| Thread Links | Date Links | ||||
|---|---|---|---|---|---|
| Thread Prev | Thread Next | Thread Index | Date Prev | Date Next | Date Index |
|
Hi Ragnar, Thanks for continuing to discuss and build understanding on this topic.
One concern I have with analyzing the probability of an RL-violating cable harness as in the slides is that it seems to take the approach that only harnesses in which all 5
cable segments exceed +/-3ohms with alternating usage are the ones that will require the relaxed RL limit. If it’s also possible to use 3 cable segments that exceed +/-3 ohms to construct a violating harness, or if the first 3 cable segments exceeding +/-3
ohms combined with any other 2 randomly chosen cable segments could violate then instead of having a “^5” it becomes “^3” and now instead of 1 bad cable assembly per 427 years we would see almost 100 per year. This would still be <1ppm, but it would be much
nicer to have something that is effectively 0ppm. If there are no RL limit violations in cable assemblies with 3 or fewer cables segments that are >+/-3 ohms combined with 0, 1, or 2 other freely chosen cable segments, then
any probability is acceptable, but unfortunately the resulting RL in these scenarios is not obvious to me. It would be reassuring to confirm that the more common cases of cable assemblies essentially cannot violate the limits even with “outlier” cables/connectors
and that the less common cable assemblies are very unlikely to violate (which you have already demonstrated to some extent). Another concern is on the temperature. It seems reasonable to assume that where 2 pieces of cable meet they are at the same temperature, but I think that temperature gradient
may still be a concern. I would expect that because of the way vehicles are constructed that a connector or cable segment could be passing from one compartment to another and different compartments could have drastically different thermal environments. So
the temperature gradient in the cable assembly could be gradual or extreme, at the connector or not at the connector, unless there are rules/standards which are in place to overcome this. How the thermal environment translates into temperature at each point
in the cable assembly is complex. I don’t intend to say whether this is important to the RL limit or not, just that ignoring temperature gradient/differences would be an “aggressive” assumption, and it seems better if everything works even with “conservative”
assumptions about temperature gradients. That way the cable assembly can work under all startup and operating conditions and in all supported environments without any strange exceptions or transient effects.
Best Regards, Scott From: Ragnar Jonsson <rjonsson@xxxxxxxxxxx>
EXTERNAL EMAIL:
Do not click links or open attachments unless you know the content is safe
Hi Rich, Regarding you statement that you have seen cables outside of +/-3 Ohm limit, I believe that. My calculations explicitly assumed that cables could have more than +/-3 Ohm variations
with a given probability. The relevant question is if you have seen an actual production cable harness that only has 12.4dB return loss at frequencies below 200Mhz? Let me clarify again what I shared in
https://www.ieee802.org/3/dm/public/adhoc/022725/jonsson_3dm_01_02_27_25.pdf At no point did I state or imply that +/-3 Ohm cables did not exist. I explicitly assume the opposite, that such cables do exist and that they have a given probability of occurrence.
If you believe that 1% of all cables violate the +/- 3 Ohm limit, then on average there would be more than 400 years between observing the worst-case cable harness in production. If you believe that 1/10000
cables violate the +/- 3 Ohm limit, then on average there would be more than 4 trillion years between seeing the worst-case cable harness in production. These calculations assume that there are 750 million cable harnesses with four in-line connectors deployed each year. In reality, most cable harnesses have less than four inline connectors. This means that
the average time interval between seeing the worst-case harness is much longer than what my calculations show. In summary, my calculations assume that cables with +/-3 Ohm do exist.
The question is if cable harnesses with only 12.5dB return loss below 200MHz exist? My answer to that question is that that is highly improbable.
Ragnar From: Long, Richard J <0000451e32405de7-dmarc-request@xxxxxxxxxxxxxxxxx>
Hi Ragnar, Thank you for your comments. I do not know that the analysis, linked below, is necessarily agreeable within the cable industry. I guarantee that I have
not been working on automotive cable assemblies for 4 trillion years and I have Hi Ragnar, Thank you for your comments. I do not know that the analysis, linked below, is necessarily agreeable within the cable industry. I guarantee that I have not been working on automotive cable assemblies for 4 trillion years and I have observed
cable with impedance outside of +/- 3 Ohm. So I would argue that I am either extremely lucky or that the time frame has been slightly overestimated.
