Paragon-neo/100G: Jitter measurements when connected back-to-back

Owned by Billy Marshall
Last updated: Apr 05, 2019
2 min read

When connected back-to-back with no jitter being applied, what jitter measurement result should I expect?

If you are connected back-to-back with no jitter being applied, then you should expect to see a long term Pk-Pk result of around 0.6UI.

When connected back-to-back with default single sine being applied, why is the result >8UI?

Because we are measuring jitter generated by the instrument, we need to consider two things – the jitter measurement, and jitter tolerance (i.e. jitter generated by the instrument)

Jitter Measurement

Jitter measurement is designed to measure the jitter generated by an EUT to the limits defined by the ITU in specification G.8262 Section 8.3. The range in the specification is:

Lane Speed

Jitter Frequency Range (Hz)

Pk-Pk Amplitude Limit (UI)

10G

20kHz to 80MHz

0.50UI

25G

20kHz to 200MHz

1.2U

The jitter measurement is able to measure up to 3UI accurately. When jitter is above 3UI we start losing measurement accuracy, until we saturate completely at 8.54UI.

Jitter Tolerance

The jitter tolerance function is designed to test how well an EUT can handle jitter (generated by the instrument), as described in section 9.2 of G.8262

Lane Speed

Jitter Frequency Range (Hz)

Max Amplitude (UI)

10G

10Hz to 40kHz

2,488UI

25G

10Hz to 100kHz

6,445UI

(Note that the formula that relates jitter frequency range and maximum amplitude is quite complex – the table above lists the the maximum jitter amplitude that a device is expected to be able to handle)

We are able to create jitter that is twice the level that the specification requires – we can generate up to 4,976UI on 10G lanes, and 12,890UI on 25G lanes.

Summary

So, when you compare these two tables, you can see that the instrument can generate jitter that it cannot measure – this is different from what you may be used to when testing wander.

The default values for QSFP28 jitter are 10Hz and 6,445UI (it is four x 25G lanes). We cannot measure this for two reasons:

  • The 10Hz frequency is outside of the ITU’s filter bandwidth of 20kHz to 200MHz
  • The 6,445UI is higher than our saturation value of 8.54UI

That is why you see 8.54UI – it is our “saturation” value, the highest that the instrument can possibly report.

Related articles



On this page: