Re: [802.3_SPEP2P] AW: Discussion about 1GBit/s for SPE

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

(with thread reassembly)

Matthias,

I have to say that I am very wary about starting to work on 1G before we get 100M done.

That being said, here are a few thoughts:

I’m skeptical about using anything with a 100m reach because of the linkage between 100m and the BASE-T/category cabling structures. This could be a significant driver of market confusion.
I think we need to be careful about claims of cable size, as it depends on your use case. I have some samples of shielded 18/2 on my desk and it’s about the same size as a cat5e cable.
I have a general concern that if we keep changing the “reach limits” of the SPE standards with every new speed.

I know that we define link segment and that’s not the same as reach, but reach is what maps into physical reality.
In the BASE-T world, we kept the cabling infrastructure physical rules the same (100m, 6-around-1, etc, etc), even as we changed the link segment definitions to support 10M->100M->1G->10G (2.5G/5G came later).
I think this was a key contributor to the success of the BASE-T technologies, and we ignore this at our peril.

I think we need to be careful about what are the “key characteristics” of the BASE-T1 technologies, and how we stay clear of the BASE-T world. We need this for 802 “unique identity”, and we need this for market clarity. I see BASE-T1L as strongly OT focused, and noise immunity is a key part of that.

Regards
Peter

_________________________________________________________

Peter Jones Cisco Systems

Distinguished Engineer 170 West Tasman Dr.

Enterprise Networks San Jose, CA, 95134, USA.

Wrk: +1 408 525 6952 Mob: +1 408 315 8024

Email: petejone at cisco.com

Web: about.me/petergjones

Webex: cisco.webex.com/join/petejone

_________________________________________________________

From: Steffen Graber <steffen@xxxxxxxxxxxxx>
Sent: Monday, July 12, 2021 3:02 AM
To: STDS-802-3-SPEP2P@xxxxxxxxxxxxxxxxx
Subject: [802.3_SPEP2P] AW: Discussion about 1GBit/s for SPE

Hi Matthias,

for me it is not fully clear, what type of cable you plan to use. Do you just intend to use AWG22 cables as defined for PROFINET (or even thicker cables) or is the idea to also use thinner AWG26-AWG23 cables in office (IT) environments? Is it expected to also work for link segments up to 100 m using these thinner types of cable with the same immunity related to external noise?

What would be the expected noise immunity in office and especially also industrial/automotive/OT environments? Background of this question is, that using a single pair instead of 4 pairs, as e.g. for 1000BASE-T, leads to a significantly higher symbol rate (or alternatively number of needed PAM-levels, but this would likely not be the way to go) compared to 1000BASE-T. Consequently this leads to a significantly higher IL for a 100 m link segment using the same wire diameter (e.g. at half-Nyquist frequency) for a SPE solution, compared to 1000BASE-T and thus reduces the SNR at the receiver for a given external noise.

1000BASE-T1 supports 40 m based on the link segment delay spec, and based on the IL definitions the class B link segment allows for 17.21 dB @ 375 MHz (Nyquist) or 11.86 dB @ 187.5 MHz (half-Nyquist). Roughly extrapolating this to 100 m would lead to an IL of 43.03 dB (x 2.5) for 375 MHz and 29.65 dB for 187.5 MHz, which is already significant. Looking into presentation https://www.ieee802.org/3/SPEP2P/public/ISOchannelsJune%2021-schicketanz.pdf, this presentation states about 20 dB @ 100 MHz for an AWG23 cable. At 375 MHz, just looking at the skin effect, this would roughly lead to an estimated IL of SQRT(375/100) * 20 dB = 38.8 dB and 27.4 dB @ 187.5 MHz, which is with some margin pretty close to the extrapolated link segment definition for a class B segment of Clause 97. For smaller diameters, the IL increases further. Having this said, this might/will lead into the room of technical feasibility, if the same (or similar) noise requirements than e.g. for a 1000BASE-T1 automotive or industrial segment are needed.

So when thinking about increasing the link segment length for e.g. a 1000BASE-T1 equivalent modulation, the expected noise environment would be one of the key points to initially answer. Alternatively having the need for larger diameter wires to bring the IL down, leads to higher cost and larger material requirements (especially copper) again and also makes installations more difficult as pretty strong cables must be handled, which would reduce the benefits compared to a 1000BASE-T implementation using 4-pairs. So for me it would be really interesting to know more about the technical feasibility related to noise immunity and the intended cable wire diameters/IL expectations. From my point of view this would also be important related to a good distinction to the already existing PHYs with a comparable 100 m reach.

