational errors when these noises pulses occur. Is that correct?


It is a very difficult measurement. In order to make accurate ones, I think we better measure mixed mode “S” parameters with careful calibrations so that we can calculate all those effects. However, there is a hope we can simplify and get a rough idea of PAM8 and 12 comparisons without exhausting measurements. Is there any possibility that you or anyone can measure the data once more and get us EXCEL format? That probably give us frequency shape (not necessary the absolute values) and enable us to proceed.



Although not very clear from PDF file but judging from the plots, the upper frequency (3dB point) of the noise energy seems ranging from 300 to 500MHz.



Hiroshi Takatori

Keyeye Communications, Inc.


From: stds-802-3-10gbt@IEEE.ORG [mailto:stds-802-3-10gbt@IEEE.ORG] On Behalf Of Dove, Dan
Sent: Thursday, July 22, 2004 4:17 PM
Subject: Re: [10GBT] "EXTERNAL NOISE" Immunity




I have dug and dug on my computer for the original data and can't find it. I was an unfortunate victim of a computer virus a few years ago that ate all of my MS application data, but I still have the PDF file of my presentation. I believe the original data was captured in EXCEL and is thus lost to an e-vandal.


I am attaching the PDF.



-----Original Message-----
From: stds-802-3-10gbt@IEEE.ORG [mailto:stds-802-3-10gbt@IEEE.ORG]On Behalf Of hiroshi takatori
Sent: Wednesday, July 21, 2004 3:00 PM
Subject: [10GBT] "EXTERNAL NOISE" Immunity


I understand the issue pointed out in your mail on EMI. It is one of our further study points for improvement.


Now, I would like to ask your “Dove test”.

As far as I understand, you are proposing to have “Dove test” for the PAM8 and 12 comparisons. This test may be used as a field test as well if it ends up with a simple test.

In anyway, the noise pulses are,

-          Randomly spaced transient impulses of a finite amplitude noise 1~20mV,

-          The noise spacing is randomly spaced but the average mean spacing is 14 ~ 20ms (50~70Hz),


And you mentioned that you could get a spectral content of those pulses. I think that is very helpful and key data for the comparison of PAM8 and 12 in terms of the “EXTERNAL NOISE” immunity.

Please, get that for us. Thanks.


Hiroshi Takatori

Keyeye Communications, Inc.


From: stds-802-3-10gbt@IEEE.ORG [mailto:stds-802-3-10gbt@IEEE.ORG] On Behalf Of Dove, Dan
Sent: Wednesday, July 21, 2004 10:01 AM
Subject: Re: [10GBT] Request for Cat6 Emissions Characteristics


Hi Jose, All;


I agree that EMI characteristics of UTP "as a general rule" increase with frequency but I don't think its as easy as applying a simple 20log10(f) curve to the PSD. Depending on cable geometries, ground plane and component balance, the performance of a cable plant will most likely not have such a smooth function.


Therefore, its not quite so easy to say that PAM12 is better than PAM8 based strictly upon Jose's assertion. If you look at the figure below, you will see that PAM12 is higher in the range from 100MHz to 180MHz and both are above 1000BASE-T's spectrum in that region.


The FCC/CISPR don't really care if you pass their specs at 600MHz if you are failing at 180MHz.


Rather than drive a conclusion from this data, one expects the rational approach would be to determine this is an issue that requires attention/resolution prior to making *any* selection. Maybe the problem is not which PAM to use, but which cable to use? (FTP/STP?)


I will be investigating whether my employer will support some research into this area. I have access to a 10m chamber and think it would be a worthwhile investigation.


Thanks to Sailesh Rao and Vivek Telang for providing the data for this graph.