The choices for RF transmission of digital data are bewildering. FSK, PSK, MSK, QPSK, many with "mark", "level", and "Manchester" qualifiers. Modern high-speed data modems may use QAM (Quantized Amplitude Modulation). For our purposes, however, simple is best. Direct FM of binary data (aka FSK) is quite robust, even under multipath propagation conditions which kill most other data transmission methods.
Data synchronization can be achieved in many ways. For example, a reserved sync pattern is used for the SDLC and HDLC protocols. Remembering that simple is best, we selected a start/stop code using a one start bit, eight data bits and one stop bit. We incurred a 20% overhead right off the bat (10 bits per byte of data), but operate perfectly with a standard PC serial port. Note: this is the same protocol that is is normally used in serial communications to a modem or other RS-232 device (8-N-1).
Spectrum occupancy was a major consideration. We wished to maximize data rate while staying within a single voice channel. FM voice using 5 kHz peak deviation is referred to as 16F3 modulation. The RF channel width equals (by Carson's rule) twice the sum of peak deviation and highest audio frequency (3 kHz is assumed for voice). Note 2 x (5 + 3) = 16. A binary data rate of 9600 bps requires baseband frequencies from DC to 4.8 kHz. Thus if transmitter deviation is reduced to 3.2 kHz peak, the FM channel bandwidth remains at 16 kHz. As before, 2 x (3.2 + 4.8) = 16.
FM voice does not require DC response, but generating FSK from binary data most definitely does. Since DC response IS mandatory, phase modulation MAY NOT be used for at least the data portion of transmission. Direct FM of the transmitter satisfies the DC response requirement for data, as does use of an FM receiver using a discriminator. Imagine a long run of ones or zeros. Without DC response, there would (eventually) be no data. Same argument for strings of mostly ones or zeros. Unless exactly balanced ones and zeros are guaranteed, DC response IS mandatory. That is why the data decoder MUST have a direct connection to the receiver. Passing the discriminator output through a series coupling capacitor prior to the data decoder will not affect the voice signal, but you can forget about receiving FSK data.
Receiver FM discriminators can have either a positive or negative slope.
That is: increasing the input frequency will either increase (positive) or decrease (negative) the discriminator output voltage.
This may have the effect of inverting the data stream (zeros become ones and ones become zeros).
Many newer receivers use an integrated circuit (such as an MC3357P) for a discriminator. These have an "unbalanced" output, that is, an on-frequency signal may output a +2 to +3 volt resting level, and vary about that value as the frequency moves up or down.
Created: 1999 August 7 Last Modified: 2002 December 7