Regards,

Steffen

To unsubscribe from the STDS-802-3-SPEP2P list, click the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-SPEP2P&A=1

[pj] – thread reassembly

From: Fritsche, Matthias 00001408f68ab65f-dmarc-request@xxxxxxxxxxxxxxxxx
Sent: Monday, July 5, 2021 10:12 AM
To: STDS-802-3-SPEP2P@xxxxxxxxxxxxxxxxx
Subject: [802.3_SPEP2P] AW: Discussion about 1GBit/s for SPE

Hello together,

Yes George, my fault. Correct sentence is: “From my point of view we have now the big chance to repeat this success story based on SPE if we complete the ecosystem for the SPE protocols.”

We have a nice and complete ecosystem of SPE protocols for automotive from 10Mbs up to 10Gbs and more for short automotive link segments.

For the industry the starting point was 10BASE-T1L. But the user expectation is like "standard" Ethernet devices that work from 1Gigabit to 10Mbit and support AUTO-NEG.

Industrial Ethernet, such as Ethernet/IP, PROFINET, etc., still prefer to use 100Mbit/s and want to make the step into the Gigabit world. SPE will definitely make this easier and accelerate this development. Main advantage here from 2-pair to 1-pair with Gigabit. With standard Ethernet we need 4-pairs and this is not very nice for robust industrial applications.

If we have a look at the copper cable details, PROFINET is a good example. To keep it simple for the user, PROFINET specify always AWG 22 wire gauge. And this will fit also to all 100m SPE link segments. SPE cables for a transmission rate up to 1 Gbit/s specified at the standards IEC 61156-11 (fixed installation) and IEC 61156-12 (flexible application). Such cables are already available from many different vendors. More details for the channels are here: https://ieee802.org/3/SPEP2P/public/ISOchannelsJune%2021-schicketanz.pdf The T1-B-100 100m, 600MHz channel covers all we need from 10Mbs up to 1Gbs.

I hope this help to understand my idea

Very nice greetings and I hope we found a good consensus for this topic

Matthias

HARTING Electronics GmbH | Postfach 14 33, 32328 Espelkamp | Marienwerderstraße 3, 32339 Espelkamp | www.HARTING.com
Vertretungsberechtigte Geschäftsführer: Dipl.-Kfm. Edgar Peter Düning, Dr.-Ing. Andreas Imhoff, Dipl.-Wirtsch.-Ing. Ralf Martin Klein
Sitz der Gesellschaft: Espelkamp | Amtsgericht Bad Oeynhausen | Register-Nr. HRB 8808 | UST-Id Nr. DE815379587 | WEEE-Reg.-Nr. DE 48334311

Von: George Zimmerman <george@xxxxxxxxxxxxxxxxxxxx>
Gesendet: Freitag, 2. Juli 2021 20:04
An: STDS-802-3-SPEP2P@xxxxxxxxxxxxxxxxx
Betreff: [EXTERNAL] Re: [802.3_SPEP2P] Discussion about 1GBit/s for SPE

Matthias & Peter, thank you for your thoughts.

Matthias - please confirm, I think there was a typo – you wrote “From my point of view we have no the big chance to repeat this success story based on SPE if we complete the ecosystem for the SPE protocols.” Did you mean, “we now have a big chance to repeat…”

While there are already 802.3 single pair standards for intrasystem, this begs:

Distinct identity from 100 and 1000 BASE-T1 as well as 100BASE-TX. Personally, I think reach has to be enabling of substantial new use cases, and, related, but more important
Technical feasibility of PHYs & systems for the targeted distinct identity reach. We have heard limited discussion on this. If Power is important to broad market potential, than this needs to include the ability to transmit power.
Broad market potential relative to powering requirements, and consistent with the copper diameter used in cabling.

For these, I suggest that those doing phy studies probably need to understand the gauge of copper that would be of interest at the reach. From what you say below at relative copper mass, I think you are considering something smaller than 20 AWG copper – since 2 pairs of 24 AWG needed for 100 BASE-TX and 4 pairs of 26 AWG have lower cross sectional conductor area. If this is true, It would also, I think, be important to consider the reach of 100BASE-TX on 24 AWG or 26 AWG copper as a competitor, as well as the practical reach (not the IL limit) of 1000BASE-T on 26 AWG 4-pair copper. These are important for establishing the market potential, in addition to the phy complexity, and the level of powering that may be achieved.

These are all good things for the next stage study group to look at.

-george

If that is true, then the powering capability becomes paramount, and needs to be considered in both tech feasibility and broad market potential as well.