I would agree with George Zimmerman’s request to have some more input from cable experts. Unfortunately, I could understand why a cable supplier would not be interested in sharing the yield data from their
factories so I would be very surprised if this data would ever be provided to the group. At least one cable expert has offered some input, and that was in the Zerna contribution.
Additionally, the approved objectives of the dm project state 4 inline connectors so we cannot discount one or two connectors because they are “not hardly ever used”. Finally, has it been determined that bad cables are just not ever seen in reality? I would argue that there is not a good reason to design the specification to include every bad cable. However, I am still
trying to understand how corner cases should be defined, and any more data or comments on this topic are most certainly welcome.
-Rich Long From: Ragnar Jonsson <rjonsson@xxxxxxxxxxx>
Hi Rich, I would like to remind you of the analysis that I provided in https://www.ieee802.org/3/dm/public/adhoc/022725/jonsson_3dm_01_02_27_25.pdf This analysis shows that the probability of the worst-case cable impedance mismatch is extremely unlikely. In fact, my calculations showed that if 1% of cables would violate the impedance requirements, there
would still be over 400 years on average between observing the worst-case echo cable. A more likely scenario would be to have more than 4 trillion years between observing such cable. To put this into context, the earth is estimated to be just over 4 billion
years old. Four inline connectors are hardly ever used in real live in the industry. It is much more common to use one or two inline connectors, with some cases of three inline connectors. This will help reduce the echo
even further. In the discussion of my presentation, it was also pointed out that for a cable to stay within the +/-3Ohm across all temperatures, the allowed variation at any one temperature is much lower. It is reasonable
to assume that both cables coming together at a connector will have essentially the same temperature at the connector. Therefore, the impedance mismatch would be less than +/-3Ohm. I agree with you that it is possible to simulate very bad impedance mismatch, and I have even shared such bad cables with the group in https://www.ieee802.org/3/dm/public/0125/jonsson_3dm_01b_01_20_25.pdf This bad cable was last used in presentation by Ahmad and Mehmet in the New Orleans meeting: https://www.ieee802.org/3/dm/public/0525/Chini_3dm_02a_0525.pdf My experience also matches yours that I have not seen any real cables show this behavior. This is not surprising given how extremely unlikely such cables are (one every 400 years to 4 trillion years). The bottom line is very simple: Cables that come close to my simulated bad cable or what was proposed in the Zerna presentation you reference below are never seen in real live. These cables are unlikely to
be ever observed in real production, even if we wait for few hundred years. Ragnar From: Long, Richard J <0000451e32405de7-dmarc-request@xxxxxxxxxxxxxxxxx>
All, The topologies presented in https: //urldefense. proofpoint. com/v2/url?u=https-3A__ieee802. org_3_dm_public_0325_boyer-5Fsharma-2D3dm-5Fxx-5F03-2D10-2D25-5FRevB. pdf&d=DwIF3g&c=nKjWec2b6R0mOyPaz7xtfQ&r=hiHgBSUj2X0k3TORVxe0NCZAlJs6SEDHhwLDz5m9MbY&m=btumaWxPxcA9-QZm2r7ogJMswJHyKLYkKv6k4pj1kUN15hLKrRIeefAx9n55OyNZ&s=29tUat1fXLUxUggBnuHh2TRssCQ5amzXo2FrdD_UF0Y&e= All,
