AMSAT-NA Digital Communications with Phase 3D

Originally published in QEX, The ARRL Experimenter's Exchange, February 1995.

by Harold E. Price, NK6K

The 400 kg Phase 3D satellite will have up to 250 watts output PEP, or about 60 watts continuous. Of that 60, the digital transponder (called RUDAK-U) will have about 20 watts allocated to it. The digital RF inputs and output come and go through an 10.7 MHz IF matrix, giving us access to a variety of uplink and downlink frequencies. The actual frequency bands are in Table 1.

P3D will be flying a plethora of DSP-based modems. While the Surrey Satellite Technology UoSATs were the first to fly DSP modems on amateur spacecraft, P3D will be flying several, eight modulators and eight demodulators. The modems can appear anywhere within the digital sub bands described in Table 1.

Most of the information below comes from the RUDAK-U project manager and lead hardware designer, Lyle Johnson, WA7GXD. The design is near final, but still subject to change. The basic configuration of the digital section is shown in Figure 1. There are two CPUs, a V53 and an 80386. Each will have 16 MB of EDAC protected program and file storage space. Each processor has a connection to the payloads, via serial port, the one Mbit CAN-bus, or both. Each processor has its own set of "low" speed DSP-based modems (up to 56 kb or so). The processors share a single 256 kb modem.

The two main processors are reloadable from the ground or each other. The DSP modems and payloads are reloadable from the ground via the main processor's on-board storage. The processors/DSP modem complex has 18 programmable processors.

The 9600 baud modems are hardwired, and will be used for the initial loading and as backup to the DSP modems. They will be on fixed frequencies in the digital sub band. At least one of the frequencies will not be published, to provide a contention-free command and loading channel.

The DSP modulators and demodulators use the ADSP2171 CPU. Each modulator and demodulator chain has a separate processor, allowing the full power of the DSP chip to be used for a single half duplex link. This will allow high baud rates, up to 56 kb, or very heavily coded low baud rates. Each DSP has 10k bytes of internal, non-EDAC protected memory. The DSPs will perform an internal CRC on program memory every second and report back to the main processor, which will reboot the DSP on bad CRC or timeout.

A demodulator is fed from a high speed video ADC. Though each demodulator could use its own ADC, each is seeing the same IF and could share an ADC with one or more other demodulators. The decision, as always, is one of redundancy versus complexity and as not yet been made. Each demodulator uses its own HSP50016 Digital Down Converter (DDC) to convert the digitized 10.7 MHz IF to a frequency range more suitable for digital processing.[1] This also allows the uplink frequency to be anywhere in the passband, under software control. Likewise, the modulators use the AD7008 Direct Digital Synthesizer (DDS) to generate 10.7 MHz IF. This allows the modulators to also appear anywhere in the passband, under software control. The DDC and DDS allow the processor to specify a phase increment to an internally generated sine wave. The DDS also allows an amplitude to be specified. Since the DSP need not generate each point on the sine wave, the input and output frequencies can be much higher than the DSP chip alone could use.

The DDS and DDC can be used to allow each ADSP2171 to process more than one low bit rate signal at a time, giving us more than 16 uplinks, perhaps as many as 32 or 48 low bit rate uplinks. What could we do with such a thing? Why, that's the fun part, of course. Readers are invited to send in application ideas for all this horsepower. Before we get too carried away with thoughts of competing with Qualcomm and Orbcomm, let's review the link budget.


The above is an acronym for "There ain't no such thing as a free lunch," a phrase I was first exposed to as a child in Robert Heinlein's "The Moon is a Harsh Mistress", and which I've encountered in real life ever since. The P3D orbit (43,000 km) has a few advantages over the Low Earth Orbit (800-1300 km) spacecraft digital users are used to. The spacecraft is much further away, meaning it moves more slowly and is visible for longer periods of time. It is also larger, meaning it can generate much more power, for louder downlinks.

Here is the TANSTAAFL part.

Since P3D is much farther away than a LEO satellite like KO-23, the path loss is also much greater. The increased path loss, in fact, just about wipes out the advantage of the higher power. For example, consider the case of a typical ground user of the 9600 UoSAT spacecraft, UO-22, KO-23, and KO-25. Most of them have what is called an "Oscar 10" class station, meaning tracking antennas with about 10 dB gain, and 10 to 100 watts of uplink power. For the LEO satellites, with about two watts on the downlink, this gives plenty of link margin.

For P3D, a similar amount of downlink power results in -0.1 dB of margin. We have a rule of thumb in the amateur satellite world that says the link margin on paper needs to be about 10 dB to give adequate performance at a typical user station. This 10 dB is usually labeled "implementation loss", and in my case, equates to the inability to correctly place an N-type connector on a piece of coax. Some link margins for P3D, and the data you need to compute your own, are in Table 2.

I want to be very clear on this. Much of the PR the P3D campaign sends out, and I'm guilty as well, has been talking about 250 watt transmitters and much improved link performance for current users. The 250 watt figure and the improvements are from the point of view of current AO-10 and AO-13 voice users, not the current LEO 9600 baud data users.

Still, the goal of the RUDAK-U module is to service the current digital satellite user community. To do this, we'll need to assign a substantial amount of our downlink power budget to a single 9600 baud downlink in the 70 cm band. This leaves us with the challenge of finding interesting applications for lower power, but presumably more heavily DSP-processed modulation schemes. We'll need matching DSP modems on the ground as well. In fact, the P3D project could well exceed the current amateur capacity to generate modem software. Volunteers?

There are other factors to keep in mind. RUDAK can transmit on more than one downlink band at a time. For example, we expect the spacecraft will often be in a mode allowing both the 435 MHz and the 2400 MHz downlink to be used. While we are servicing old-style users on 435 at 9600 baud, we could be handling gateways with big antennas and higher data rates at 2400 MHz. Back in the TANSTAAFL category, while each user will have a longer access time, several hours instead of ten minutes per pass; more users can see the satellite at the same time. Will this lead to more contention, or will the long access times lead to less contention, since users aren't all trying to download in the same ten minute interval.

All in all, we believe that we can provide access to P3D from current UoSAT 9600 baud FSK class stations. We may also provide access to Microsat 1200 baud PSK users. We can also, simultaneously on the same or other bands, provide access with new modulation schemes and new protocols. There will be fun for protocol developers and well as software modem designers on this mission.


[1] The DDC has been mentioned previously in QEX, see Anderson, P. T., "A Simple SSB Receiver Using a Digital Down Converter", QEX, March 1994, for an overview.

Some photos of the RUDAK design team are available.

Original article by Harold E. Price, NK6K (, published in QEX, The ARRL Experimenter's Exchange, February 1995. Hypertext conversion by and feedback to KB5MU.