I would like to implement the 100Mbps single pair automotive ethernet specified in 802.3bw, otherwise known as 100BASE-T1. I was told that in order to troubleshoot and test on this standard, the scope must be capable of 1 Ghz. Is this accurate? Why would this be necessary as opposed to an oscilloscope that is well above 100MHz such as one that is 200MHz, but still 1 GSa/s ?

  • 1
    \$\begingroup\$ What type of analysis do you want to do? Eye diagrams or just something like network traffic analysis? \$\endgroup\$ – Ron Beyer Jan 14 at 20:56
  • 2
    \$\begingroup\$ I was told that in order to troubleshoot and test on this standard, the scope must be capable of 1 Ghz So you didn't immediately ask why? Also write 100Mbps and 200 MHz, as m = milli meaning 1/1000th while M = Mega = 1,000,000 \$\endgroup\$ – Bimpelrekkie Jan 14 at 20:56
  • \$\begingroup\$ Only looking for network traffic analysis at this point. Is there a limitation for either type of analysis given much slower specifications? As for why I wouldn't immediately ask why - it's because I read it on an online article written on behalf of a supplier of expensive oscopes. I didn't want to specifically call them out, but it seemed like their reasoning may have been a bit biased. Will make that correction - thank you. \$\endgroup\$ – catmandrew Jan 14 at 21:05
  • \$\begingroup\$ The 100Base-T1 seems to use PAM3 to send 3 bits as group of two PAM3 symbols. Bandwidth requirement from cable is about 66 MHz (as per wikipedia). So you could use any scope fast enough to see the waveforms. The 1 GHz scope is needed for making precise enough measurements to see if the device passes compliance tests required for electrical parameters or fails them so you can't sell it. \$\endgroup\$ – Justme Jan 14 at 22:36
  • \$\begingroup\$ Thank you @Justme - that definitely clarifies the requirement discrepancy between compliance testing and basic troubleshooting. \$\endgroup\$ – catmandrew Jan 14 at 23:10

Sounds like you want a protocol tester more than you want an oscilloscope. But oh well, that's not the question here.

I was told that in order to troubleshoot and test on this standard, the scope must be capable of 1 GHz

No. Even wikipedia will tell you that it only requires CAT3 cabling, and that doesn't guarantee a bandwidth even close to that.

Ah, with your comment:

Only looking for network traffic analysis at this point

yeah, well, then an oscilloscope in itself is no use at all. You'll need a device that speaks that protocol. For some higher-end, there might be add-ons that implement such a protocol decoder. But that again has nothing to do with the scope's 1 GHz bandwidth or not – it's something that extends the signal processing capabilities of the scope, and "converts" it into a protocol analyzer.

Since this is an ethernet standard, in principle, a network card connecting your device A to a PC, which sniffs and forwards the packets in both directions, and another network card for PC <-> device B would sound like a saner investment than an oscilloscope.

  • \$\begingroup\$ That makes sense. I hadn't thought of that as an option. \$\endgroup\$ – catmandrew Jan 14 at 21:32
  • \$\begingroup\$ @asdfex 100Base-T1 is not good old ethernet, and there's no "good old Ethernet hubs" for that (as it's incompatible with the very basic operational principles of that type of Ethernet) \$\endgroup\$ – Marcus Müller Jan 15 at 16:55
  • \$\begingroup\$ "Even wikipedia will tell you that it only requires CAT3 cabling, and that doesn't guarantee a bandwidth even close to that." - Where are you reading that? Wikipedia doesn't mention required cable category, but does state the cabling should support at least 66 MHz. CAT-3 only supports up to 16 MHz, so it doesn't look like you can run 100BASE-T1 on CAT-3. And to do conformance analysis etc, a 1 GHz scope doesn't seem too wild a requirement. \$\endgroup\$ – marcelm Jan 15 at 16:58
  • \$\begingroup\$ @MarcusMüller Right, I mixed up TX and T1. But Cat3 doesn't fit for T1 (should be around 33 MHz actual BW), only for T2/T4. \$\endgroup\$ – asdfex Jan 15 at 17:26

There are two aspects to Ethernet connectivity: signal integrity and content. You’d use an oscilloscope with suitable differential probes to capture the raw signals and analyze them with (usually) an Ethernet signal analysis package on the oscilloscope. That’s probably over $10k worth of equipment - if you aren’t in that line of work, you’d do a rental. Without that, you won’t probably be able to verify that the signal is fully within specs since you won’t have bit error rate measurements and such, but at least you can get an eye diagram and make sure it fits the masks prescribed for the physical layer you chose. For 100BASE-T-anything I’d probably look for a 5Gs/s at least - otherwise there are way too few samples per clock period to make it convincing (or even usable if you are diagnosing signal integrity issues vs. merely confirming that things work).

The content aspect is taken care of by wireshark and a PC :)

  • 1
    \$\begingroup\$ why on earth would you need 5 GS/s oscilloscope to analyze a 33 Mbd signal? That's oversampling by a factor of 75; nobody needs that to do signal integrity analysis. This isn't Gigabit Ethernet or something, it's 100Base-T1, which has electrically very little to do with the standard computer Ethernet you're thinking of. \$\endgroup\$ – Marcus Müller Jan 15 at 16:57
  • \$\begingroup\$ I guess I didn't look into 100Base-T1 :) Something new. \$\endgroup\$ – Kuba hasn't forgotten Monica Jan 15 at 19:48
  • \$\begingroup\$ It's a 33Mbd signal, yes, so there's 15ns between transitions. 15 samples (at 1Gs/s) is maybe OK to see whether the eye is good enough, although it can be a bit painful at much higher bandwidths. Maybe 2Gs/s would be adequate (30 samples between transitions). \$\endgroup\$ – Kuba hasn't forgotten Monica Jan 15 at 19:56

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.