The receiver-reported clock rate and phase errors might have errors or misbehaviors under certain scenarios. One way to verify the reported Rate and Phase errors either in Clock mode or Position update Mode is to monitor the Receiver’s clock with 8.4 Verifying the Veracity of Reported Receiver Clock Rate and Phase Errors 149
START
LOCK ON TO SELECECTED SV SIGNAL
RECOVER SV CLOCK FROM RECEIVED DATA SV REPLICA CLOCK, NOW READY
RECOVER SV EPHEMERIS DATA TO USE TO CALCULATE SV POSITION
WAIT FOR SNAP_SHOT SIGNAL
SNAP_SHOT PULSE RECEIVED
READ SV REPLICA CLOCK INFORMATION
REORDED AT SNAP_SHOT INSTANT, SV_CLK_TIME
COMPUTE SV POSITION AT SNAP_SHOT INSTANT
& COMPUTE LOS DISTANCE FROM SV TO USER, U - SV
ESTIMATE LOS PATH DELAY FROM SV POSITION TO USER, tlos READ RECEIVER REFERENCE CLOCK TIME, REF_CLK_TIME AT SNAP_SHOT INSTANT
ESTIMATE ATMOSPHERIC DELAY, tatm
ESTIMATE GPS_TIME = SV_CLK_TIME - Terr_sv + tlos + tatm FORM DIFFERENCE BETWEEN GPS TIME AND REC REF. CLK TIME CLOCK ERROR = GPS_TIME - REF_CLK_TIME
ADVANCE OR RETARD REFERENCE CLOCK BY ERROR AMOUNT
c DKD INSTRUMENTS
Fig. 8.7 Flow chart for estimating and correcting receiver time to GPS time using one SV
150 8 GPS Time and Frequency Reception
external instrumentation. The idea is to measure receiver clock rate and phase errors with the external instruments and then compare that collected data to the data reported by the receiver at the same time instants. Such a check, properly done, can reveal many interesting phenomenon associated with GPS receiver clock performance and indeed overall receiver system performance.
The precision of the 1 PPS signal (hopefully) and reported rate and phase errors requires quality instrumentation and careful setup. Figure8.8shows our receiver model of Fig.8.1with time interval counter and frequency counter attached to the receiver clock.
The local atomic clock of Fig.8.8should be Rubidium type or better. A Cesium clock is optimum but these are very expensive. A well-behaved Rubidium will have adequate performance and is rate stable over days of operation once it is warmed up.
As shown, the 1 PPS output from the local Atomic Clock is offset from GPS time by approximately 1 ms. This offset insures that the measured phase difference between receiver 1 PPS and Local Atomic Clock 1 PPS should not go through zero and reverse sign. Some TICs can deal with this sign change while others may change scale or worse report an error during the zero crossing data points.
Assuming the Receiver is producing a corrected 1 PPS its offset to true GPS, 1 PPS is usually not of interest but rather its variance around a mean value and excursions from that value is the primary data of interest. As a note to the reader getting a 1-PPS signal in the lab that is coincident with GPS, 1 PPS is fraught with unaccounted delays, which produce phase offsets! Here are some other tips for getting the best data:
• Carefully tap off the receiver’s master clock. A buffered 50Ωoutput is prefera-ble but a scope probe is also acceptaprefera-ble.
• The counters should have a selectable input termination. If possible, use 50Ω setting and 50Ωcables. If a scope probe is used, use the high impedance setting.
Keep in mind rise times will change readings. Select thresholds carefully!
• Keep All Cables as Short as Possible
• Offset the 1 PPS from the local Atomic Clock so as to avoid a zero crossing phase error as noted above. The trigger edge polarity may be enough offset.
• Set up the TIC to read the smallest time possible. If we reversed the TIC connections shown in Fig. 8.8, the interval measured is near a second. The longer the TIC counts, the more time errors in the TIC have to accumulate.
• The 10 MHz must be connected to both instruments, especially the frequency counter. If you use the freq counter with its internal reference, you may end up measuring the rate error of the counter instead of the receiver’s clock rate error!
• Be careful splitting a single 10 MHz, 50Ωoutput from the Atomic Clock. Better to use the 10 MHz output from the instruments in daisy chain fashion.
• Make the Frequency counter the first instrument in the daisy 10 MHz chain. It is the most sensitive to issues that could occur on this reference signal.
8.4 Verifying the Veracity of Reported Receiver Clock Rate and Phase Errors 151
Fig. 8.8 Instrumentation setup to measure receiver clock rate and phase errors
152 8 GPS Time and Frequency Reception
8.5 Using a DDS Based Receiver Clock to Introduce Precise Rate and Phase Errors
Most commercial L1 receivers have limited rate control (if any) of there receiver clock. In particular the Master Clock or oscillator is often a fixed frequency TCXO or for some systems a limited analog rate adjustment is provided. A very useful method for investigating reported rate errors and possibly enhancing 1 PPS output performance is to provide replacement for the receiver’s TCXO with a clock based on a local Atomic clock. Standard signal generators do not typically have the precision and resolution needed for such a clock.
An atomic clock-based reference using a 48-bit DDS is shown in Fig.8.9and can fulfill our desired needs of precision, resolution, and accuracy. It is shown connected to our receiver of Fig.8.1providing a precise, low rate error 16.8 MHz clock signal with extremely fine rate and phase control. The DDS-based master clock shown can be setup for other output frequencies by changing the values programmed into the DDS rate word and the output bandpass filter.
A discrete8 multiplier is used to achieve the 80-MHz clock needed by the DDS. Some DDS devices have internal clock multipliers. Often these are PLL-type multipliers with an on chip VCO that is locked to the applied reference frequency. It is the author’s experience that often these types of clock multipliers are not suitable for use as DDS clock when the final goal is to drive a GPS receiver master clock.
The reason is that master clock (16.8 MHz in this example) is effectively multiplied up to the first LO by a receiver-based PLL – Synthesizer subsystem. This multipli-cation will increase the phase noise as seen at the first LO. In short, the use of internal multipliers may compromise DDS output phase noise and result in degraded first LO phase noise.
The rate resolution of the 48-Bit DDS output is given by;
DDS Rate Resolution¼80 MHz/2**48, or 2.842170943107Hz This rate resolution would correspond to rate error referenced to 16.8 MHz of;
2.842170943107Hz ¼e16.8 MHz; ore¼1.6917681014s/s The rate error resolution limit calculated above is about two orders of magnitude smaller than what can typically be expected from GPS-reported rate error precision.
If we use a rubidium oscillator for the Atomic Reference, it should be possible to set the reported rate error to near zero for many hours and observe the receiver’s performance in terms of 1 PPS phase error and reported Rate errors. In addition, a precise step of rate (or phase) can be commanded to the DDS and the reported rate inspected to verify accuracy. Large, precise rate errors can also be introduced to the receiver to verify performance and accuracy. Lastly, the phase control can be used to move the 1PPS signal with typically much finer resolution than many commer-cial receivers provide in their phase step type corrections.
Combining this system with the independent measurement system shown in Fig.8.8enables very sophisticated investigation of Receiver Clock error performance.
8.5 Using a DDS Based Receiver Clock to Introduce Precise Rate. . . 153
Fig. 8.9 DDS Signal generation of 16.8-Mhz master clock with ultra precise clock rate and phase control
154 8 GPS Time and Frequency Reception