Precision External Phase Measurement Using the ptf 3207A

Application Note – 44

Precision External Phase Measurement Using the ptf 3207A

Introduction

Certain applications require the precise measurement of the phase of an external 1PPS (one pulse per second) signal relative to UTC (Universal Time Coordinated).  The Precise Time and Frequency, LLC ptf 3207A GNS receiver uses the Global Positioning Satellite system – GPS (or other constellation) to very accurately discipline an internal oscillator such that it produces an internal 1PPS signal that is typically within a few nano seconds of UTC.

By inputting an external 1PPS signal, the instrument is able to very accurately compare the phase of the two signals (the internally generated 1PPS and the externally derived 1PPS) and provide a phase comparison between the two.

This application note describes a method for making such measurements with a very high degree of precision using an RF clock source augmented with a precision phase interpolator.

Overview

The system elements required to make the measurement are as follows:

  • Accurate RF clock source (10MHz)
  • Digital counter to count up to 1 second with a resolution of 100 nano seconds or better (i.e. 10,000,000 steps)
  • Precision Interpolator capable of measuring up to 100 nano seconds with a resolution of 0.1 nano seconds or better (i.e. effective gain of 1,000)
  • Processing module, capable of combining the readings and initiating a regular calibration cycle.

The measurement is made in two parts. The first part is a coarse measurement based on the number of RF clock source cycles counted from a start pulse (from an internal 1PPS) to a stop pulse from an external 1PPS source. The start pulse is always exactly synchronized with the rising edge of the internal 1PPS, however the stop pulse is unsynchronized and may well occur between two successive reference pulses, therefore the second part of the measurement is to measure the error between the closest edge of the RF clock and when the actual stop pulse occurs.

System Implementation

A simplified schematic of the system is shown below in figure 1 :

The heart of the system is a low noise 10MHz oscillator. The oscillator is precisely tuned using a control signal developed by measuring the phase difference between a 1PPS derived from GPS and an internally generated 1PPS signal derived by dividing the 10MHz RF signal by 10,000,000.

The resulting RF frequency accuracy, determined by measuring the frequency offset from a precision reference such as a high performance cesium standard, is of the order of 1 E -12. As important for making the precision measurements described here, is that the noise of the oscillator in terms of phase noise (or jitter) is very small. If we are making measurements to a resolution of better than 0.1 nano seconds (100 pico seconds) we need to know that the input signals themselves are stable to this or better. In this case the jitter of the 10MHz clock is well under 1 pico second and therefore should not impact measurement accuracy.

As mentioned above, the measurement is made in two parts. First a coarse measurement is made by initiating a counter on the rising edge of the internally developed 1PPS signal that is also synchronized to UTC. The rising edge of the externally applied 1PPS is used to stop the count, resulting in a coarse count measurement.

As a 10MHz RF signal has a cycle period of 100 nano seconds, the accuracy of the coarse count measurement is limited to a resolution of 100 nano seconds. This is demonstrated by the diagram below, figure 2;

The leading edge of the 1PPS start pulse is coincident with a rising edge on the 10MHz clock. The counter then continues to count until the leading edge of the stop pulse. As the next clock leading edge is after the stop, it does not get counted, giving and error equivalent to the time from the last counted leading edge to the leading edge of the stop, in this case approximately 80 nano seconds.

In order to reduce the error, the system utilizes an interpolator to provide a “fine” measurement, as shown in figure 1.

In order to measure the phase error, the interpolator is used as shown below in figure 3:

A precision charging circuit is used to charge a capacitor for a time proportional to the coarse count error. The stop pulse is used to initiate the charging, and the next clock rising edge is used to stop the charging. A precision discharge circuit that has a much reduced rate of discharge (1000 times slower than the charge rate) is used to discharge the capacitor. A fine count is initiated coincident with the commencement of discharge, and the total time to discharge is measured, giving and effective magnification of the resolution of approximately 1000 times, equivalent to a resolution of 100 nano seconds / 1000 = 0.1 nano seconds (100 pico seconds).

The coarse measurement and the fine measurement are then added together to give the overall measurement, with a resolution of 0.1 nano seconds.

As the interpolator utilizes analog electronic components (e.g. capacitors, transistors etc.) the exact consistency can be impacted by environmental variations, in particular changes in temperature. In order to eliminate these external effects, it is necessary to calibrate the interpolator on a regular basis.

Interpolator Calibration

The interpolator converts a pulse width (representing the count error) into an analog level. The analog level is then discharged at a very slow rate, and the length of time for the discharge is measured by a counter. The response of the interpolator (measured count) for different start/stop pulse widths can therefore be represented on a simple X-Y graph. The slope of the graph represents the “gain” of the interpolator, and the offset of the graph represents the fixed offset error of the interpolator response, see the graph below, figure 4:

In order to calibrate the interpolator, both the offset and the slope of the interpolator response are characterized and compensation coefficients stored to correct any error in gain and offset. This is done by measuring the count for a 150 nano second error pulse width (c2), and then measuring the count for a 50 nano second pulse width (c1), not forgetting that there are 10 counts per nano second (resolution = 0.1 nano seconds).  The calculations are as follows;

The standard equation describing the graph is y = mx + b ,

        where y=counts, m=slope, x=error in nano seconds, and b=offset

First calculate the slope, m.       m = (c2-c1)/100

Next solve for the y intercept:   b = c2 – 100 * m   and b = c1

To calculate the time interval from counts, solve for x in terms of y:

x = (y – b)/m                   ideally: x = (y – 500)/10

In order to calculate the calibrated fine phase we use the coefficients c1 and c2 calculated from the above equations. This gives us a corrected fine phase calculation of:

fine phase error time t               t(ns) = 1000 * (counts-c1)/(c2-c1)

Note the 1000 in the previous equation. This is because 1000 counts represent 100 nano seconds (resolution is 0.1 nano seconds)

To convert to nano seconds we simply divide t by 10, and when added to the coarse phase count the result gives total phase difference to a resolution of 1 nano second.

Conclusion

In principle, the phase measurement is relatively straight forward, and measurement of fine phase by the technique described is also quite straight forward. To implement such a system however requires very careful selection of components to insure an accurate and representative result that can be impacted by many things (leakage, “on” resistance of charge circuits, temperature sensitivity and many others) that can render the result unusable.

Also, to minimize these effects and insure the highest possible accuracy, continuous calibration of the interpolator to update the correction coefficients is essential.

The new ptf 3207A instrument includes three such interpolators, two to measure the 1PPS phase error of each of the two front end GNS receivers against the internally generated 1PPS, and one additional interpolator used for external phase measurements to compare the external phase against the 1PPS UTC reference.

For questions or further information please contact : info@ptf-llc.com

Scroll to Top