By Jim White (WD0E), Bruce Rahn (WB9ANQ), Paul Williamson (KB5MU)

Beginning with the turnover of AO-16 to the command team in early 1991, we have been tending to its care and feeding and occasionally that of the other microsats. Among other tasks this includes gathering and analyzing telemetry for the purpose of watching over the health of the spacecraft bus. During the past several months we have greatly increased our understanding of the spacecraft systems. However there were a few things that we were never able to understand to our satisfaction. The crash of the file server software on 27 July provided the opportunity to review spacecraft performance and systems and answer a few of those bothersome questions. It turned out that we learned much more than anticipated. This paper discuses the analysis methods used, the results of that analysis and the expected effects on microsat users.

The four microsats (AO-16, DO-17, WO-18, LO-19) were launched in January 1990. Operational control of LO-19 transferred to the Argentine group who sponsored and funded it a few months later. Control of WO-18 was assumed by the Center for Aerospace Technology at Weber State University at about the same time.

Control of AO-16 was gradually assumed by a command team led by Bruce Rahn, WB9ANQ. The first year was a period of intense software development and a parallel effort to collect and analyze large amounts of telemetry that led to a fairly through characterization of the spacecraft systems. During that time Bruce, Jim White, WD0E, Courtney Duncan, N5BF, and (after Courtney's departure from the team) Paul Williamson, KB5MU, learned a great deal about the operational characteristics of AO-16. However at the time the file server software became operational some questions remained. Since the bulk of the downlink channel time was being used by file exchanges it was difficult to collect telemetry at a fast enough rate to allow detailed analysis. While some work continued, such as characterization of seasonal temperatures and power generation cycles, other efforts were put on hold. By July of 1992 several months had gone by with little activity of this nature.

When the file server software crashed (probably because of a double bit error in the EDAC memory) it had been operating continuously for 316 days without a problem. At that point the server software developers were a few days away from making the latest version available for AO-16. The new release contains several enhancements including the directory broadcast feature which will greatly increase throughput. The team decided we would not load the old software only to reload the newer version a few days later. Rather, we would wait for the new code and take the opportunity to run higher speed telemetry and possibly even Whole Orbit Data (WOD) for a few days. Hopefully this would allow us to answer several questions. The puzzle list looked something like this:

- Some of the data from some telemetry channels had never been completely understood. The results did not match expectations. These were not particularly important channels (or so we thought) so we had not been too concerned about the funny results. These were temperature channels and generally indicated the design targets had been met even though some patterns in the data were puzzling. But Bruce recently had begun to suspect two or more of these channels were reversed or otherwise incorrectly labeled.

- We had received reports that the AO-16 PSK transmitter was hard to copy. Bob Diersing, N5AHD, had measured the bit error rate (BER) from AO-16 several months before and found it quite acceptable, so these reports were puzzling. However, some of us had experienced the problem ourselves so it appeared as though this would be worth looking into.

- We wanted to know how much the space environment had reduced the output of the solar arrays. Whole orbit data is necessary to make this determination.

- Reports of phase noise or jitter on various microsat transmitters (especially from James Miller) caused us some concern. Preliminary observations had indicated that there was indeed more than we would have liked.

- We were looking for ways to tune up the power control software in the spacecraft to improve the downlink signal margins, which is an continuos effort.

We felt all or most of these questions could be answered with a few days of high speed telemetry from the spacecraft and a few experiments. Of course it took longer than expected, but we found out more than we anticipated. During this period of engineering tests we concentrated on AO-16, but did some work with WO-18 and LO-19 as well, in cooperation with their respective command groups.

It might be worth reviewing why answering these questions was important.

- It's always desirable to do a detailed validation that spacecraft systems are working correctly from time to time so you can watch for long term trends that might not show up in routine coarser observations. Catching these can sometimes allow corrective action to be taken before a problem becomes serious. Validations are valuable to spacecraft designers and builders because they show what worked well and what can be improved in future satellites. With at least two more spacecraft of the microsat design in the works results can be applied during construction and hopefully contribute to an improved spacecraft.

- It was necessary to determining why the AO-16 PSK transmitter was difficult to copy before corrective action or adjustment could be considered. The ideal result would be a better signal and more usable satellite. This is especially important for ground stations using omni-directional antennas, for whom any degradation in signal could mean the difference between acceptable and unusable down-link signals.

- Clearing up the confusion over the temperature telemetry channels was not considered to be particularly important since it was clear that the temperatures were well within design specifications and were not a threat to the batteries or other spacecraft systems. We did not realize how important this issue would become.

- Lastly, and certainly of lesser importance to the amateur satellite user, the command team just can't stand it when we don't understand something.

