Flole wrote:For me the 100M Full Duplex work better, but that is not a solution, it's only a dirty workaround.
I also use a HP Switch (Model 1910-48G) but also with a simple unmanaged Switch the issue is the same.
I am in no way trying to defend Raise3D, my only intent is to get you working- if that's even possible.
The 2 known limitations are the extension wiring inside the printer to extended the RJ45 cable to the rear of the printer and then the custom pcDuino mainboard that is in fact the front panel computer. It was speculated and then more or less proven, that somehow in the design, the the board has some less than ideal signal problems with the ethernet interface. The "fix" would be a costly redesign of the board. At best, the workaround was to:
#1 eliminate the extension wiring as a source of further signal degredation by directly plugging known good cable into the mainboard port.
#2 Further limiting both sides of the interface to 100mbps to prevent the corrupt packets and eventual interface failure.
It's a really simple concept. The chip might be capable of gig speeds, but due to manufacturing and or design "limitations", the board simply is known to produce corrupt packets. Corrupt packets then cause issues in the OS and network stack when the poor little ARM processor is trying to make sense of all this noise just fills the network stack of the OS trying to get a good set of information and eventually the network stack crashes. That's why you might at first see such a connection act like it's going to work, but in real world usage it just craps out.
This also easily explains why the Wifi interface does not have the same problem. I'm not saying it's perfect either, but again, we know, it's proven, it's not been fixed to customer satsifaction, there is a limit on the Ethernet port.
Again, the simple answer is, if you connect a modern switch, hub, router, or other network device that either attempts to auto negotiate a rate, or just flat out defaults to a high rate, then the wired network connection is flooded with corrupted packets, that causes major issues and ultimately the connection fails. The workaround is to rate limit the physical interface in such a way to not invoke the level of corruption that comes at higher speeds and kind of "get by". Since this is a physical layer problem, depending on firmware, hardware, length, reflections and so forth all determines how affected any given network might be. Also being a physical layer problem, not exactly something you can fix.