The significance of the EFC count

There has been considerable debate concerning the meaning of the EFC count as reported by the Z3801A. In order to clarify my thoughts and to learn more about my own receiver, I have done some experiments.

The first experiment went something like this: I synchronized a local standard to the running and stable Z3801A. This was done with some precision using a frequency comparator which by a process of multiplying up and mixing down allows the frequency difference to by magnified by 1000 times.

The Z3801A was put into holdover mode. It was then turned onto its side. This caused its frequency to move, because of gravitational error, by about 1E-9. The difference frequency was noted. Then it was taken out of holdover, and the EFC graph was monitored until it stabilized the frequency as confirmed by the comparator. The new value of EFC count was noted. The process was than repeated but this time restoring the orientation of the Z3801A to normal.

The outcome was that twelve counts of EFC, on my receiver, represented one part in ten to the eleven.

Some time passed, but the discussion still continued. So I devised another way to explore the meaning and the possible range of values of the EFC count.

The data was obtained by the following method. The 10811 clock oscillator output was disconnected from the receiver board. It was used instead to feed the external standard input of an HP 8660C synthesized generator. The generator output was connected to the 10Mhz input of the GPS receiver board. This provided the means to shift the frequency of the clock oscillator in 1 Hz increments. The GPS receiver automatically corrected for this by moving the EFC in the appropriate direction to oppose the shift introduced by the generator. The EFC voltage was measured at the oscillator. NOTE that the values at +4 Hz and -4 Hz should be disregarded as the required EFC fell outside of the range of control of the receiver. For all other values the receiver operated normally, was able to track satellites and record sensible values for EFC and TI.

During this experiment, once the receiver was locked the generator would always produce exactly 10MHz out. The receiver was offsetting the external reference of the generator in an equal and opposite direction to that of the generator's frequency setting.

Freq offset EFC value EFC percent EFC voltage Comment
-4 hz 1048560 +100 -2.060 Not able to lock
-3 hz 1041672 +99 -2.016 Locked
-1 hz 0809633 +54 -0.540 Locked
0 hz 0685230 +31 +0.236 Locked
+1 hz 0544265 +3.8 +1.135 Locked
+3 hz 0145702 -72 +3.646 Locked
+4 hz 0 -100 +4.567 Not able to lock

The results, albeit over a much greater range than the first experiment, confirm that one count of EFC represents a frequency change of slightly less than 1E-12. The actual exact sensitivity is of little importance to me since the oscillator's control characteristic is not quite linear.

The results also confirm that the EFC count is a twenty bit number, and that the relative EFC range goes from -100% to +100%, both as explained by Joe Geller.

Since the holding range appears to exceed what would be expected of a 10811E, I recorded the approximate frequency offset as a function of control voltage.

The following table shows the relationship between EFC control voltage at the oscillator and the oscillator frequency in open loop.

Efc voltage Frequency offset Freq offset normalized
0.0 +0.2 Hz 0.0 Hz
+1.0 -0.9 -1.1
+2.0 -1.9 -2.1
+3.0 -2.5 -2.7
+4.0 -3.3 -3.5
+5.0 -3.9 -4.1
-1.0 +1.6 +1.4
-2.0 +3.0 +2.8
-3.0 +4.2 +4.0
-4.0 +5.5 +5.3
-5.0 +6.7 +6.5

My conclusion is that in the case of my receiver the clock oscillator has a greater range of control than that expected of a 10811E.

Do other Z3801As differ? Well, here is a table prepared by Steve Smith from his Z3801A.

Efc voltage Frequency offset
0.0 +0.3 Hz
+1.0 -0.8
+2.0 -1.6
+3.0 -2.3
+4.0 -3
+5.0 -3.5
-1.0 +1.6
-2.0 +2.8
-3.0 +4.2
-4.0 +5.4
-5.0 +6.6