Models for 100base-tx measurements or latency

Which product is right for your exact requirements
Post Reply
Posts: 1
Joined: Mon Aug 02, 2021 1:52 am

Models for 100base-tx measurements or latency

Post by nonoriri »


I am considering purchasing a pico scope.
(I've tinkered with oscilloscopes, but I'm not familiar with them..XD)
Right now I'm thinking of a "3405D MSO".
Which model would be good for the following use cases?

Q1) 100base-tx (Ethernet) signal measurement
If I want to measure a 100base-tx (Ethernet) signal, what would be the lowest bandwidth and sampling rate I need?
I want to check the data by dividing it into "preamble / SFD / packet information". (TX and RX)
For example, is a 1 GS/s sampling rate sufficient for a 100 MHz bandwidth?

Q2) Calculating the latency of the switch (port forwarding)
If I don't need to measure 100base-tx signals and just want to find the latency of a local switch (port forwarding), what would be the minimum bandwidth and sampling rate I need?
For example, the configuration shown in the following figure.
If I connect the tx of transmit and rx lines of receive, is it correct that the oscilloscope needs at least 2 CH analog channels? (4 passive probe or 2 differential probe maybe?)

Thanks for reading the this. Have a nice day!

Posts: 1034
Joined: Mon Aug 11, 2014 11:14 am

Re: Models for 100base-tx measurements or latency

Post by Gerry »

Hi nonoriri,

To answer your questions:

1/ There is a post from a member of our team that discusses the answer to your first question here: viewtopic.php?t=22511, and as you can see from the example 1GS/s is sufficient for decoding 100BASE-TX (this post also answers your questions regarding what Scope to use and minimum bandwidth).

2/ The sampling rate that you would need to determine the latency across the switch by measuring the delay between the TX and RX differential signals will entirely depend upon the precision with which you want the measurement to be made, and the quality of the signal captured (e.g. primarily how much noise is in the signal). 1GS/s would give you a precision of 1ps (if the signal was perfect, which, of course, it won't be).

3/ Ideally you should use a differential measurement if you want to minimise the noise (e.g. for your delay measurement), however, it's not easy to find a reasonably priced, differential probe for lower voltages, and a wide enough bandwidth for analysing 100-BASE-TX data (we certainly don't have any). So, as an alternative, you could:
(i) use a conventional passive probe for the non-inverting differential signal (as was done in the linked example).
(ii) use a Math Channel of A-B from 2 conventional passive probes, used to capture the non-inverting and inverting signals on Channels A and B (which may give you reduce or eliminate errors if you are getting any from (i).
(iii) search for a Differential Instrumentation Amplifier, to provide you with the noise reduced, difference signal (if the noise is too large from [ii]. Although this would be much more cost effective than a low voltage, high bandwidth, differential Probe, it would likely be as difficult to find one with a suitable bandwidth).

Option (ii) would require 4 Channels, while any other option would need just 2 Channels.


Technical Specialist

Post Reply