[From QnA]Two WIZ810MJ connected crossover cable for ping

Question

I am a professional test developer, trying to use two WIZ810MJ modules for passive loop-back production tests. I am using one module as a master (sends a ping and checks the response), and the other as a slave (responds to a ping from the master). Basically, if two modules are connected via a conventional network switch the modules start up (Magjack LEDs active) and the ping works perfectly. However, if I directly connect the same two modules via crossed network cable after powering them up, the modules do not ‘wake up’ (no activity on the Magjack LEDs) and the ping test fails. I must first connect the modules to the network switch to wake them up, for the ping to then work through direct connection. Please note that I do not have the option to connect the modules via a switch - they must work the first time they are powered up and directly connected by crossed network connection. If possible I would prefer to leave the software unchanged and physically modify the module(s) if possible. Please advise what I must do to modify the modules for ping tests with crossed network cable. Thank you. Brendan

wiznetmuseum · Answer

Dear customer, Without interfacing MCU, the two LED of WIZ810MJ must be activated if the power (3.3V) and reset signal were successfully supplied. (Link LED must be ON, ACT LED must be blinking). So please check the power line (3.3V, GND) and reset signal. The reset signal timing must be ‘low’ more than 2us and set to ‘high’ and maintains ‘high’. For more detailed information about resset signal, please refer to the ‘AC characteristics Reset Timing’ of W5100 datasheet. Thanks, WIZnet

wiznetmuseum · Answer

Thank you for your reply. Following investigations and hard-wire modification to the WIZ810MJ board, the problem was solved. The problem appears to be that two W5100 chips will not negotiate with each other in ‘Auto’ mode, but this particular board is actually hard-wired for Auto mode (all mode pins tied directly to ground) with no user control over this. We fixed this issue by lifting the OPMODE-2 pin 65 and tying it to rail, forcing the chip to 100Base-T Full-Duplex. Then, when connected to the second (un-modified) WIZ810MJ with crossed cable, there was no problem at-all and everything immediately worked repeatedly with 100% reliability. Before this modification, initiating a ping request could not wake the chip at-all and communication was impossible. I believe that this completely rules out any possible problem with power supplies or reset, as nothing else was changed other than taking OPMODE-2 high. This was a challenging modification to the fine-pitch QFP, requiring professional skill and equipment. Can I suggest that future iterations of this board take the mode-setting pins to available unused pins on the header connector, enabling two boards to be connected back-to-back where required (e.g. data extenders, etc) without need of board modification or intervening network to wake the chip first. Thank you for your time and support. Kind regards, Brendan

wiznetmuseum · Answer

Dear customer, Thanks for your interest in wiz810mj. I want to request test and test result. When OPMODE2-0 change by bit, I want to know the FDXLED(pin70), SPDLED(pin67)and LINKLED(pin66) status. thanks, WIZnet