Another essential thing for the CW or SSB operator is the method of listening to own sent signal. There are two options:
1. SIDETONE – the resulting beat note does not need to be exactly the same as the offset between the exact signal frequency and the frequency to tune the receiver to obtain an audible beat note, however it is a BIG ADVANTAGE when the offset is identical to the resulting beat note. MORE IMPORTANT is that there should be ZERO LATENCY between paddle touch and sidetone audio output. The main purpose of this is NOT monitoring the actual signal quality, it is to provide the necessary feedback from hand to ear (brain) while composing the character. To send out perfect, error-free Morse code, a high level of synchronization is needed. Every CW operator has his internal “clock” which must be synchronized with resulting signal (ie. frequency of dits). Any latency presumes to be subtracted from the audio output heard. As a result of observation of hundreds of CW operators (any knowledge level) is that any latency between a paddle touching and audio output > 10 ms is a disadvantage at speeds above 35 WPM, the operating become to be tiresome and exhausting.
2. MONITOR – the actual state. The actual output signal is MONITORED concerning frequency, shape and sidebands however the is some latency. My impression from operating at speeds > 35 WPM is that the output signal is “sampled”, ie. should be exactly synchronized with a “clock” which is independent on the used speed (this is only option possible when using an external keyer). It sounds like the resulting dits does not have an uniform length but are “variable” depending on the used speed. Like how the speed matches the internal “clock”, in some cases the dits CAN have an uniform length. When using the internal keyer this problem disappears.
It seems that the ideal solution will be the pure MONITOR when using the internal keyer but when using an external keyer a SIDETONE with zero latency between key-on state and audio output. Of course, there should be two selectable options - MONITOR and SIDETONE. Anyway if the relationship is “sampled” in either way use the highest possible sampling frequency, always > 1 kHz (remember that 100 Hz means reciprocally 10 ms which is TOO MUCH!).
This latency also badly affects the SSB operation. The resulting delay may force the operator to “unnatural voice performances” like yelling, speaking with artificial, strained voice and even falter (stutter) speech. The MONITOR latency seems to be one of the worst, badly and deeply affecting the operator’s performance.
This article is a stub. You can help by expanding it providing more/new ideas and suggestions. Don’t hesitate to borrow a good idea from an older proven radio.