Digital OSCAR Communications

By: John A. Magliacane, KD2BD


This paper was originally published in the January/February 1997 issue of Satellite Times magazine.


The Federal Communications Commission allow Radio Amateurs the privilege of using a myriad of digital communication modes in various forms on the ham bands. Tuning across amateur frequencies, one might hear communications utilizing continuous wave (CW) telegraphy, radioteletype (RTTY), AMTOR, PACTOR, ASCII, and packet radio, just to name a few. These digital signals are modulated onto RF carriers using such methods as on/off keying (OOK), frequency shift keying (FSK), and phase shift keying (PSK).

In the world of amateur satellites, early OSCARs used CW and on/off keying to transmit telemetry information to groundstations through low-powered beacon transmitters they carried into earth orbit. As technology advanced and home computers became popular, the digital communication modes, and modulation methods that supported them, changed. Soon, CW shared telemetry downlink time with RTTY, ASCII, and packet radio. Specialized satellites were later designed that contained only digital communication transponders. These satellites are setting an important trend in amateur satellites.

Digital communications have a significant and growing impact in the telecommunications industry, and the non-commercial world of amateur radio is no different. In fact, amateur radio provides opportunities for research and experimentation in the digital communications field not possible or practical in the commercial world. It is, therefore, important to have a basic understanding of digital communications and the OSCAR satellites that support them if the Amateur Radio Service is to continue to grow with advances in technology.


Radio Basics 101

Basically, radio transmissions are accomplished by modulating a radio frequency (RF) carrier with intelligence in a circuit known as a modulator. The modulated RF carrier is then amplified, and radiated into space through an antenna system. The intelligence the RF signal carries may contain analog information, such as in the case of voice or video, or digital information, such as in the case of morse telegraphy, packet radio, or any other form of binary data sent in a serial fashion.

There are three aspects of an RF carrier that may be modulated in order to carry intelligence. Either the amplitude of the RF carrier is modulated, or its frequency is modulated, or its instantaneous phase is modulated. In some cases, such as commercial broadcasts of color television, AM stereo, or even modern telephone modems, a combination of phase, frequency, and amplitude modulation is used to convey several discrete channels of intelligence on a single carrier. However, this combination of modulation methods is rarely seen in present day amateur radio communications.


Morse: The Essential Language

The first form of amplitude modulation used by amateur radio operators consisted of on/off keying of an RF carrier by morse telegraphy. Commonly referred to as CW, morse telegraphy is still in wide use today for both terrestrial and satellite communications. CW was the first mode of communications adopted by OSCAR satellites for the transmission of telemetry information through beacon transmitters carried on amateur spacecraft, mainly because of its simplicity and communications effectiveness. It is still widely used today.

Morse telegraphy is typically copied by ear. Satellite telemetry conveyed in CW is interpreted through the use of look up tables and calibration equations. One of the advantages of morse telegraphy is that decoding is done by ear, and requires no data demodulator or specialized electronics for reception.

During the mid 1970s, OSCAR satellites adopted more advanced techniques for making telemetry transmissions. OSCAR 7, launched on November 15, 1974, was the first amateur satellite to use radioteletype (RTTY) for the transmission of spacecraft telemetry. Frequency shift keying (FSK) modulation was used for the transmission of RTTY signals. Radioteletype provided the advantage of transmitting information at a higher rate than was possible using CW at speeds that could be comfortably copied by ear. The speed at which data can be transferred between transmitter and receiver is an important consideration for low-earth orbiting satellites with limited access times. The disadvantage of RTTY is that it required expensive or bulky terminal equipment for reception, and was later shown to be relatively inefficient in terms of data rate per unit of power, another important consideration when dealing with low-powered OSCAR satellites with limited supplies of power. For this reason, RTTY is rarely used today by OSCAR spacecraft.

During the 1980s, the FCC approved the use of the American Standard for Information Interchange (ASCII) for communications on amateur frequencies. This started a trend toward using more advanced data encoding techniques directly compatible with home computers that were growing popular at the time. This compatibility allowed PCs to be used to interpret, store, and directly process data received from space. Like RTTY, ASCII required a data demodulator to be used at groundstations to convert the signals received from the satellite by the groundstation receiver to voltage levels compatible with the computer performing the data processing functions. Like RTTY, frequency shift keying was the modulation most often used for making ASCII transmissions.

