[DrJarmin]

Today had a bit of an odd problem which I would like to understand a bit better.

I have cat 6 cabling with some sockets in each room, terminating at a switch. As it turned out, the electrician crimped the ends of each cable wrong (each wire pair was connected to the RJ45 straight through and not with the green wires around the blue pair (as per 586-B wiring).

The odd bit is, whilst no PC would work (ie get an IP address), the Playstation 3 would work and I could stream video from my NAS (also plugged into the switch) to the PS3.

My question is - why does the PS3 work when a PC doesn't? I am guessing that the PS3 doesn't use as many wires as the PC but would like to understand this a bit better.


Thanks!

[Heady1977]

Originally Posted by DrJarmin

...
My question is - why does the PS3 work when a PC doesn't? I am guessing that the PS3 doesn't use as many wires as the PC but would like to understand this a bit better.
...

I don't have a PS3 - but...

For 100M Ethernet ports they only use two pairs (one each for RX & TX) and expect it to be wired one way. For GigE ports they use all pairs (two each for RX & TX) and therefore, have the smarts to determine how the cable is wired.

The reason why is that Gig-E ports uses all pairs and can use MDX and MDX-I connected cables and will train against all pairs. Therefore, in most cases GigE is more forgiving than 100M Ethernet.

Some manufacturers decided that it was a nice feature to have for 100M too - so now some 100M ports will also train against all pairs.

The dirty little secret is that GigE has lower requirements on a cable than 100M from the point of view of the electrical modulation used and electrical characteristics from the signal being spread across many pairs.

In your case as long as the cable is short it shouldn't matter. The only real issue will be your straight through cable will be more susceptible to electrical interference as the pairs are not balanced anymore.

Many years ago I had GigE running (with some errors) on cable used for explosives detonation - it would not have even been Cat1.

[DrJarmin]

thanks for the detailed reply.

Sort of understand what you are saying, but I don't understand why the PS3 would be more reliable than PC's or Laptop NICs plugged into the same port?

Am I missing the point here?

[Heady1977]

Originally Posted by DrJarmin

thanks for the detailed reply.
...
but I don't understand why the PS3 would be more reliable than PC's or Laptop NICs plugged into the same port?
...

I don't have a PS3 - but a quick Google search shows the Ethernet port is a GigE port.

What ports do you have on the other devices? 100M or GigE?

If the ports are 100M then - 100M ports assume that the wiring of the cable matches the standards 586-A/B as 100M ports only use two pairs. If the pairs are incorrect then one wire from a pair might not be connected at all or both wires from a pair might not be connected at all. In the first case the signal path is now unbalanced where as it should be balanced - all this does is make the signal more susceptible to noise and interference. In the second case an "air-gap" and therefore, no signal at all.

If the ports are GigE then - GigE ports use all pairs and therefore, have to train against (test) all pairs as wiring (if correct) could be either 586-A/B (pairs could go to different pins). In the normal case the port has to detect whether the cable is 586-A/B and does so by sending pulses across each pair in sequence and listening for the result. Depending on the order of the received sequence the port can then tell if the cable is 586-A/B and can map the signals appropriately. However, this can also detect split pairs. I cannot remember if detecting split pairs is part of the IEEE standard or is a feature added by some manufacturers. If the PHY doesn't detect split pairs then all this does is typically make the signal path now unbalanced and is therefore, like the case described above for the 100M. GigE also trains against the pairs and can modify the pre-equaliser to cope with different cable responses therefore, making it a bit more robust.

You have to pay for IEEE standards and it seems that I no longer have copies... I cannot remember what the training details are for the 100M and GigE but I can safely assume that as the GigE standard came after the 100M standard the training procedure would use what was learnt from field experience with 100M.

The training sequence is fixed - but what the manufacturers do with the information that is discovered during the training sequence is their chipset "feature" and value add. Some I expect try to be very smart and try to do the best possible under as many situations as possible. However, the cheaper end of the scale would only do what is required and would only typically work in the normal case.

So to answer your original question: it would appear that the GigE chips on the PS3 have features that allow for oddball wiring and are better quality and most probably more expensive then some of your other devices.

[Sborrone]

I think I might be barking up the wrong tree here but thought this might help?

I could be wrong but it almost sounds as if the installer turned your "straight" cable to "twisted"?

For arguments sake; you want to copy data from a Laptop to a Desktop -

You can't just connect any CAT5e/6 cable as it would fail to use the correct cables and protocol would not be receive the correct responses and connection will fail (Protocols are another story)

The answer is to connect a twisted cable (twisted in a sense that a couple of cables are swapped over)

Back to your question;

The newer equipment (networking comms / computers / consoles) have the ability to compensate for the cable whereas older or cheaper items don't have the ability built in.

What's your PC network card's negotiation set too? 10/100/1000 half/full?