IEC 62368-1 compliance of TI IEEE 802.3bt solutions

Hi Michael,

We are currently looking into this from our colleague who is heavily involved in the standard. We will update this thread when an answer is available. Thanks!

Hi Darwin,
Many thanks for taking on the task of providing an answer to my IEC 62368-1 question.
The problem is a lot of the voltage stress testing is based on the eternal circuit transient voltage levels. Table 13 – External circuit transient voltages of IEC 62368-1 lists 1.5 kV, 10/700 as the assumed transient voltage on twisted pair conductors. But for Ethernet this makes no sense.
The 1.5 kV 10/700 is an average value that comes from world-wide measurements coordinated by the ITU-T for long distance telephone lines earth bonded at one end. Waveform dispersion along the long line causes the transient front to slow to 10 µs and the duration to increase to 700 µs. The study concludes that for rural areas the 1.5 kV value can be exceeded, at the subscriber end, about once each 5 or every thunderstorm days. Clearly, Ethernet is different as the cable runs are short distance, up to 100 m, and there isn’t an earth bond.
A get out from Table 13 is “If the external circuit transient voltages are known to be higher than indicated in Table 13, the known value shall be used.”. In this respect, both the IEEE.802.3 and the ITU-T have standardised on 2.4 kV, 1.2/50. It would certainly help if a company like TI would give some recommendations on the transient voltage levels to be used.

Hi Michael,

As the standard is not released yet I can't say it will specifically be included in the BT standard documentation. However, I can say it has been discussed.

As for these PoE systems in strict transient conditions, I recommend looking at the app notes below on how it can be approached on the PSE/PD side.

Presumably by “As the standard is not released yet I can't say it will specifically be included in the BT standard documentation” you mean IEEE 802.3bt. But as TI has been a major development player and the latest ballot has finished, TI has a good idea what will be in the actual published version. This is further evidenced by TI confidently issuing a range of ICs that are claimed to be IEEE 802.3bt compliant.

Both IEC 60950-1 and IEC 62368-1 get mentioned in the last balloted draft of IEEE 802.bt and the IEEE still adds “The resistance after the test shall be at least 2 Mohm, measured at 500 V dc”. In testing insulation withstand it is permissible to remove some shunt components, but it is not clear if these are reinstated for the insulation resistance test. The real question is how IEC 62368-1 in total is applied to the Ethernet equipment as IEC 62368-1 uses different assessment procedures than IEEE 802.3. For example, if the impulse level is 1.5 kV, IEC 62368-1 would use an AC test voltage of 1 kV rms and the DC test voltage of 1.5 kV d.c. compared to the IEEE 802.3 values 1.5 kV rms and 2.25 kV d.c.. Worked the other way around using the IEEE 802.3 1.5 kV rms and 2250 V d.c. values would drive an IEC 62368-1 impulse test of 2.25 kV. Clearly there is a lack of understanding on the relationship of a.c., d.c. and impulse voltage test values and the actual surges that occur on short lines

It’s very much up to market leaders and designers to work out Ethernet equipment solutions that are IEC 62368-1 compliant, something I hope you can help with.

I’ve seen the PDF files you mentioned before and have the following comments on TIDA-01411.

TIDA-01411 references IEC 61000-4-5 2005, yet the latest version is IEC 61000-4-5 2017. The group that maintains IEC 61000-4-5 assume that magnetic surge induction occurs in similar circuits and so the current equally divides into each conductor. For linear matched circuits this is true. But for circuits that have voltage limiters applied, often the lowest voltage limiter will conduct first and in extreme circumstances hog all the available current. In an Ethernet circuit there are typically four twisted pairs, meaning the actual current in the hogging pair can have four times the current of the IEC 61000-4-5 test. The ITU-T gets around this by having two tests; a hogging test where all the conductors are joined and surged together and a second test like the IEC 61000-4-5 to check for any common-mode to differential-mode conversion.

The 1.5 kV, 10/700 waveform came about from an ITU-T world-wide survey on long distance telephone lines. Long lines dispersed the surge waveshape slowing the rise to 10 µs and a lengthening the duration to 700 µs. The study concludes by stating that the 1.5 kV level is is likely to be exceeded once every 5 thunderstorm days. This waveform is not appropriate for short distance (100 m) Ethernet lines.

