[Ethernet] bunch of bugfix and improvements

[Fede85]

Version 1.1 of the Ethernet library brings the following improvements:

• support for W5500 and W5200 Wiznet chipsets
• automatic detection of the Wiznet chipset
• SPI speed set to 14MHz, maximum allowed by the W5100 chipset
• generalized SPI_CS pin handling functions
• added getMaxSockets() method
• improved 3rd party compatibility using the SPI Extended API as a criterion of choice
• removed redundant configurations and useless variable modifiers

[cmaglie]

@ArduinoBot build this please

[ArduinoBot]

Build successful. Please test this code using one of the following:
http://downloads.arduino.cc/javaide/pull_requests/arduino-PR-2325-BUILD-56-linux32.tgz
http://downloads.arduino.cc/javaide/pull_requests/arduino-PR-2325-BUILD-56-linux64.tgz
http://downloads.arduino.cc/javaide/pull_requests/arduino-PR-2325-BUILD-56-windows.zip
http://downloads.arduino.cc/javaide/pull_requests/arduino-PR-2325-BUILD-56-macosx.zip

[cmaglie]

/cc @BAG28

I see, on another thread, that you're searching for an Ethernet library with support for W5200, may you try this one? It should support W5100/5200/5500 with auto detection. I've tested it with a W5100 and W5500 but not with a W5200.

[embeddist]

@Fede85

I will check/test "ide-1.5.x-ethernet-W5500-2" and feedback you until this week.

[BAG28]

Hello, I dowloaded the whole SW today and checked the following with Example "Webserver"
Due + W5200 board 1 = assigned adress : most of times 0.0.0.0 sometimes erratic ones, in all cases no ethernet answer
Due + W5200 board 2 = no serial print answer
Due + W5100 = KO (assigned adress stick to 255.255.255.255, even after many resets)

Uno + W5100 = OK
Uno + W5200 board 1 = assigned adress : most of times 0.0.0.0 sometimes erratic ones, in all cases no ethernet answer
Uno + W5200 board 2 = assigned adress : always 0.0.0.0

Tries with Uno + W5200 board 2 and ethernet V2.0 found on internet is workin, so I think W5200 boards is working
For information
Uno + W5200 board 2 with SD Cardinfo is working fine.
Due + W5200 board 2 with SD Cardinfo is working fine.

All tests done in same(~) conditions : ethernet cable, USB port and dvt environment : arduino-PR-2325-BUILD-56
I hope this will help. Thanks for having involved me, even if I can't help much more.

[BAG28]

Hi,
I'm still doing some trials.
I changed the ethernet cable, after seeing poor refresh of my DHCP server.
After many trials, reets and so on I got
Due + W5100 board is working fine reapeatable.
Same Sketch with Due + W5200 still not working
I observed that W5200 ship is heating-up on my two boards, and cannot touch it on board 1 after few seconds.
MAC adress is not seen on the network switch.
If I put some traces in W5200.cpp I can see that board is well detected.

[cmaglie]

@BAG28
thanks for the report! we got W5100 and W5500 working during our tests so, at least for them, we are now aligned with the results.

We didn't tried with a W5200 yet, after your test report we consider it not-working for now, unless someone else come up with a successful test.

[BAG28]

Hi,
I am trying this version.
It is better : MAC adress is recognized by my switch. Some packets are issued by W5200 on the network. When trying to connect to the board, some packets are received but no connexion.
When scanning the network with "FING" sw on android phone, IP adress and MAC adress are correctly seen, but scan services gives no services (should see port 80).

I will pursue testing, maybe with DHCP given IP adress.
If you have some suggestions for testing, please let me know. If you need further information on my network layout, or other, please do not hesitate.

[BAG28]

Hi again,
I did the test with DUE + Board 2, and I just tested on Uno + Board 1 : Same result.

[deonp]

I believe the problem with the W5200 is because CH_BASE is 0x0400 for W5100 and 0x4000 for W5200 but both currently have a fixed 0x0400. I'll leave the implementation up to you @Fede85

[diverseg]

I agree with deonp, I believe the problem with the W5200 not working is because CH_BASE is 0x0400 for W5100 and 0x4000 for W5200.
but the .h file has it fixed
static const uint16_t CH_BASE = 0x0400;
I have a W5200 and will change this to 0x4000 and try this tonight.
if it works I'll look at a better solution.

[PaulStoffregen]

Some time ago I created a modified Ethernet library that supports both W5100 & W5200 and automatically detects either. The code is here, if anyone's interested.

https://github.com/PaulStoffregen/Ethernet

One small gotcha I ran into with the W5200 chip is the reset signal really is required. I put it on pin 9. I found the W5200 can get out of sync if it hears other SPI communication and never recovers, never properly realigns by only the CS signal. If you don't use the reset signal, you can get lucky most of the time when other SPI chips aren't present. But unlike the W5100, reset before initializing the W5200 seems to be the only 100% reliable way to make the W5200 work.

[deonp]

From memory there is also some issues with MAX_SOCK_NUM which is defined in Ethernet.h
The original MAX_SOCK_NUM was removed from the 5100.h file but there were two copies.
Therefore some changes are required to Ethernet / EthernetClient / EthernetUdp / EthernetServer objects as well to suit the 4 or 8 available sockets.

[cmaglie]

@diverseg @deonp @BAG28
I've rebased the pull request and added the suggested fix for w5200.

[agdl]

This issue was moved to arduino-libraries/Ethernet#36

[agdl]

Sorry closed by mistake

[PaulStoffregen]

I believe nearly all of this is now implemented in version 2.0.0, except getMaxSockets().

Going to close this old pull request.