The topologies presented in https://urldefense.proofpoint.com/v2/url?u=https-3A__ieee802.org_3_dm_public_0325_boyer-5Fsharma-2D3dm-5Fxx-5F03-2D10-2D25-5FRevB.pdf&d=DwIF3g&c=nKjWec2b6R0mOyPaz7xtfQ&r=hiHgBSUj2X0k3TORVxe0NCZAlJs6SEDHhwLDz5m9MbY&m=btumaWxPxcA9-QZm2r7ogJMswJHyKLYkKv6k4pj1kUN15hLKrRIeefAx9n55OyNZ&s=29tUat1fXLUxUggBnuHh2TRssCQ5amzXo2FrdD_UF0Y&e= not being suitable for evaluation of RL limits is agreeable to me. If you take a look at some of the past contributions, such as presented here for RL limits, see slide 11, the limits presented by Rich and Rohit would not pass:
https://urldefense.proofpoint.com/v2/url?u=https-3A__www.ieee802.org_3_dm_public_0724_Zerna-5F802.3dm-5F01b-5F240717-5FIL-5FRL-5FLimits.pdf&d=DwIF3g&c=nKjWec2b6R0mOyPaz7xtfQ&r=hiHgBSUj2X0k3TORVxe0NCZAlJs6SEDHhwLDz5m9MbY&m=btumaWxPxcA9-QZm2r7ogJMswJHyKLYkKv6k4pj1kUN15hLKrRIeefAx9n55OyNZ&s=562L611gpb4wJpwqkgdvKTW-AxCNpXjsfqLNCK2yWcg&e= Here is another work with additional topologies presented that would not pass the limits proposed by Rich and Rohit:
https://urldefense.proofpoint.com/v2/url?u=https-3A__www.ieee802.org_3_dm_public_0924_bergner-5F3dm-5F01a-5F18-5F09-5F24.pdf&d=DwIF3g&c=nKjWec2b6R0mOyPaz7xtfQ&r=hiHgBSUj2X0k3TORVxe0NCZAlJs6SEDHhwLDz5m9MbY&m=btumaWxPxcA9-QZm2r7ogJMswJHyKLYkKv6k4pj1kUN15hLKrRIeefAx9n55OyNZ&s=X2Yh2NOjSyDFNIbnpCCney7MKJjuoEjDqmpW4s4BKs8&e= Additionally keep in mind some of those topologies were created via simulation. So if you agree with those topologies, and the simulations, as being realistic then the RL limit should be relaxed. I would agree with something closer to what is presented by Zerna above.
Thanks.
-Rich Long
-----Original Message-----
From: Boyer, Rich <000012e205f410df-dmarc-request@xxxxxxxxxxxxxxxxx> Sent: Thursday, May 15, 2025 2:06 PM
To: STDS-802-3-ISAAC@xxxxxxxxxxxxxxxxx Subject: Re: [802.3_ISAAC] Questions on "boyer_sharma-3dm_xx_05-14-25_2.pdf"
[CAUTION] This email originates from outside of TE. Be cautious when clicking links, opening attachments and providing sensitive information.
Hello Conrad,
This email is from Both Rohit and Rich. We both worked on this reply.
The reason the IL changed is that we are working to a compromise in 802.3dm as a new asynchronous technology of what is feasible from the IC transceiver and what is doable by links. Our goal is to find out what is doable within reason so that is why we work and collaborate with as many people as want to work together for good balance in transceiver technology and what may come in the future of links.
For the RL, we have evaluated many different realistic topologies from measured links. What we mean by "realistic" is from our experience of implementing complete links into vehicles. We realize that there maybe cases that do not meet in the <400 MHz and those cases need to be examined as to applicability and what could be done to make them meet.
We will put together a presentation for the next 802.3dm meeting and show RL data from some link topologies.
Best Regards,
- Rohit and Rich
-----Original Message-----
From: Conrad Zerna <conrad@xxxxxxxxxxxxxx> Sent: Wednesday, May 14, 2025 4:19 PM
To: STDS-802-3-ISAAC@xxxxxxxxxxxxxxxxx Subject: [802.3_ISAAC] Questions on "boyer_sharma-3dm_xx_05-14-25_2.pdf"
CAUTION: This email comes from a non Aptiv email account Do not click links or open attachments unless you recognize the sender and know the content is safe.
