0
\$\begingroup\$

IEEE 802.3's clause 40.8.3.1 specifies a certain a MDI return loss. The specification states that this return loss shall be maintained at all times when the PHY is transmitting data or control symbols. How does one measure the return loss, when the PHY is transmitting data or control symbols?

The Gigabit Ethernet Consortium (of the IOL from the University of New Hampshire) has published a reference PMA test suite, which lists MDI Return Loss as a test (found in page 11 of the document here). The test procedure there states to put the PHY in Test Mode 4 while performing the Return Loss measurement to satisfy the "when the PHY is transmitting data or control symbols" requirement. But wouldn't this mess up the VNA S11 measurement? If the PHY is already transmitting symbols, how would the VNA differentiate between what the PHY is transmitting versus a reflection of what the VNA is sending into the channel?

\$\endgroup\$
0
\$\begingroup\$

VNAs have a signal separation stage that separates the reflected the signal from the incident signal. Without an incident signal there would be no reflection and no measurement possible.

One possibilty is the VNA is transmitting a signal at the same time the DUT is transmitting. The two signals can be differentiated as explained from a Tektronix MOI: A return loss measurement is performed by stimulating the device under test with a sine wave at the desired frequency. If the DUT is generating signals at the same frequency as the stimulus then an error in the measurement will occur unless steps are taken to minimize that error. This implies careful testing to avoid harmonic interference. This problem is sometimes overcome by injecting a signal with higher power. However, care must be taken when driving a DUT with a signal larger than 2.8Vpp. It is possible that some devices have ESD structures that will start to interfere and become nonlinear at about 5 Volts. The Tektronix SW application configuration described in this paper minimizes the error in two ways. First sine wave frequencies in the AWG are chosen so that most of them fall in between the harmonics of the DUT. Second the acquired waveform is averaged so that much of the DUT signal is greatly reduced.

\$\endgroup\$
  • \$\begingroup\$ Yes. My question is, if the PHY (my DUT) is transmitting data, how will the VNA know to differentiate between the reflected signal and the data transmitted? \$\endgroup\$ – lp66 Aug 13 '18 at 17:23
  • \$\begingroup\$ The VNA somehow has to know the power of the incident signal. It is not clear from the test procedure, but my guess it is assumed to be a certain amplitude defined by the Ethernet specification, or it was previously measured. \$\endgroup\$ – EE_socal Aug 13 '18 at 17:49

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.