I built my K2 in March 2003, followed by the KNB2, KAT2, KSB2, KDSP2, KPA100 and KAT100. Building, testing and operating the K2 has proved to be a very enjoyable and rewarding experience. The receiver works exceptionally well, and, after some adjustments to the transmit audio I now get very good reports on my transmitted signal. All in all I am was well pleased with my K2 setup, with one exception - the VOX operation.
I operate SSB exclusively, and for most part use PTT. There are times, however, mainly when using the K2 for nets when the use of VOX is advantageous, giving the other station an opportunity to break in and comment without having to listen to a 'long rambling over'.
The basic VOX design in the K2 is flawed. In all other rigs that have a VOX facility the VOX is derived from a separate microphone audio path from that of the transmit audio. This gives a degree of control totally independent of any speech processing. In the K2 the VOX audio and transmit audio come from the Speech Compressor U3 (SSM2165). This IC has a noise gate which reduces the compressor output on low level microphone signals. The effect of this is to give erratic VOX action when the compressor is at the 1:1 or 2:1 settings. This happens even if the microphone audio is sufficient for the correct level into U5 the balanced modulator.
I have investigated running a separate audio path into the VOX comparator U4. I achieved some success, but had difficulties in keeping RF out of the audio circuitry. It really needs to be implemented in a redesign of the KSB2 audio path (KSB2 MKII - how about it Wayne/Eric)?
I have not given up entirely on my approach, but it will have to wait a while. In the meantime I needed to use the VOX.
In common with a number of others I found that the VOX operation was much better at higher compression levels (3:1 and 4:1). This is because the noise gate gain reduction is masked by the increased gain of the compressor. The VOX triggers well - too well... If a station with strong audio came on the transmitter would trip and the resultant sound would be like 'shooting rabbits'. The lack of Anti-VOX became a real pain.
I spent a considerable amount of time and effort with analog simulation and breadboards until I came up with the current solution (MK7). This is the Anti-VOX project described in this site.