In this section the overnight stability tests of the 8 GHz band two-way timing link measurements of May 26-27, 1999 are discussed. The test was run over a 16 hour period with an orbit prediction file that would produce a residual ``Doppler'' frequency of about -1.5 Hz. The exact times when this test was performed are 21h15m16s May 26, 1999 until 13h15m18s May 27, 1999. The measured residual phases, residual ``Doppler'' frequency and the RMS noise level determined using a linear fit to the phases are shown in Figures 17 and 18.
The first thing to note is that the VIRK locked to the uplink signal almost instantaneously once it was present. This was determined from the Russian analog read-back system which had an indicator for the phase loop lock in the VIRK 7/8 GHz transponder. However, the VIRK did not produce any downlink signal which could be used for valid two-way timing measurements for another 257 seconds, i.e. the phases on the downlink were not stable enough for the earth station's Costas loop to lock to the carrier. This results in no valid two-way timing phases being measured and thus no two-way timing information. It was generally observed for all radiative tests that the VIRK would take four or five minutes for the earth station`s Costas loop to lock to the downlink signal. So the first four or five minutes of any established link with the VIRK would not produce any two-way timing measurements even though the VIRK locks to the uplink signal almost instantaneously. Although the source of this delay was never identified it is similar to the amount of time that it takes the traveling wave tube amplifier in the VIRK to ``warm up'' ([3]).
In Figure 17 it can be seen that there are several small glitches in the residual phase measurements. These glitches are of unknown origin and are jumps in phase of order 0.1 cycles - 11.8 picoseconds in the residual timing measurements. Tests of a similar length of time were performed using the NRAO satellite simulator in which no glitches were seen. This suggests that the source of the glitches is within the VIRK modules or the Russian test set. The measured phase residuals were run through the earth station program doppler. This program creates a ``Doppler'' file. During normal on-orbit operations, such a file would be sent to the JPL NAV team for orbit determination; here we merely use the program to perform certain data quality checks. 57212 of 57602 seconds of valid two-way timing phase residuals were measured that were acceptable for producing doppler data. None of the glitches described above resulted in a new clock setting event being needed. All of the invalid data were at the beginning of the tests while the VIRK was not producing a valid downlink signal.
The VIRK produced residual timing data that were similar to but generaly
noisier than what is observed
for the Japanese HALCA satellite (see Figures 16 and 18). We can compute the
expected coherence 
of radio astronomical observations from the RMS timing noise
(in seconds)
in the timing residuals. This is calculated using the relationship
| Parameter | Measured Value |
| VIRK power output - 8 GHz | 400 milliwatts |
| VIRK power output - 15 GHz | 11 Watts |
| VIRK minimum power level for lock |  dBm
= 8.9 x 10-14 Watts |
| VIRK lock capture range | 7215.2670 to 7215.2320 MHz |
| VIRK lock hold range | 7215.26724 to 7215.232240 MHz |
| VIRK Bit Error Rate | > 10-5 |
| Average VIRK delay before producing valid downlink phases | 4-5 minutes |
| Average noise in phase residuals | 0.02 cycles |
| Coherence at 330 MHz | ~1 |
| Coherence at 1.6 Ghz | ~1 |
| Coherence at 5 Ghz | 0.99993 |
| Coherence at 22 Ghz | 0.99865 |