Bruce and Jim collected and analyzed telemetry from their own locations and from several other AMSAT members and microsat users who were kind enough to respond to their requests for data from various other locations. Internet electronic mail and telephone discussion and consultation took place between the command team members and with Dick Jannson, WD4FAB, Jan King, W3GEY, and Harold Price, NK6K. In about three weeks we gathered and dissected over 1 megabyte of telemetry, exchanged several hundred thousand bytes of email and ran up our phone bills considerably with brain storming sessions well into the night. The primary method of telemetry analysis was to gather the data, often while commanding changes in various spacecraft parameters, then drop the calculated engineering values into spreadsheets and graph or otherwise manipulate the results. This activity peeked with several hours of observations and analysis on August 8.

On that day Jim took a car load of equipment to Jan King's OSC lab in Boulder. A few hours of preparation went into setting up lab quality equipment to observe in real time and make photographs or plotter charts of spectrum displays, phase jitter, and eye patterns of the microsat transmitter signals. As passes took place for the remainder of the day conference calls were established with other team members. Often one member would command changes to transmitter power or other parameters and observe and report on telemetry while another would transmit special test files and Jan, Jim, and Greg Hein, WT0M, would observe and record the results.

By special arrangement with the LO-19 command team that satellite was transmitting a test file to enable spectrum analyzer measurements. A great deal was accomplished during a few short passes thanks to the close cooperation and flexibility of the entire team. It is worth mentioning that no prior arrangements had been made with Weber State University, but Bob Argyle noted what we were doing on the other microsats and did exactly what was needed to WO-18 while we were observing it. Further coordination was then accomplished on the phone. Special thanks to Bob for being so observant and responding to the situation.

As of this writing (mid August) the following issues have been resolved by the work that led up to the August 8 effort and the analysis which has followed. More work remains and further results will be reported in the AMSAT-NA Journal.

The AO-16 PSK transmitter carrier suppression has deteriorated to the point where it is essentially 0 when cold, and is only about 5 to 6 dB when warm. Figures 1, 2 and 3 are spectrum analyzer displays of the AO-16 PSK transmitter before launch, on August 8 when cold, and the same day when warm respectively. Carrier suppression can be determined by measuring the vertical distance between top of the carrier signal and the modulation peeks in the plots. Please note in these and the figures which follow that the horizontal and vertical scales of the plots are not all the same. The TAPR PSK modems in use at OSC and at WB9ANQ were able to copy data reliably at power levels of 1W and 1.3W when suppression was measured at 6.7 dB and were able to copy only occasional packets at a power level of 3.2W when suppression was measured at 5.3 dB. It appears that when carrier suppression is 5 dB most current PSK modems are unable to properly recover the data. This carrier suppression problem probably results from transmitter component values drifting slightly with time. The Q of these stages may be high enough that they are overly sensitive to small component value changes that have occurred in orbit. It is not likely this will get better for the life of the mission. One option for the continuation of the mission was to operate the PSK transmitter at no more than about 1 Watt and dump the rest of the power into the batteries. This would lead to over charging, which was a concern. While this option might result in a usable signal for a while, there was the distinct possibility that the carrier suppression problem would get worse because of further component drift, rendering the signal useless at any power level. The other option was to switch to the raised cosine (RC) transmitter. There were several mitigating factors considered regarding the RC option.

The power amplifier transistor (HPA) in the RC transmitter was replaced at the last minute in the lab prior to shipment to Kourou because the original was damaged. The new one was the most efficient of the entire set of devices purchased for the project, but was inadvertently tuned into a bad load during repair. Additionally, there was no time to re-calibrate the RC transmitter power output telemetry channel. So we have a very efficient device less than optimally tuned and an uncalibrated power output telemetry channel. However it was Jan's judgment that there was little risk in operating this transmitter full time and it would seem we can live with the telemetry calibration question. The team concluded that the PSK transmitter should be removed from service and the RC transmitter used for routine operations.

The AO-16 RC transmitter carrier suppression was measured at between 18 and 20 dB and was essentially insensitive to power level, which is excellent. Figure 4 is the RC transmitter spectrum prior to launch and figure 5 is the same display captured during these tests. Additionally the phase jitter or phase noise was examined. The jitter (or deviation from exactly 0 and 180 degrees of phase shift) was approximately 20 degrees and about 5 degrees asymmetrical. It will probably be worth attempting to adjust the modulator in order to reduce the jitter and/or make the shift more symmetrical. However even in its current state this transmitter should perform quite well. (The best microsat transmitter observed had about 15 degrees of jitter and was symmetrical.) The advantages of RC over PSK have been discussed in pervious papers and articles. However we should also note that we confirmed Jim's hypothesis of a few months ago that the second sideband product of the RC transmitters is about as strong as the same sideband on a straight PSK transmitter - even though remaining sideband energy is very low. With very large link margins this may make it easier to achieve false lock when receiving an RC transmitter.