The advice that “Therefore, based on standard principle, the MOV used for the earthed PSE must have an allowable operating voltage of at least 216 VRMS when installed in an area where the nominal AC mains is less than 130 V” does not harmonise with IEEE 802.3. IEEE 802.3 requires a 500 V d.c. insulation test and that would require an MOV with an operating voltage above 400 V r.m.s to meet the insulation resistance requirement. Interestingly, the ITU-T is moving towards instituting a power contact test for 500 V d.c. insulation resistance failures as they could represent a safety hazard in the event of a power cross.

Regards
Mick

Hi Mick,

See the following snippets from the IEEE 802.3bt standard regarding IEC 62368-1:

Hi Tom,
Sorry to be late in my reply, I had a deadline to get contributions done for an upcoming standards organisation meeting.

1. Does 802.3bt meet the IEC 62368-1 safety standard for external circuits not connect to AC mains? How does IEC 62368-1 apply to ethernet equipment?

Answer: I probably overemphasised IEEE 802.3bt. My question was “does your recommended designs comply with IEC 62368-1 (Edition 3 now)”. The reason for asking this is that, being a hazard-based standard, the user is expected to interpret the expected hazards and provide by design appropriate and justifiable safeguards to prevent injury or damage. IEC 62368-1 doesn’t recognise IEEE 802.3bt, the IEEE standard isn’t a normative or bibliography item.
There are two types of ports in IEC 62368-1

mains: AC or DC power distribution system (external to the equipment) that supplies operating power to the equipment and is PS3 (PS3 means >100 W continuous so IEEE 802.3bt does not qualify)
NOTE 1 to entry: Mains include public or private utilities and, unless otherwise specified in this document, equivalent sources such as motor-driven generators and uninterruptible power supplies.

external circuit: electrical circuit that is external to the equipment and is not mains.

Thus, the Ethernet port should be evaluated for external circuit hazards and TI’s field experience knowledge is needed here in defining hazard levels and events.

2. Should we use the 1.5kV 10us/700us transient test defined by ITU-T, or 2.4kV 1.2us/50us defined by IEEE 802.3?

Answer: Although for long distance external cables the ITU-T SG 5 does use 1.5 kV, 10/700, but for Ethernet type ports the ITU-T uses 2.5 kV (basic) and 6 kV (enhanced) charge voltages on a 1.2/50-8/20 surge generator. The 6 kV setting comes from equipment manufacturer’s analysis of field ONT equipment failures in severe customer premises situations. The 2.5 kV generator charge voltage setting outputs an open-circuit voltage of about 2.4 kV and so agrees with the IEEE 802.3 2.4 kV.

3. For the IEEE 802.3 standard, how did you determine 1.5kV rms/2250Vdc for impulse testing?

This is the crazy thing that IEC 62368-1 has retained the AC and DC prescriptive test source approach of IEC 60950-1. This is unfortunate as it leads people not skilled in insulation testing cherry pick parts of IEC 62368-1 and come up with the wrong answer as in IEEE 802.3bt.

In IEC 62368-1 clause 5.4.9 Electric strength test really defines the test levels you are talking about for mains ports (this is why IEEE 802.3bt is wrong to call out a mains port as an Ethernet port). For 230 V rms mains and basic insulation the highest result from the three required evaluation methods is 2.5 kV (AC 1.77 kV rms and DC 2.5 kV). For routine test these values can be reduced to DC 2.25 kV and AC 1.6 kV, which are not far away from the IEC 60950-1 values of DC 2.25 kV and AC 1.5 kV rms. Now these levels are mainly set by peak transient voltage and for surge testing the appropriate peak voltage would be 2.5 kV (type test) and 2.25 kV (routine test).

Other things to consider is that Ethernet port has functional insulation NOT safety insulation and the main safety hazard event is AC mains power cross.

Mick,

I am following up on this thread as it has been some time since our last response. We have not forgotten about this thread and are planning to continue the conversation once our committee representative returns to the office in mid-September.

Given the technical depth of this thread, are you okay with moving this conversation to email/phone calls? Once resolution is reached, we could post a summary of the core points here as well.

Thank you for your interest in PoE products from Texas Instruments,

Thomas,

I'm fine with your idea and would like email correspondence. Not much can happen in September as there is an ITU-T meeting and I will be running 27 hours of meetings dealing with 42 topics. I'll establish contact with you and send some background material.

Mick