Hi Rich, Rohit,
I was wondering about the specific reason, that the IL formula change from
Insertion loss(f) ≤ 0.0015 - 0.001325*f- 0.3785*√f - 1.1785/ √f (dB) in March to
Insertion loss(f) ≤ 0.0015 - 0.001325*f- 0.3645*√f - 1.1785/ √f (dB) now in May.
You said, it was due to "technology". What specific requirement/limitation of technology is that?
I also wanted to reiterate my comment from March, that the topologies in slides 3 here https://urldefense.proofpoint.com/v2/url?u=https-3A__ieee802.org_3_dm_public_0325_boyer-5Fsharma-2D3dm-5Fxx-5F03-2D10-2D25-5FRevB.pdf&d=DwIF3g&c=nKjWec2b6R0mOyPaz7xtfQ&r=hiHgBSUj2X0k3TORVxe0NCZAlJs6SEDHhwLDz5m9MbY&m=btumaWxPxcA9-QZm2r7ogJMswJHyKLYkKv6k4pj1kUN15hLKrRIeefAx9n55OyNZ&s=29tUat1fXLUxUggBnuHh2TRssCQ5amzXo2FrdD_UF0Y&e= are not suitable to validate a proposed RL limit, especially in the lower frequency range.
If you have evaluated your new RL curve from May against other topology simulations?
Thanks,
Conrad
________________________________________________________________________
To unsubscribe from the STDS-802-3-ISAAC list, click the following link: https://urldefense.proofpoint.com/v2/url?u=https-3A__listserv.ieee.org_cgi-2Dbin_wa-3FSUBED1-3DSTDS-2D802-2D3-2DISAAC-26A-3D1&d=DwIF3g&c=nKjWec2b6R0mOyPaz7xtfQ&r=hiHgBSUj2X0k3TORVxe0NCZAlJs6SEDHhwLDz5m9MbY&m=btumaWxPxcA9-QZm2r7ogJMswJHyKLYkKv6k4pj1kUN15hLKrRIeefAx9n55OyNZ&s=sHwr87kb0I0FEjRJ_BqFp1g9B67ivd7hYMtBD1MOlI0&e= ________________________________________________________________________
To unsubscribe from the STDS-802-3-ISAAC list, click the following link: https://urldefense.proofpoint.com/v2/url?u=https-3A__listserv.ieee.org_cgi-2Dbin_wa-3FSUBED1-3DSTDS-2D802-2D3-2DISAAC-26A-3D1&d=DwIF3g&c=nKjWec2b6R0mOyPaz7xtfQ&r=hiHgBSUj2X0k3TORVxe0NCZAlJs6SEDHhwLDz5m9MbY&m=btumaWxPxcA9-QZm2r7ogJMswJHyKLYkKv6k4pj1kUN15hLKrRIeefAx9n55OyNZ&s=sHwr87kb0I0FEjRJ_BqFp1g9B67ivd7hYMtBD1MOlI0&e= ________________________________________________________________________
To unsubscribe from the STDS-802-3-ISAAC list, click the following link: https://urldefense.proofpoint.com/v2/url?u=https-3A__listserv.ieee.org_cgi-2Dbin_wa-3FSUBED1-3DSTDS-2D802-2D3-2DISAAC-26A-3D1&d=DwIF3g&c=nKjWec2b6R0mOyPaz7xtfQ&r=hiHgBSUj2X0k3TORVxe0NCZAlJs6SEDHhwLDz5m9MbY&m=btumaWxPxcA9-QZm2r7ogJMswJHyKLYkKv6k4pj1kUN15hLKrRIeefAx9n55OyNZ&s=sHwr87kb0I0FEjRJ_BqFp1g9B67ivd7hYMtBD1MOlI0&e= To unsubscribe from the STDS-802-3-ISAAC list, click the following link:
https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-ISAAC&A=1 To unsubscribe from the STDS-802-3-ISAAC list, click the following link:
https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-ISAAC&A=1 To unsubscribe from the STDS-802-3-ISAAC list, click the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-ISAAC&A=1 |