- The WO-18 straight PSK transmitter was known to be bad from tests conducted immediately after launch. During the August 8 tests the phase jitter was in excess of 45 degrees at 1.4W output power. Figure 7 is an oscilloscope display showing the phase deviation from 0 and 180. A perfect PSK transmitter would create a display with tight circle traces at the top and bottom of a vertical line. The angle from the center line to the edge of those circles is the amount, in degrees, the phase shift varies from 0 and 180 degrees. In figure 7 it can be seen that this transmitter is producing so much phase noise that it would be very difficult for any demodulator to recover the data. We did not look at the WO-18 PSK transmitter spectrum because of time constraints. The jitter observations were useful as comparisons to properly operating transmitters.

- The WO-18 RC transmitter has a carrier suppression of about 7.5 dB at power level of 1.4W (figure 8), and the phase jitter is about 15 degrees and quite symmetrical. Figure 6 is a spectrum display of this transmitter prior to launch. The phase jitter is equal to the best microsat transmitter at this time. There is probably be little to be gained from adjusting the modulator except for observing the effect of those adjustments on a properly operating transmitter.

- Figure 9 shows the LO-19 RC transmitter carrier suppression prior to launch. During the tests we measured it's carrier suppression at 12dB as can be seen in figure 10.

- No observations were made of the LO-19 PSK transmitter, but a carrier suppression observation may be made at some future date.

The following conclusions were reached after a couple of hours of examination of WOD, single pass telemetry, Jim and Bruce's earlier test results, and records made during design and construction.

- Conflicting notes exist in the available documentation as to which of the thermistors is connected to each AART (Asynchronous Addressable Receiver Transmitter) multiplexer input. Only sketchy information is available that directly relates AART addresses to thermistor locations. However in every case CH 36 is related to the +Y (originally -Y in some data) thermistor. After examination of telemetry data we concluded channels 34 through 37 were actually connected as shown in the chart below. It can be clearly seen in figures 12 and 13 that channel 36 is actually the +Y panel temperature. This cleared up a long term question about the behavior of that channel. Now that we understood this channel to be the temperature of a side panel the reasons for the 10 to 15 degree swings during the illuminated portion of the orbit were clear. The side panels warm and cool as the sun angle on them changes. CH 35 has been used to obtain the +Y panel temperature for the array-side power management software. We have been using a channel (CH 35) that at times is up to 15 degrees different from the actual side panel temperature.

- It is likely that these channel assignments are the same in all four spacecraft because the thermistors were physically connected to the same points on the AART boards in each. However this could not be verified from the available data. Further testing is planned to substantiate this.

- We determined CH 14 "Rx Temp" and CH 15 "+X (RX) Temp" channels are probably reversed (figure 14).

- The makeup of the panel sandwich for the side and top surfaces was verified. The team has ask Dick Jannson to check his thermal conductivity study of these panels so we may be able to more closely adjust the +Y and +Z temperature measurements and calculations.

- Some quick calculations indicate for every 5 degrees of improved accuracy in measured panel temperature we will gain .15W at the output of the Battery Charge Regulator (BCR). That is .45W for the 15 degree improved accuracy that will occur during portions of each orbit. An additional 1 Watt may be produced for short periods once all software tuning is completed. It remains to be seen how much this will impact power generation over an entire orbit. However, even a small improvement will have a positive impact in the microsat power system because it has an overall orbit average power production of only 6 Watts.

Several improvements to microsat operations and future microsat designs will result from these tests and investigations. We have already switched to the RC transmitter on AO-16 which will provide a downlink signal that most stations will find easier to reliably decode. Expected or proper operation of several other microsat transmitters has been verified. Power control programming will be modified to correctly sample and calculate the temperature of the panels providing the majority of current to the battery charge regulator. This will allow the transmitter to be run at higher power which will improve downlink signal strength. Additional operating time for the AO-16 and DO-17 S-Band transmitters may also be possible. A great deal of information about the quality of the system design(s) was gathered and documented for future satellite builders.

As of this writing several tests and measurements remain to be done. Bob Diersing is gathering data in order to determine downlink bit error rates while the RC transmitter is in use. We have not yet loaded whole orbit data gathering software so we can measure panel power output deterioration (however glimpses of portions of orbits confirm our expectation that little reduction has taken place since about the end of 1990). We would like to look at the spectrum of the LO-19 PSK transmitter and both of the transmitters aboard DO-17.

Thanks those who assisted during these tests, and especially to the microsat user community for their patience while AO-16 has been out of service. With luck the changes noted in this paper will be implemented shortly and the AO-16 file serve back on line with improved signal levels and quality. Improvements will be implemented in the other microsats as operations allow.