Forums

Split Pair Detection

Started by Tom Del Rosso August 20, 2018
Crosstalk could be time domain or frequency domain, but how could TDR 
detect split pairs?


https://www.flukenetworks.com/knowledge-base/cableiq/ethernet-split-pair-error-detection-cableiq

How does CableIQ detect a split pair error?
The CableIQ takes advantage of its DSP (Digital Signal Processor) to 
look in the time domain for excessive crosstalk. The term time domain 
indicates that we are looking down the cable and regular intervals to 
see the sources of crosstalk. Split pairs exhibit a large amount of 
crosstalk. They can also be detected using time domain reflectometry or 
just plain old frequency domain measurement.


-- 


Tom Del Rosso wrote
>Crosstalk could be time domain or frequency domain, but how could TDR >detect split pairs? > > >https://www.flukenetworks.com/knowledge-base/cableiq/ethernet-split-pair-error-detection-cableiq > >How does CableIQ detect a split pair error? >The CableIQ takes advantage of its DSP (Digital Signal Processor) to >look in the time domain for excessive crosstalk. The term time domain >indicates that we are looking down the cable and regular intervals to >see the sources of crosstalk. Split pairs exhibit a large amount of >crosstalk. They can also be detected using time domain reflectometry or >just plain old frequency domain measurement.
Funny, some time ago I bought a long length of ethernet cable with connectors on ebay. The idea was to cut it into several pieces with length I needed, add my own connectors (have a nice tool for that), and at least there already were 2 connectors on the ends, so that saves time... Anyways the (very nice) tool had problems: the copper in the wires, was very thin... And the whole cable was so thin it would not clamp very well in the connectors... But while stripping I did not see twisted pairs, colors were non-standard too.. Stripped some more, no just a cable with 8 wires with different colors. YMMV Crosstalk what is crosstalk ;-) Here to one room I have ethernet over 2 pairs, and 2 analog video cameras over the other 2 (1Vpp). There is no interference from the ethernet (also a camera) on the analog camera pictures, but sometimes you see some very slight interference from the one analog channel into the other, appearing as a slow moving color bar (NTSC crystal difference frequency in the cameras). But that link uses real high quality screened ethernet cable bought locally >10 years ago. So.. anything that works goes. ;-)
On Mon, 20 Aug 2018 01:30:11 -0400, "Tom Del Rosso"
<fizzbintuesday@that-google-mail-domain.com> wrote:

>Crosstalk could be time domain or frequency domain, but how could TDR >detect split pairs?
The impedance of the two pairs in a split pair will be greater than 100 ohms. A TDR can be an impedance measuring device.
>https://www.flukenetworks.com/knowledge-base/cableiq/ethernet-split-pair-error-detection-cableiq > >How does CableIQ detect a split pair error? >The CableIQ takes advantage of its DSP (Digital Signal Processor) to >look in the time domain for excessive crosstalk. The term time domain >indicates that we are looking down the cable and regular intervals to >see the sources of crosstalk. Split pairs exhibit a large amount of >crosstalk. They can also be detected using time domain reflectometry or >just plain old frequency domain measurement.
It uses the TDR function to determine if the impedance of the pair is close to the 100 ohm spec (+/- 15%). A split pair will have a higher impedance than 100 ohm, especially with CAT6 where the pairs are physically separated more than with CAT5e. Then it checks for NEXT (near end crosstalk) or possibly the PS NEXT (power sum near end crosstalk) to any of the other pairs. The NEXT spec at 100 MHz is: CAT5 32.3 dB CAT5e 35.3 dB CAT6 44.3 dB I don't recall the numbers, but in a split pair, one split pair will couple rather nicely to the other split pair and fail the NEXT spec. I believe (not sure) that the actual decision is made from the PS NEXT calculation; <https://www.flukenetworks.com/knowledge-base/dtx-cableanalyzer/power-sum-next-ps-next> Finding where the crosstalk is coming from is the next nightmare. One could have a cable run, consisting of multiple shorter cables spliced together with RJ45 patch panels and Type 66 or 110 punchdown blocks with only one segment having split pairs. I've been there and it's not fun. Then, there's HDTDX (High Definition Time Domain Xtalk). HDTDX - DTX CableAnalyzer <https://www.flukenetworks.com/knowledge-base/dtx-cableanalyzer/hdtdx-dtx-cableanalyzer> which methinks is most useful when someone uses the wrong type of connector but everything else -- Jeff Liebermann jeffl@cruzio.com 150 Felker St #D http://www.LearnByDestroying.com Santa Cruz CA 95060 http://802.11junk.com Skype: JeffLiebermann AE6KS 831-336-2558