UoSAT-OSCAR-9, launched on October 6, 1981, was the first amateur satellite to adopt advanced forms of digital transmissions such as ASCII for transmitting spacecraft telemetry on its beacon transmitters. UoSAT-OSCAR-9, and later its younger sister, UoSAT-OSCAR-11, assigned audio tones of 1200 Hz and 2400 Hz to represent the digital ones and zeros (marks and spaces) that comprised the ASCII serial data stream used to transmit spacecraft telemetry information to groundstations. This form of modulation is known as audio frequency shift keying, or AFSK. The early UoSAT satellites used AFSK audio tones to modulate a narrowband FM beacon transmitter, making groundstation reception as easy as turning on a 2-meter FM rig. The problem, of course, was demodulating the mark and space tones in ways that would be least affected by noise, and converting those audio tones to the appropriate logic levels suitable for interpretation by a home computer.

Also during the early 80's, AFSK-FM was adopted by early packet radio pioneers who used FM voice transceivers and surplus Bell 202 telephone modems that were widely available at the time. This combination provided a quick, easy, and low-cost method of transmitting digital information using a commonly available voice-grade FM voice transmitters without the need for making any circuit modifications to the communications equipment.

There was, however, a price to pay for this simplicity. AFSK-FM is essentially two FM signals in one, and as such, is terribly inefficient in terms of bandwidth and communication effectiveness. AFSK-FM requires a very strong signal for reliable communications, much higher than if other modulation methods are used. Steve Goode, K9NG, studied this situation with respect to 1200 baud packet radio communications and concluded that a signal level that produced at least 25 dB of receiver quieting was required for reliable packet radio communications using AFSK-FM. This strong signal level is not easy to realize from OSCAR satellites operating many miles from earth with only several hundred milliwatts of transmitter power. Clearly, a better solution was needed if digital communications were to be effective via OSCAR satellites.


Coherent Communications and BPSK

For many years, a small group of amateur radio operators and low frequency radio enthusiasts (LOWFERS) have been experimenting with coherent continuous wave (CCW) communications. CCW permits the detection of radio signals at levels too weak to be copied by ear, and allows transmitter power levels to be reduced dramatically as a result. CCW achieves its stellar weak signal performance by being sensitive to not only the frequency and amplitude of the desired radio signal, but also its phase. Coherent CW achieves high signal sensitivity while remaining relatively immune to interfering signals and noise since these undesired signals have random frequency and phase characteristics compared to the desired signal.

Coherent CW is great for morse telegraphy communications, but it can be shown that if the phase of the RF carrier were modulated instead of its amplitude, an additional 6 dB in signal-to-noise ratio could be realized over CCW's on/off keying technique without any increase in transmitter output power. The 6 dB advantage comes when the phase modulation is antipodal -- that is, it shifts by exactly 180 degree intervals. Phase shift keying comes in many different varieties. Antipodal PSK is commonly referred to as binary phase shift keying (BPSK).

BPSK is a very robust form of digital modulation and has been used to good advantage in copying extremely weak signals from several interplanetary deep space probes. BPSK was adopted by the AMSAT-OSCAR-10 and AMSAT-OSCAR-13 amateur satellites for telemetry transmissions, and later by several other OSCARs carrying digital transponders.

The problem with coherent communications is that proper signal demodulation requires a reference carrier with which to compare the received signal in a synchronous phase detector. A phase detector functions by comparing two signals and producing an output voltage based on how well those two signals match in terms of frequency and phase. If the frequency and phase of both signals match, a voltage of a certain polarity is produced, while if both signals are completely out of phase, an output voltage of opposite polarity is produced. Finally, if both signals are 90 or 270 degrees out of phase, the phase detector produces no output voltage.

Although BPSK offers a tremendous advantage in terms of signal-to-noise ratio over FSK and on/off keying, its reception can be a bit tricky. If viewed in the frequency domain, a BPSK signal may be thought of as being a double sideband (DSB) suppressed carrier AM signal. This is not too much different from single sideband suppressed (SSB) carrier AM that is popular on the HF ham bands for voice communications. Proper demodulation requires re-insertion of a locally generated carrier having identical frequency and phase characteristics as the carrier suppressed in the BPSK transmitter. This is similar to the technique used to receive single sideband voice transmissions, except that in the case of SSB, the phase of the locally generated carrier (beat frequency oscillator, or BFO) is unimportant since only one sideband is transmitted. Compounding the problem of requiring an accurate frequency and phase reference for properly demodulating BPSK signals is Doppler shift that is inherent in all amateur satellite communications. Doppler shift causes the radio frequency of the signal received from the satellite to constantly drift lower in frequency as the slant range to the satellite changes during a pass, making it even more difficult to perform proper phase detection.

