Paragon-X: How is SyncE Wander Transfer Performed?

In version 10.40, the method used for Synce Wander Transfer was changed. How does the new method work and what are the changes?

 


Purpose of the Test

G.8262 defines wander transfer testing in terms of gain limits on a low-pass filter. The purpose of the test is to confirm that the wander at the input is filtered using the correct filter bandwidth and type, and that the clock does not amplify any wander due to gain peaking in the filter.

This testing can be performed using a set of tones of specified amplitude applied to the device input and then measuring the output.

For an Option 1 clock, the filter is defined in clause 10.1 of G.8262 as being low-pass with a minimum bandwidth of 1Hz; a maximum bandwidth of 10Hz with maximum gain peaking in the pass-band of 0.2dB. The specification does not define other details such as the order of the filter or the steepness of the cut-off.

For an Option 2 clock, the filter is defined in clause 10.2 of G.8262 as being low-pass with a maximum bandwidth of 0.1Hz. No minimum bandwidth is defined.

 

A transfer measurement involves applying a stimulus signal to the input of a DUT and then measuring the response on the output. The transfer characteristic of the DUT determines how the output changes for different stimulus signals.

For noise (or wander) transfer, the stimulus signal has specified characteristics: frequency and pk-pk amplitude. The pk-pk amplitude is then measured on the output of the DUT. The transfer characteristic for the specified frequency is given as gain (in dB) and is calculated from:

gain (dB) = 20.log10 (Amplout / Amplin)

To be able to measure DUT performance to the level specified in the ITU-T standards, the test equipment must be able to ensure that it can accurately measure the difference in amplitude between the applied stimulus and the measured response. This requires that the amplitude of the generated stimulus is accurate and that the response from the DUT can be accurately measured relative to that input.

This can be achieved in two ways:

  1. The test equipment generates the required stimulus and then measures that signal with no DUT present. This provides a baseline measurement that takes account of any inherent inaccuracy in signal generation and inaccuracy or noise in signal measurement. When the transfer measurement is repeated with the DUT present, then the difference in measured output is the transfer characteristic of the DUT with instrument variation removed.
    This is known as calibration.

  2. The test equipment has an absolute accuracy for both generation and measurement such that the transfer calculation can be made to the accuracy required without a calibration step. In other words, any variation in generation or measurement due to the test equipment is not significant given the accuracy required.

Calibration Changes in X.10.40

In Paragon-X software version prior to X.10.40, calibration was mandatory step prior to performing the transfer test. This results in very long test times because each frequency point is effectively tested twice: once for calibration and once with the DUT in place.

In X.10.40, calibration has been made an optional step. The absolute accuracy of Paragon-X generation and measurement is such that the G.8262 tests can be made accurately without calibration.

For the highest accuracy possible, calibration may still be performed. This may be useful in the case where there are marginal failures. In addition, calibration values are now stored. This allows multiple wander transfer tests to be performed with only one calibration run (compared to previous versions where calibration had to be performed each time).

Testing using a Single Frequency

Using Paragon-X, a wander transfer measurement can be made using either a single frequency or a table of frequencies.

For a single frequency, the frequency and pk-pk amplitude are specified in the UI. Clicking Generate Wander starts the generation of this tone; the measurement then runs for a duration dependent on the frequency. The pk-pk output amplitude is measured across the full capture. The gain is then calculated and the result presented on the UI.

The measurement method for single sine has not changed in X.10.40.

Note that if there is any transient following the application of the single sine wander, this will be included in the measurement result i.e. there is no facility for a recovery time

Testing using a Table of Frequencies

Using Paragon-X, a wander transfer measurement can be made using either a single frequency or a table of frequencies. Details about table operation is given below.

The main changes in X.10.40 are:

  • New Enhanced Defaults for Option 1 clocks with optimised dwell times

  • New Lower Limit, with both Upper Limit and Lower Limit checking now optional

  • A Recovery Time has been added to each step

Enhanced Defaults

A new set of defaults has been added to provide a quick way to get extended frequency coverage and optimised test time. These defaults are recommended for testing Option 1 clocks to G.8262.

Limits

G.8262 does not completely specify the filter to be implemented. In order to establish performance limits, the following assumptions have been made:

  • The minimum gain in the pass-band is -3dB

  • The filter is first-order with 20dB/decade roll-off

These assumptions then allow us to set lower and upper limits as shown in the diagram below:

Recovery Time

The frequency and amplitude changes between steps in the table will cause a short transient in the DUT output due to the action of the internal filter. This is a normal response of a filter to the change in stimulus, but it will result in incorrect pk-pk amplitude measurements and so should be ignored when making measurements. Therefore a “recovery time” is used to ignore the output during the first few seconds after each change in frequency.

In X.10.40, the recovery times at different frequencies are defined in the table below:

Frequency (Hz)

Recovery Time (s)

Frequency (Hz)

Recovery Time (s)

<4Hz

50s

4Hz to 15Hz

20s

15Hz to 35Hz

5s

>35Hz

0s

For example, for any frequency less than 4Hz, the first 50 seconds of the measurement will be ignored and the pk-pk amplitude will be calculated over the remaining capture period (as defined by the Dwell Time)

Since each step has a recovery time, this also changes the minimum allowed Dwell Time for a step.

Notes:

  • Frequencies above 15Hz are not necessary for G.8262 testing. The information above is included here for reference purposes only.

  • The restriction on recovery time based on frequency is due to an internal hardware constraint in Paragon-X.