Teledyne LeCroy QPHY-ENET User Manual

Page 58

QPHY-ENET Operator’s Manual Rev G

1000Base-T Slave Jitter - Mode 3 without TX_TCLK

Note : The following test must only be done after the Mode 2 without clock test (previous) is done on each pair. This is because the two results
from the Mode 2 without clock test are remembered and used for this one.

Approximating the slave jitter tests with only the MDI port available is harder than approximating the master jitter
tests. Slave jitter measurements are meant to be made with both devices connected to each other using their MDI
ports and operating normally. However, since a slave device must receive a signal at the MDI from which it
recovers the TX_TCLK, it’s not possible to measure the slave’s output jitter because each pair is bi-directional.

This procedure assumes that a device with one MDI port is put into Test Mode 3, recovers the clock from its own
master clock, and generates the Test Mode 3 signal without requiring an incoming signal at its MDI. The
procedure then proposes the following measurements:

1. Measure the unfiltered peak-to-peak jitter on the Test Mode 3 signal from the DUT, relative to an un-

jittered reference. Now, subtract the unfiltered TM2 peak-to-peak jitter.

2. Filter the Test Mode 3 jitter with the 32 kHz high-pass filter (specified in 802.3 section 40.6.1.2.5) and

measure the remaining peak-to-peak jitter. Now, subtract the Filtered TM2 peak-to-peak jitter.

3. This differs from the tests specified in 802.3, section 40.6.1.2.5 in two significant ways:

•

First, it is not possible to insert the jitter test channel (802.3, section 40.6.1.1.1) between whatever the
DUT is using as its timing reference and its recovered clock.

•

Second, in the 802.3 specified procedure the jitter between the TX_TCLK and the signal at the MDI
Jtxout is also sent into the jitter test channel by the master. Some different (and probably lower) jitter
is present if the DUT is using an internal reference as the source from which it “recovers” the clock.

Both of differences make it easier for the DUT to recover a clock. So, the limits from the test specified in
the 802.3 standard (1.4 ns and 0.4 ns, respectively) cannot be used as conformance limits. This
procedure suggests the numerical results of these slave mode tests should be reported “for purely
informational purposes without judging them on a pass/fail basis.” LeCroy follows this suggestion.

Perform the test by doing the following:

Note: Remember, you should have already run the Master jitter test on the pair you intend to measure slave jitter.

1. Make the DUT generate the Test Mode 3 signal.

2. Connect the MDI of the DUT to J64 of the TF-ENET text fixture.

3. Connect to oscilloscope channel 2 and channel 3 as described in

Table 9. Jumper configuration for

each pair measurements

(as follows) to test a given pair (A, B, C, D). See

Figure 45

for an example

connection diagram.

4. On the oscilloscope menu, select Analysis

Ethernet tests. Set up the test and press the Set Up and

Start Test button. Both filtered and unfiltered measurements are made on the same data.