0
\$\begingroup\$

I refer to this cable because they also include a test certificate (3rd icon down on left).

The cable is tested to 236.5 MHz (250 MHz nominal). Yet it's a Cat 6 cable capable of 10 Gbit/s (if length < 55m). So how is that possible given the sampling theorem and problems with anti-aliasing? I would've expected \$ \ngtr \$ 125 Mbit/s. Clearly it is, so is the 236.5 MHz something else?

Cat 5 cable 100MHz bandwidth meaning and How are 200 GBit/s over twin-axial copper cable achieved? (esp. last para of answer) have only deepened my confusion.

\$\endgroup\$
1
  • \$\begingroup\$ Symbol rate vs data rate. \$\endgroup\$
    – Andy aka
    Jul 18 at 15:27
3
\$\begingroup\$

10GBASE-T uses all four pairs, transmitting 800 megasymbols per second, using PAM-16 to send about 4 bits per symbol (4 recuded to 3.125 bits per symbol).

4 lanes * 800 megasymbols/second * 3.125 bits/symbol = 10 gbps.

Yes, it uses bandwidth exceeding the rated specs of the wiring, but still CAT 6A can provide full 100 meters, while CAT 6 cable only goes up to 55 meters.

\$\endgroup\$
2
  • \$\begingroup\$ But if you're sending 800 megasymbols/second at ~16 stepwise levels of discrete amplitude, there'll be harmonics well into the GHz band. What's the point of testing the cable at a nominal 250 MHz then? \$\endgroup\$
    – Paul Uszak
    Jul 18 at 19:19
  • \$\begingroup\$ @PaulUszak I don't know how 10Gbps Ethernet works at that level, but why do you expect the stepwise levels to be sent without proper pulse shaping - I expect there is some serious signal processing going on to pull that off. And cables have specs for cables, they are used communication channels with certain parameters and requirements. The Ethernet specs are just made to require a certain level of communications channel, and designed to work with it. \$\endgroup\$
    – Justme
    Jul 18 at 20:14
2
\$\begingroup\$

Gigabit ethernet cables use the four twisted pairs to increase the datarate. Additionally the encoding of the data has a role in how many bits can be transferred per second. It is not just a one to one relationship.

More information on PAM (pulse amplitude modulation) used for gigabit ethernet can be found here: PAM on Wikipedia

\$\endgroup\$
4
  • \$\begingroup\$ Gigabit ethernet cables use the four twisted pairs to increase the datarate., um no. The twisted pair cabling is necessary to transport the 250 MHz of bandwidth; there's no "increasing" of anything going on here. \$\endgroup\$ Jul 18 at 16:13
  • \$\begingroup\$ @MarcusMüller, Using four pairs increases the aggregate data rate compared to using only one pair (as was done in earlier generations of Ethernet). \$\endgroup\$
    – The Photon
    Jul 18 at 16:40
  • \$\begingroup\$ Besides the answer is about gigabit signaling, which is very different from 10 gigabit ethernet. \$\endgroup\$
    – Justme
    Jul 18 at 16:56
  • \$\begingroup\$ I was stupid: I didn't read the "four" in "four twisted pairs": My apologies. \$\endgroup\$ Jul 18 at 18:15
1
\$\begingroup\$

250 MHz is not the carrier frequency, it is the test bandwidth.

So, within 250 MHz, you can transport up to \$C=250\,\text{MHz}\log_2(1+\text{SNR})\$ bits per second. That doesn't need more symbols per second - it simply needs more bits per symbol.

How are 200 GBit/s over twin-axial copper cable achieved?

Twin-Axial means a shielded waveguide, not just twisted pair, and such cables have bandwidths in the multiple Gigahertz, not just 100 or 250 MHz. These have nothing to do with Cat.6 cabling.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.