Since the BPSK carrier is suppressed by the BPSK transmitter, the key to demodulating BPSK emissions from satellites lies in the regeneration of the needed carrier reference from the upper and lower sideband components of the BPSK signal. Two popular methods exist for doing this. One method uses a circuit known as a Costas Loop, while the second method uses a Squaring Loop. In theory, the Costas Loop offers slightly better performance at the expense of added circuit complexity. The simpler Squaring Loop, however, can be made to provide performance comparable to that of the Costas Loop with additional filtering.

Other issues come into play when comparing the sensitivity of a BPSK receiver. Radio signals received via satellite often fade in amplitude as the slant range to the satellite changes, and signal polarization changes occur during a pass. In order to accommodate a wide range of signal levels, some BPSK demodulator designs incorporate amplitude limiters. Limiters are often used in FM receivers, and are responsible for a characteristic in FM communications known as the "capture effect". If more than one signal is received at the same time on the same frequency, the stronger signal will "capture" the limiter and be the only signal detected by the receiver. If the stronger signal happens to be noise, the desired signal will be completely lost. Designing a BPSK demodulator with non-limiting automatic gain control (AGC) has been shown to have a significant weak signal sensitivity advantage over designs incorporating hard limiting. Furthermore, post detection filtering that closely matches the bandwidth of the transmitted signal to allows the detection of BPSK signals even if they are received at a level below the ambient noise level.

BPSK signals are typically copied with receivers designed for reception of single sideband (SSB) voice transmissions. SSB receivers are essentially frequency converters that linearly convert radio frequency signals to the audio spectrum where they can be copied by ear. Virtually all groundstations that access the 1200 baud Pacsat satellites use SSB receivers and BPSK demodulators for downlink reception. Uplink transmissions, however, use FSK and a data encoding technique named after its birthplace of Manchester, England.


Manchester Encoding

Pacsat satellites operating at a data rate of 1200 bits per second (BPS) respond to FSK uplink signals that have been Manchester encoded. Manchester encoding modulates the transmit data with a carrier equal in frequency to the bit rate (1200 Hz). The resulting product is actually BPSK with a suppressed carrier frequency of 1200 Hz, although it is not necessarily handled as such. Manchester code is also known as "split phase".

In terms of Pacsat communications, signals from a standard packet radio terminal node controller (TNC) are combined to produce Manchester code in a phase modulator, the output of which is low-pass filtered and used to frequency modulate an FM voice transmitter, producing FSK. Manchester encoding eliminates the DC component of a digital data stream, allowing it to pass without distortion through the speech circuits of an FM voice transmitter. A direct connection to the FM varactor diode is not needed for 1200 bit per second Pacsat communications. Even phase modulated voice transmitters can be used to generate Pacsat uplinks. The lack of a DC component also makes a Manchester receiver immune to small changes in carrier frequency such as those caused by Doppler shift, and allows for easy bit clock extraction by the receiver.


Pacsat Receivers

The Pacsat receivers carried on-board the AMSAT-OSCAR-16, WEBERSAT-OSCAR-18, and LUSAT-OSCAR-19 satellites were designed around a Motorola MC3362 dual conversion single chip FM receiver, with one complete receiver chip employed for each of the four uplink channels on each satellite. A single GaAsFET-based frequency downconverter with an intermediate frequency (I.F.) in the 40 to 50 MHz range is shared by all Pacsat uplink receivers.

The first I.F. in each FM receiver chip operates at a center frequency of 10.7 MHz, and is preceded by a filter with a very sharp skirt. The frequency response of the filter rolls off by approximately 20 dB at 20 kHz, and achieves a total of 70 dB of adjacent channel rejection. Impedance matching techniques used on the input and output of each I.F. filter minimizes passband ripple to approximately one decibel.

Rather than use a final I.F. frequency of 455 kHz as is typical in applications using the Motorola MC3362, Pacsat designers instead used a final I.F. frequency of 1.8 MHz in an effort to realize a highly linear FM discriminator characteristic across the entire 20 kHz bandwidth of the uplink channel. While Pacsat receivers employ no automatic frequency control to compensate for Doppler shift, groundstations employing approximately 3 kHz of peak frequency deviation need not be concerned with Doppler shift of their uplink signals.


High Speed Data Communications

High speed data communications by satellite is currently no different than terrestrial high speed data communications, except for the fact that satellite communications is full duplex, while terrestrial communications, unless crossband, typically are not. The same hardware is used in either case, and data is transferred at 9600 bits per second (also referred to as "9k6") using frequency shift keying (FSK). Because high speed serial data occupies a wider frequency bandwidth than audio frequencies typically used for voice communications, standard FM voice communications equipment can be used for high speed data communications provided that direct connections to the receiver's discriminator and transmitter's varactor diode are made. These connections are needed to ensure a response characteristic that is flat in terms of frequency and phase across the entire bandwidth of the receiver. Such characteristics are a necessity for high quality data communications.

It is interesting to point out that 9600 bit per second FSK communications require no more RF bandwidth than that required by 1200 bit per second AFSK-FM. In fact, 9600 bps data communications requires slightly less. However, in order to realize this bandwidth, the transmitted data must be first processed through a low-pass filter having a low group delay, and sharp cutoff prior to modulation. In addition, high-speed data is often "scrambled" to randomize the data pattern and provide for rapid bit clock recovery in the receiver and minimize the DC component of the signal, making automatic frequency control (AFC) both practical and effective.


Pseudo Random Scrambling

As stated earlier, 1200 bps FSK data is Manchester encoded to eliminate the DC component of transmitted data. In doing so, however, it effectively doubles the bandwidth of the transmitted signal. This is not a major concern in terms of 1200 bit per second data communications, but carries a negative impact in terms of 9600 bps communications. In an effort to conserve bandwidth, a different approach is taken to reduce the DC component of high speed binary data. Instead of Manchester encoding, high speed is randomized or scrambled prior to transmission, and unscrambled during reception. Scrambling significantly reduces the chances of transmitting long strings of zeros and ones, which by virtue of their constant and unchanging polarity, carry a strong DC component. It is the DC component of a digital signal that makes receiver clock extraction slow and unreliable, data carrier detection difficult, and receiver automatic frequency control (AFC) nearly impossible.

The Federal Communications Commission allows several forms of data scrambling to facilitate high speed data and spread spectrum communications on amateur frequencies. Note, however, that scrambling is not used to obscure the meaning of the transmitted signal since the methods used to scramble data on amateur frequencies are well published and FCC approved for use by amateurs.

The method most often used to scramble high speed data normally employs a cascaded string of shift registers. 17 such devices are typically used. Taps made to the output of the fifth and 17th shift register are combined with the 9600 bit per second data in a pair of exclusive OR (XOR) gates. During transmit, the shift registers are driven by a bit clock derived from the sending TNC. On receive, the process is reversed, and the bit clock is derived from the clock extraction circuits of the FSK data demodulator. It is interesting to note that if scrambled data transmissions are monitored on a narrowband FM voice receiver, the pseudo random nature of the data's content makes the transmissions sound very much like the white noise of an unsquelched FM receiver.

Data scrambling is not a panacea by any means. While scrambling techniques reduce the DC component of serial data transmissions, they do not eliminate it completely. As a result, a frequency response down to 5 Hz, or preferably DC, is required by the modem and the radio equipment with which it is used. There is also some concern as to whether scrambling reduces the bit error rate of a communications channel since an incorrectly received bit fed into a string of shift registers will corrupt bits received at a later time. If a corrupted bit is received at the end of one packet, it can destroy the next packet received, even if the second packet is received without error. Nevertheless, the data scrambling and unscrambling technique described here is highly effective and is used by most 9600 bit per second terrestrial and all satellite data communication links.


Conclusion

The purpose of this discussion was to show that there is a method to the madness of amateur data communications, and that data transmission protocols and modulation methods are not subjects that are shrouded in secrecy. In some areas, modulation methods are selected for simplicity. In other areas, they are chosen because of their performance superiority over other methods.

The entire subject of digital communications and the OSCAR satellites is far too detailed to effectively cover in this column. However, those interested in a more in-depth understanding of this subject matter are encouraged to consider reading "Packet: Speed, More Speed and Applications", published by the ARRL. This book not only covers the subject of data communications in depth, but also provides block diagrams and schematics of projects that can further the understanding of the concepts used to convey digital information on amateur frequencies.

See you on the birds!


Frequency shift keying uses discrete carrier frequencies to represent each binary logic level. This example represents coherent FSK. Note that both mark and space tones are harmonically related. UoSAT-OSCAR-9 and -11 used AFSK of this type.



Binary phase shift keying uses reversals in carrier phase to represent different binary logic levels. An accurate phase reference is required for proper demodulation.



Manchester encoding is similar to BPSK in that changing logic levels are represented by reversals in carrier phase. In this example, a falling edge within a bit interval represents a logic "1", while a rising edge within a bit interval represents a logic "0".



Circuit representative of those used to scramble or randomize 9600 bps FSK data transmissions. A complementary circuit is used at the receiver to restore the data stream to its original format.

John A. Magliacane, KD2BD