From: stds-802-3-spep2p@xxxxxxxxxxxxxxxxx <stds-802-3-spep2p@xxxxxxxxxxxxxxxxx> On Behalf Of Fritsche, Matthias
Sent: Friday, July 2, 2021 8:07 AM
To: STDS-802-3-SPEP2P@xxxxxxxxxxxxxxxxx
Cc: Peter Fischer <p.fischer@xxxxxx>
Subject: AW: Discussion about 1GBit/s for SPE

Hello all together,

From my point of view we have already a wide consensus for 100BASE-T1L.

I like to add some points also for 1000BASE-T1L (min. 100m) too.

The big success story of Ethernet in the last decades is based on the typical 100m link segments and the availability of PHYs and devices supporting multiple speeds from 10Mbs up to 10Gbs.

The starting point of SPE was the global car industry with the special needs. From my point of view we have no the big chance to repeat this success story based on SPE if we complete the ecosystem for the SPE protocols.

But why SPE should be extended to 1Gbs with at least 100m link segments? Why not using the existing 4-pair solutions?

With SPE 1000BASE-T1L we can address a lot of new applications for Ethernet. SPE is perfect for all mobile applications with a strong focus on weight reduction, such as transportation (shipbuilding, railway industry) and mobile machines. (See: https://ieee802.org/3/SPEP2P/public/Fritsche_3SPEP2P_01_05122021.pdf)

Compared to the 4-pair solution, SPE requires a lot less resources (copper and plastic). A good solution for a green Ethernet solution.

We see a growing number of network connections and this is also an important point we should keep in mind.

Let us discuss this topics and I hope this idea for work on 100BASE-T1L and 1000BASE-T1L on one project found a lot of supporters.

Many greetings and have a nice weekend

Matthias

Best regards / Mit freundlichen Grüßen

Matthias Fritsche
HARTING Electronics GmbH
- Senior Specialist & Global Product Manager

Ethernet Connectivity -
Marienwerderstraße 3
32339 Espelkamp - Germany
E-mail: Matthias.Fritsche@xxxxxxxxxxx
http://www.HARTING.com
Phone: +49 5772 47-9104
Fax: +49 5772 47-909104
Mobile: +49 171 3389010

Von: Peter Fischer <p.fischer@xxxxxx>
Gesendet: Donnerstag, 24. Juni 2021 15:49
An: STDS-802-3-SPEP2P@xxxxxxxxxxxxxxxxx
Betreff: [EXTERNAL] [802.3_SPEP2P] Discussion about 1GBit/s for SPE[Avira checked]

Dear all

As time was running out yesterday, we could not deeply discuss our proposal (https://www.ieee802.org/3/SPEP2P/public/Fritsche_draft_objectives_longterm_3SPEP2P_06232021.pdf), therefore as requested by George, we use the reflector. During the last meeting we have seen presentations focusing on factory automation. We completely agree that we need 100Mbit/s there. The length we suggest is 400m as there the ISO/IEC channel shows that PoDL has already some restrictions with the current diameter, the thinner cable. In the presentation of Dieter Schicketanz (https://www.ieee802.org/3/SPEP2P/public/ISOchannelsJune%2021-schicketanz.pdf) we can find a short summary about the channels/ link segments and a picture showing the difference between the two cable types currently under development at IEC.

The idea for 100Mbit/s is too short term for us. IEEE is not only making applications for factory automation but for many other industries too. Therefore, we need 1Gbit/s for industries like:

Railway and Transportation
Robotic
Building automation

IEEE needs to look also into IIoT and IoT which is expected to be in the range of billions of devices. If we miss this opportunity now, we will lose large sections of these industries which can give us all a second wave of success as seen in the late 90ties and 2000 years, where applications in a similar range were developed for 4 pairs with a lot of success.

In the case we do not start to work on 1Gbit/s now we expect that much will be lost to a bunch of wireless and other applications. Please keep in mind, the development of such applications will take 2 years and till we will see active components it can take some more time. Devices might be available in 2025. For future SPE applications we should not preclude even higher data rates. Our proposal included autonegotiation between all speeds from 10Mbit/s up to 1Gbit/s as this will increase the market success as entering at any level is future proof.

We want to hear your thoughts about this idea and discuss with you how it can be realised to bring profit to all working in these fields.

Best regards

Peter Fischer

Head Business Solution & Innovation

Tel.: +41 32 681 54 54

Direkt: +41 32 681 54 56

Email: p.fischer@xxxxxx

BKS Kabel-Service AG

Fabrikstrasse 8

4552 Derendingen

Schweiz

www.bks.ch

To unsubscribe from the STDS-802-3-SPEP2P list, click the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-SPEP2P&A=1