Ethernet magjack problem maintaining standards between POE and LED pads

[HVManiacc]

Good morning eeVblog fans

I am new to your forum but I have always been curious about all technology. Recently I have been learning a lot about designing devices and I decided to try to design something of my own.

A problem that you have probably seen hundreds of times but I cannot find the answer to it on the "big internet"

That's enough introduction.

When designing a device using an RJ45 port with a built-in transformer and POE taps, I encountered a problem that I am unable to maintain the 1.5kV isolation standard between the sides. I do not understand why the manufacturer created a connector that does not meet these IEC 60950-1:2001 standards.
According to what I have read, there are two test methods

a) 1500 Vrms steady-state at 50-60 Hz for 60 seconds, applied as specified in subclause 6.2 of IEC60950-1:2001.
b) An impulse test consisting of a 1500 V, 10/700 μs waveform, applied 10 times, with a 60 second interval between pulses, applied as specified in subclause 6.2 of IEC 60950-1:2001.

The connector model is BEL SI-52003-F
I have marked the place I have concerns about in blue on the attachment. The gap between the connections is close to 1 mm

that is not enough to meet IEC standards.

According to what I have read, there should be at least 2 mm there.

To sum up, I have two more questions.

1) What should I do to maintain IEC standards?

2) If I cut a gap in the place of the "LED" pins so that they hang in the air and the gap is >2 mm, will it then comply with the standards? Not everything can be solved this way, but this is the only idea I have.

3) Am I right that there should be at least 2mm between the POE pins and the system side (I'm attaching a photo of IEC62368 which shows that there should be at least 1.8mm for 1500V)

4) I know that there are people among you who have been designing for decades so how do you deal with such problems? Are there any good practices?

Thank you for any help.
If I didn't add the topic to the correct section, I apologize.

Best regards HVManiacc

[ajb]

Welcome to the forum!

I have marked the place I have concerns about in blue on the attachment. The gap between the connections is close to 1 mm

that is not enough to meet IEC standards.

It's enough to meet the basic isolation level per your chart, and that's probably fine here.  There are basically four levels of isolation (five if you count "not isolated"): Functional, Basic, Double, and Reinforced.  Ethernet requires functional isolation to prevent current flowing between devices in a network as a result of normal offsets in ground voltage between locations.  This is to protect the devices from damage rather than user safety.  Basic isolation is a step up from that, and as the name suggests, provides a basic level of safety (eg between primary and secondary voltages) in an otherwise low-risk situation as well as functional isolation.  Double isolation adds a second layer of isolation for increased protection, and Reinforced isolation is meant to provide a similar level of protection in only a single layer.  So unless you have some specific reason to need the extra level of safety provided by the reinforced isolation level, you should be fine with the standard footprint.   

This article from TI has a decent overview: https://www.ti.com/lit/SSZT342

[Rafiki]

Isn't the 'working voltage' in such standards usually defined as the voltage which is usually applied over the isolation? Whereas the 1500V is the peak voltage the transformer is tested to withstand.

Not sure about that so I welcome this discussion. But to my understanding the 'working voltage' over a Ethernet transformer is a few volts. Where I think in such cases the standards define a minimum 'working voltage' being at least the supply voltage of the device. Maybe 48V-60V when powered by this (POE) or maybe 230V~ when powered from AC mains. Do the 1500V peak you marked in the table really apply here?

[HVManiacc]

Isn't the 'working voltage' in such standards usually defined as the voltage which is usually applied over the isolation? Whereas the 1500V is the peak voltage the transformer is tested to withstand.

Not sure about that so I welcome this discussion. But to my understanding the 'working voltage' over a Ethernet transformer is a few volts. Where I think in such cases the standards define a minimum 'working voltage' being at least the supply voltage of the device. Maybe 48V-60V when powered by this (POE) or maybe 230V~ when powered from AC mains. Do the 1500V peak you marked in the table really apply here?

As far as I know, the "working voltage" you describe is different from the "insulation voltage".
In the case of the first one, it is within the paths and components working in the same zone, e.g. between the pins and POE paths.
Let's call it the primary side.

The insulation voltage will be between the primary and secondary sides.
I have attached a photo that will show it better
blue secondary zone
red primary zone.

Ethernet requires functional isolation to prevent current flowing between devices in a network as a result of normal offsets in ground voltage between locations.

Are you sure? I read that "Functional" isolation can only be used when the user does not have access to any conductive part of the device. That is, the system is completely isolated, e.g. with a plastic casing.
Then it takes over the protection function.

[Siwastaja]

Ethernet isolation is functional only! In other words, you are definitely not allowed to design a product which uses an unisolated (e.g., capacitive dropper) power supply from mains with Ethernet connector facing the outside world, relying on its isolation.

And for functional isolation, there are no standards to follow, because it has nothing to do with safety; from safety perspective it's not isolation at all, it is "as if" there is galvanic connection. Really not different from e.g. RS232 in that sense. Your product of course needs to pass EMC, ESD and so on.

[ajb]

Ethernet requires functional isolation to prevent current flowing between devices in a network as a result of normal offsets in ground voltage between locations.

Are you sure? I read that "Functional" isolation can only be used when the user does not have access to any conductive part of the device. That is, the system is completely isolated, e.g. with a plastic casing.
Then it takes over the protection function.

The main factor is what the consequences of an isolation failure would be.  If the primary-secondary isolation in your mains-powered phone charger fails, then the output would expose the user to hazardous mains voltages; therefore it requires at least basic isolation to protect the user.  But if the isolation on your Ethernet link fails then the device might stop working or even be permanently damaged, but probably nothing else will happen.  Certainly it would not endanger the user, since even PoE voltages are not really considered a safety hazard; therefore the only isolation requirement is functional. 

Isn't the 'working voltage' in such standards usually defined as the voltage which is usually applied over the isolation? Whereas the 1500V is the peak voltage the transformer is tested to withstand.

Not sure about that so I welcome this discussion. But to my understanding the 'working voltage' over a Ethernet transformer is a few volts. Where I think in such cases the standards define a minimum 'working voltage' being at least the supply voltage of the device. Maybe 48V-60V when powered by this (POE) or maybe 230V~ when powered from AC mains. Do the 1500V peak you marked in the table really apply here?

The "Working Voltage" would typically be the voltage (peak, in this case) expected across the isolation boundary in question during normal operation.  The voltage between primary and secondary sides of an Ethernet transformer isn't really defined, because it's just whatever results from differences in ground potential between the two ends of the cable.  To the extent that the supply voltage matters, it's the supply voltage on the PHY side of the magnetics that matters, not the main supply to the box.  A mains-powered device will generally need isolation between the primary and secondary sides of whatever reduces that mains voltage down to the ~1-5V that powers the Ethernet circuitry -- THAT is the isolation that matters to user safety, and the isolation does need to be designed for at least 240VAC (or whatever the mains voltage is).  If there is no isolation between mains and low voltage, like in Siwastaja's example of a capacitive dropper power supply, then you would absolutely need to provide suitable isolation between the secondary circuitry and the external Ethernet port -- and any other circuit terminals the user can access, like USB ports, for that matter. 

[HVManiacc]

I decided to remove the pads from the right LED in the RJ45 Connector, this way I gained space. Now I have >2.5mm of clearance between both the POE and PHY paths
The same between the connector housing and the components.

Thanks for your help.