High percentage of Ethernet Failures - Pi 3B

[gdouglas]

I'm beginning to wonder if there's a serious issue with the Ethernet part on the 3B.

I have 8 - 3B units and of those 8, 3 of them have experienced Ethernet hardware failure. All are relatively new units (<1 year old). One died the day I received it.

While the kernel still sees the part -- and the drivers load, Ethernet link and communication always fails -- and in fact the Link LED on the pi illuminates regardless of the presence of a cable or link partner. Eth0 comes up and gets an avahi autoconf IP address.

Yes -- I've tried different OS images (I'm using the latest Raspian lite in general), cables, switches, config tweaks etc, and even hard locking switch ports to all of the various speed/duplex combinations. The symptom is always the same.

I'm looking to understand why this is happening in such high number. The USB ports all still work just fine.

Is there a process by which I can send these units back for analysis -- and/or is there any documentation regarding how or why the part (and I suspect specifically the Ethernet PHY's analog side) might fail?

Is there a known design problem / errata that might explain my experience?

Thanks in advance!

[pcmanbob]

Hi.

I think if this was a common issue then with the millions of pi3B's that have been sold we would have seen a lot of people coming to the forum for help and a lot of suppliers complaining due to the number of returns.

First I would look at the network you are connecting them to ( assuming they are all connected to one network ) to see if there is something like a bad switch or a switch with poe turned on, that could be causing these failures.

[gdouglas]

I agree with your response -- but i've 'been there done that' with cables etc!

Everything's been checked -- and in fact I quarantined my original switch just in case -- and changed over to another but it still happens.

FYI: in addition to the units I have, a partner has another, larger number of pi 3B's (out of an order of about 100pcs) that exhibit the same symptoms. Completely different network.

FYI2: Our process is:

1) Enable USB boot on the pi with a stripped down image (Sadly the 3B doesn't come with USB boot enabled)
2) Netboot the pi with a blank SD card
3) Have the pi image itself (using dd) from an imaged stored on NFS.

The units usually break between 1 and 2.

I'm mostly looking for guidance on what, from a hardware perspective, could kill the Phy in such a manner -- and could this somehow be a side-effect of enabling network booting/power/etc.

[mahjongg]

copy sent to moderators.

[dobova86]

I read somewhere, that microchip lan951x chip production has changed recently for some weakness in the industrial model due to temperature. I can’t remind where I got the news, but this was not related to Pi.

[gdouglas]

Thanks for the lan951x info. I'll look into errata. All help is greatly appreciated.

[jamesh]

Hi gdouglas,

we would be interested in getting one or two of your defective boards back for testing, as this is not a problem we have seen before.

Can you send an email to info@raspberrypi.org, FAO JamesH, and we can organise something over email rather than on here.

Thanks

James

[gdouglas]

Thanks JamesH -- email sent.

[jamesh]

Thanks JamesH -- email sent.

I've got a few days off over Easter, so don't expect a reply until next week!

[gdouglas]

Thanks for letting me know -- and enjoy your time off.

Reply when you can. I have devices ready to send.

[gdouglas]

Hi James -- hope you had a good Easter break!

I haven't seen a reply from you via email. I re-sent a message this morning.


Thanks.

[jamesh]

Hi James -- hope you had a good Easter break!

I haven't seen a reply from you via email. I re-sent a message this morning.


Thanks.

Hmm, nothing made it to me first go , I'll see what happens with the one this morning. Make sure you address it FAO James Hughes

[gdouglas]

Hi again James -- I sent the emails again ( a couple of different ways) to info@raspberrypi.org -- did you receive them?

Is there some other way we can connect?

Thanks~!

[BespokeTech]

I'm beginning to wonder if there's a serious issue with the Ethernet part on the 3B.

I have 8 - 3B units and of those 8, 3 of them have experienced Ethernet hardware failure. All are relatively new units (<1 year old). One died the day I received it.

While the kernel still sees the part -- and the drivers load, Ethernet link and communication always fails -- and in fact the Link LED on the pi illuminates regardless of the presence of a cable or link partner. Eth0 comes up and gets an avahi autoconf IP address.

Yes -- I've tried different OS images (I'm using the latest Raspian lite in general), cables, switches, config tweaks etc, and even hard locking switch ports to all of the various speed/duplex combinations. The symptom is always the same.

I'm looking to understand why this is happening in such high number. The USB ports all still work just fine.

Is there a process by which I can send these units back for analysis -- and/or is there any documentation regarding how or why the part (and I suspect specifically the Ethernet PHY's analog side) might fail?

Is there a known design problem / errata that might explain my experience?

Thanks in advance!

Hi,

Any PoE come into play here? I note you mention that you have tried different switches cables etc, but presumably that was after each one had died! Did the 3 faulty pi's all die when plugged into the same switch hardware? If so I would be interested to know the make/model of network switch used.

[jamesh]

Hi again James -- I sent the emails again ( a couple of different ways) to info@raspberrypi.org -- did you receive them?

Is there some other way we can connect?

Thanks~!

info@ is the appropratie way, I'll ask around to see what's happened to them.