Hey,
I've got two SF Rush the Rock games which work great by themselves, but the link is very flakey.
<back story>
The last time I used them linked was about 2.5 years ago, and they worked perfectly. Then one hard drive died, and I never got around to replacing it, so I just played one player for the next year. Then I moved, and the cab with the bad drive sat in my garage for 1.5 years, until a few weeks ago when I moved it down with my other cab. The hard drive on the other cab also died a few months ago, so I replaced both of them with Compact Flash drives.
</back story>
They're linked together with a standard crossover cable, and what happens is about two out of three times when starting a race, the two cabs get out of sync and each cab plays by itself. I figured it was probably a bad cable, so I replaced the cable... but it made no difference. I checked/adjusted the voltage to the board, and made sure the Ethernet connectors are clean. I also swapped the hard drives to see if anything changed, but it didn't.
I found through the network test that everything claims it passes, though the network receive count on one cab is about half that of the transmit on the other, yet the other cab has the transmit and receive nearly identical. So, it seems that one board is having problems receiving, or the other is having problems transmitting.
Here's where it gets a little goofy. I decided to try putting a switch in place, thinking maybe it needed something to "condition" the line. Plugging in a D-link switch caused the one cab to still receive all packets, but the other received none, claiming it was waiting for link. Then I plugged in a Cisco switch and it seemed both were receiving close to perfectly.
So, with the Cisco switch in place, I tested the game, and the majority of the time it works perfectly, though about one out of five races it fails to link. So... I think there's gotta be some physical layer signaling problem.
Has anyone run into a problem like this, or have any ideas? My guess is maybe the Ethernet transformer (SF1012), though I think those are usually pretty robust.
Thanks,
DogP
I've got two SF Rush the Rock games which work great by themselves, but the link is very flakey.
<back story>
The last time I used them linked was about 2.5 years ago, and they worked perfectly. Then one hard drive died, and I never got around to replacing it, so I just played one player for the next year. Then I moved, and the cab with the bad drive sat in my garage for 1.5 years, until a few weeks ago when I moved it down with my other cab. The hard drive on the other cab also died a few months ago, so I replaced both of them with Compact Flash drives.
</back story>
They're linked together with a standard crossover cable, and what happens is about two out of three times when starting a race, the two cabs get out of sync and each cab plays by itself. I figured it was probably a bad cable, so I replaced the cable... but it made no difference. I checked/adjusted the voltage to the board, and made sure the Ethernet connectors are clean. I also swapped the hard drives to see if anything changed, but it didn't.
I found through the network test that everything claims it passes, though the network receive count on one cab is about half that of the transmit on the other, yet the other cab has the transmit and receive nearly identical. So, it seems that one board is having problems receiving, or the other is having problems transmitting.
Here's where it gets a little goofy. I decided to try putting a switch in place, thinking maybe it needed something to "condition" the line. Plugging in a D-link switch caused the one cab to still receive all packets, but the other received none, claiming it was waiting for link. Then I plugged in a Cisco switch and it seemed both were receiving close to perfectly.
So, with the Cisco switch in place, I tested the game, and the majority of the time it works perfectly, though about one out of five races it fails to link. So... I think there's gotta be some physical layer signaling problem.
Has anyone run into a problem like this, or have any ideas? My guess is maybe the Ethernet transformer (SF1012), though I think those are usually pretty robust.
Thanks,
DogP

