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

[802.3_SPEP2P] AW: Questions regarding Insertion Loss and Return Loss proposals

Dear George and all


Thanks for all the questions and there might be more. I had today already a very good discussion with Stephan Schreiner, and you mentioned most points we have discussed. I have added as bullets below to not mix up/ interfere with the numbering of George. If something is missing, anybody can come back to request more clarifications or explanations, therefore we have to reflector to improve the understanding between the meetings.


  1. Return Loss (slide 4)

You are right. To be honest, I don’t know how this could happen as I had many relevant standards in front of me and obviously in Excel we see the correct values, e.g., RL=9+10*f (from 0.1 – 1 MHz, nothing like log should be therein, but the same error is already in the copy from the 802.3cg standard). We will discuss who will do which correction and send you an updated file.

  1. Insertion loss (slide 10)

The factor in front, the first digit, is as you have mentioned to scale the length to 500m, 300m and 100m. In a link we have a large portion, which is fixed installed and in between or at the end patchcords, consolidation point cords and many other names are used describing the same. Such cords have normally a derating of 50%, means 50% higher insertion loss per meter. In the proposed example we have for all lengths a total length of 10m of patchcords. The resulting fixed portion is then 490m, 290m and 90m summing up to the specified lengths with the patchcords.

In the proposed case we are describing an AWG23 cable for the fixed part of the link and an AWG26 stranded cable for the patchcords. The question if it should be constant is a very good one. The total length of all patchcords summed up is proposed as 10m, however, it can be more of course. This is normally covered in the respective standards by explaining how to calculate the resulting fixed portion if the sum of all patchcords is 20m, 30m or 50m.

As reference the 802.3cg standard was used which is slightly below 60dB at 20MHz. As we have not fixed the upper frequency yet, the yellow proposal for 300m is close to 60dB at 100MHz, they grey close to 60dB at 80MHz. Both are below the cg value if we decide for 60MHz upper frequency.

As last parameter we must fix the number of connectors, here we have 5 for the 500m link and 4 for the 300m and the 100m link.

  1. Insertion Loss for the motor control link (slide 10)

Maybe there is a misunderstanding on my side. The motor control link has the same parameters in the formula but not the first one which relates to 100m. This is represented with the lowest blue curve, reaching 20dB at 100MHz. The huge gap to the yellow and grey curve can be used to run the same application on a shorter length and the option to switch off with a specific bit the FEC to reduce the cycle time. With the much larger signal received it should be possible.

What is the difference between the building automation and the motor control link? The common mode suppression level, in our case coupling attenuation, is different. The motor control link is E3, building automation E1. The proposal for the motor control link consists of only the data portion of the cable. Of course, there is also a power portion depending on the needed power for the motor. It can be AC or DC, 2poles up to 5 poles (3P+N+PE). Maybe Bernd Hormeyer can make a follow up presentation on the different options which will be covered by the connector and the respective power transmitted to the motor.

  1. Wire diameter for process automation (slide 10)

I have used AWG18 for the grey curve that is correct. A larger diameter is always possible. The idea behind is to have the same cable diameter for 10Base-T1L and 100Base-T1L. For the cable to work up to 60MHz or 100MHz the construction is the same as working up to 20MHz. An improvement in shielding I expect to end up with the same value at the maximum for 802.3dg as for 802.3cg. The same is valid for the connector, there is no additional cost because of the increasement of the maximum frequency. The connectors in the IEC 63171-x series are today, as far as I know, all capable to be assembled to an AWG18 cable, I am not aware of larger diameters for SPE connectors. With a AWG16 cable we could reach longer lengths, an option we have today also for 802.3cg. I remember that I have heard numbers of approximately 1700m for AWG14.


Discussed today with Stephan:

  • Confusion on FEC (slide 3ff)

This was corrected in version a of the presentation just uploaded by George. FEC was replaced by echo cancelling (EC).

  • Return Loss of the connector (slide 5, informative)

Do we need to add the plateaus of 30dB which are normally used as measurement plateaus? The given behaviour above 30dB is theoretical and is used for scaling sensitivity for Return Loss. Values above 30dB are close to be perfect and therefore are not contributing to the link Return Loss which is dominated by the cable behaviour.

  • Common Mode suppression (slide 8)

The 802.3cg values were taken as reference with the usual frequency dependency starting from the maximum frequency of 20MHz. Here the question about the plateaus applies too. The approach was to have the same value for 802.3dg at the maximum frequency as for 802.3cg for E3, 60dB, which is represented by the yellow graph. The grey graph for E1/E2 is 10dB below the E3 requirement.


I hope that helps. If not, just drop a line with the parameter/ slide number and a description of the issue/ question which must be further clarified/ discussed.


Best regards


Peter Fischer

Head Business Solution & Innovation


Ein Bild, das Zeichnung enthält.

Automatisch generierte Beschreibung




Von: George Zimmerman <george@xxxxxxxxxxxxxxxxxxxx>
Gesendet: Donnerstag, 13. Oktober 2022 21:28
An: STDS-802-3-SPEP2P@xxxxxxxxxxxxxxxxx
Betreff: [802.3_SPEP2P] Questions regarding Insertion Loss and Return Loss proposals


Dieter/Peter/Matthias –

Thank you for your presentation the other day.  I was in the process of looking at the effects of these on PHY parameters for a follow up, and noticed a few odd things in the proposal that needed some clarification.  I am sending this to the reflector in case others had the same questions.


All references are to slides in , since I haven’t gotten around to posting your update.

First, the return loss – the proposed values on slide 4 (Return Loss for a link) drop below zero at the low end of  the frequency range (9 + 10log(f) = -1 dB at 0.1 MHz).  Is this intended?


Second, the insertion loss has an odd form.  Each limit has a factor in front of it (5.05 for the 500m link, 3.05 for the 300m link, and 1.05 for the 100m link).  I can understand the desire to scale by length, hence the 5, 3, or 1, but what is the 0.05 for?  If it is for a percent margin, then shouldn’t that be an  additional factor of 1.0x, where x is the % margin?  Why is it a constant, .05 for all the lengths?


Third, it seems odd, and possibly incorrect to consider the 100m motor control link to have exactly the same form as the building automation link.  The cabling described by Dayin Xu for these application has a substantially different construction than two wire cabling used in building automation constructions.


Finally, the cabling you propose for process control seems substantially lossier than what Steffen had described.  I presume you are assuming 18 AWG rather than the 16 AWG Steffen based his estimates on.  Is that correct?


George Zimmerman, Ph.D.

President & Principal

CME Consulting, Inc.

Experts in Advanced PHYsical Communications




To unsubscribe from the STDS-802-3-SPEP2P list, click the following link:

To unsubscribe from the STDS-802-3-SPEP2P list, click the following link: