Packet Radio communication on 9600Bd and higher is becoming increasingly popular. However, the technology involved - while "simpler" than the widespread 1200Bd AFSK - is quite a bit more complex in its implementation.
Apart from experimental setups, there is currently only one 9600Bd technology that has become a de-facto standard for amateurs worldwide. It was first introduced by James Miller, G3RUH, and is popularly known as "G3RUH-FSK". Over the years, a number of G3RUH-compatible modem designs have become available.
Why can't we simply use AFSK for 9600Bd? Remember your amateur radio class if you still can: you would have learned that the bandwidth of an FM or AM signal depends - perhaps among other things - on the frequency of the modulating audio. If you do this for 1200Bd with an auxiliary carrier (the tones that 1200Bd modems produce), your total signal will fit nicely within the 15-or-so kHz bandwidth that filters an amateur receiver will pass.
If you did this for 9600Bd with an auxiliary carrier, you'd have a much wider signal. Apart from raising the ire of your fellow amateurs on adjacent channels, you are likely to get garbage on receive: your receiver filters will remove essential parts of your signal (the high frequency components) and screw up the phases all over the place.
Worse still, most FM rigs have pre-emphasis on TX and de-emphasis on RX, which does wonders to the nice digital signal you put into the microphone input. Not ideal.
G3RUH decided leaving the auxiliary carrier aside. Since that on its own still won't work, he further decided to do away with the pre-emphasis and de-emphasis filters in the radios. He also wanted to make the best of the available bandwidth, i.e., transmit (roughly) about the same power in every Hertz of his transmit channel. For this purpose he wanted to have a roughly flat audio spectrum going into the transmitter.
Unfortunately, most TTL or CMOS serial data just doesn't correspond to a flat audio spectrum. A flat audio spectrum sounds like open squelch noise. The solution was to scramble the data stream from the computer.
The scrambling is done by a shift register and a few XOR gates - the process is reversible at the receiver. So the data - while "scrambled" - is still readable without anyone having to know a special "code".
If you look at a TTL signal bit stream - even a scrambled bit stream - on an oscilloscope, you will still see "edges". It's this "edginess" that - much like the morse key clicks that you were warned about when you became a ham - causes out-of-band interference and therefore has to be eliminated.
G3RUH accomplished this by means of a digital filter - essentially an EPROM that takes the current bit plus the last few bits and uses them as an address. The EPROM content at that address is then used as the input to a digital-to-analog converter which produces the actual transmitter audio. Its output is sent through an analog low-pass filter that removes the DAC "step" edges and from there straight into the modulator of the transmitter.
The transmitter in question must be FM (PM won't do) and ought to be crystal controlled as most synthesized rigs (even some rigs that pretend to be "9600Bd-ready") don't perform very well.
Why? Consider that our audio signal is "flat", i.e., uses all audio frequencies between DC and 10kHz or so to about the same extent. Now think about your synthesizer PLL. If you want to modulate it, you have two options:
If you modulate the VCO, the loop will regard low frequencies as frequency drifts and will try to compensate for them. That's OK for voice but often fatal for 9600Bd. If you modulate the reference oscillator, the loop inertia will absorb the higher audio frequencies. You could modulate both, but that's not done in radios built for voice. In any case, your TX/RX switching times are likely to be astronomical with a synthesized radio.
So crystal-controlled radios such as the TEKK transceivers available through BayCom are often the best choice.
The receive side of G3RUH modems is somewhat simpler, with the RX clock recovery being the main source of complexity in the modem. The receivers may need a slightly wider IF filter. The audio must be taken straight off the FM discriminator.
The 9600Bd modems by BayCom are refined versions of James Miller's design, by DF9IC and DK9RR.
This paragraph is based on the author's own experience and makes no claim to completeness or correctness. Follow my suggestions at your own risk. I presume that you know how to set up the BayCom terminal software and that your modem is connected to the right bits on/in your computer. It helps to have another modem/working setup present so you can test RX/TX separately without having to use a radio.
Thanks to their complexity, G3RUH-type modems are generally a bit difficult to troubleshoot. The main obstacle to easy fault finding is the scrambler, which makes it extremely difficult to trace a signal through the modem. The filters are doing their extra bit to prevent you from tracing signals, and the circuitry for clock recovery and DCD isn't trivial logic either. Generally, you will be restricted to verifying that there is some kind of signal of the kind you'd expect.
If your modem is full-duplex capable, chances are it will have a loopback jumper which feeds the TX audio into the RX part of the modem. If you modem doesn't get a response from your partner station, make sure the loopback jumper has been removed - otherwise your own modem transmit branch will swamp the receive path, even when it's just idling along.
If that doesn't fix it, disconnect your radio and put the loopback jumper on. Send a few unproto packets through the modem from the F10 monitor screen. You should be able to tell if they're transmitted because most modems have a TX indicator LED. If you can't see the packets RX'd on the monitor (make sure you're actually monitoring that channel in the software!), try to adjust the audio level trimpot on the modem. Johannes DG3RBU has tales to tell about modems sent in for repair that merely had their audio turned down to zero.
If the loopback still doesn't work, you have a real problem. Unless you have a PIC-based modem, check clocks and follow the data to the scrambler. Make sure all signals surrounding the scrambler look feasible. If you like tampering with the PCB, put a jumper wire across from the scrambler output (last bit) to the descrambler input (past all the RX audio circuitry). Be careful to avoid any coupling with other outputs there as this could damage components. Remove the loopback jumper and repeat the experiment with the scrambler-to-descrambler wire.
If this doesn't work, try working another modem linked up by audio wire and determine if the problem is on the TX or RX side. Candidates for trouble apart from dry joints are defective GALs (have seen one or two of those).
If the scrambler-to-descrambler loopback worked, the problem has to be either in the TX or RX filter circuitry. Use a scope to ensure the EPROM is working. Swap if required. Check output of DAC and the op-amps for correct DC levels.
If the modem loopback works, there must be a problem either with the radio, or with the way the modem is interfaced to it. It helps to know whether the problem is on TX or on RX. A friend's working station nearby comes in VERY handy in such circumstances.
Wiring OK? I've come across weird documentation in the past, so make sure you're connecting to what you think you're connecting to.
TX problems: an obvious source of trouble is overdeviation - rather roughly and easily checked with a hand-held RX (if need be with a dummy load as aerial). Make sure you don't break the squelch if more than 10kHz away from center frequency. This measurement should also indicate if your TX is off frequency.
RX problems: audio level into the modem? Too high/too low? RX off frequency (one of the few situations where synthesized rigs with 5kHz increments do make a positive difference...)?
Ulrich ZL1DDL
Not all of the following modem products have web pages. If you are interested in a particular one, please contact us:
If you understand German, you may be interested in the book "9600 Bd FSK-Technik nach G3RUH-Standard" by Johannes Kneip DG3RBU, which is available from BayCom.
