Setup for TNC Tests
Here's the setup I used to verify the facts, and generate the 'scope pictures, in Setting Your TNC's Audio Drive Level.
An MFJ-1270C TNC was hooked to a Kenwood TW-4100A transceiver (this is a 1988-vintage dual band FM rig) using an MFJ cable kit. The RF output of the radio was fed into an HP-8920B service monitor, which measured the deviation and was used to generate the oscilloscope pictures.
The TW-4100's pre-emphasis circuit isn't perfect; it provides a bit too much boost. The desired pre-emphasis between 1200 and 2200 Hz is 5dB; the TW-4100 boosts the high tone by 7.2dB. As a result, when the high tone deviation is set to 3kHz, the low tone will deviate about 1.3kHz instead of the 1.65kHz that would be expected.
The receiver used for the received-audio pictures was an Icom W-32 HT; the speaker jack was connected to an external 8 ohm speaker, and the audio was taken across the speaker terminals. The W-32 also has a bit more deemphasis than it should (about 6dB from 1200 to 2200Hz), but not enough to completely cancel out the over-boost provided by the TW-4100.
The fact that the transmitter's boost is a bit more than the receiver's cut accounts for the received tones showing a bit of twist toward the high frequency.
The HP-8920 has a single sweep oscilloscope mode, and a very nice trigger delay function. Since it's a digital display, the captured sweep remains visible indefinitely, and can be dumped to a printer. Using these features, I was able to capture the audio waveform in the middle of the packet, after any keying transients had settled down. Each picture required several tries, as I was trying to capture an image that included several cycles of each tone. In a few cases, I moved the trigger delay around a bit to try to find a good spot to sample.
If you're reading this note, you're probably at least a little interested in the technology of packet radio, so I'll mention here that there's no fixed relationship between data values 1 and 0, and the high and low tones in a packet signal. Packet uses "NRZI" (Non-return to zero, inverted) coding, which relies on transitions between states, and not the actual value of the state itself.
As a result, you can't talk about one tone representing "one" or "zero", or even "mark" or "space." Although the relationship remains the same within one packet transmission, the modulation in the next packet may be inverted. In one packet, it may seem that the high tone predominates, but in another transmission the same bit pattern may be inverted, with the low tone predominating.