[Elecraft] On the K3's CW keying envelope shaping, rise/fall times, and CW ALC

[Jan Erik Holm]

Good my 5 ms measurement was the closest so far.

Thank´s for the explanation on the construction,
appreciate it.

After all, why should I bother, if my K3 clicks it
doesn´t disturb me.

/Jim SM2EKM
----------------
On 2010-12-03 17:56, Wayne Burdick wrote:
> There will never be a K3 menu option that allows for clicky CW. K3s
> are at or near the top in every CW contest already, thanks to the
> skill of our customers rather than bandwidth-hogging signals.
>
> There are two factors that control how clean and click-free a CW
> keying envelope will be: explicit shaping by DSP or analog circuitry,
> and incidental shaping due to ALC or transmit chain effects. The K3's
> CW is extremely clean and narrow-banded and click-free because we took
> both factors into consideration. (Note the this is also true of the K2.)
>
> DSP Shaping:
>
> The rising and falling edges of the CW waveform are shaped by the DSP
> using an optimal raised-cosine envelope. Rise and fall times are
> approximately 4 ms, varying only slightly over the entire available
> power output range of 0.1 to 110 W. We experimented with other
> sigmoidal envelope shapes, but the raised cosine was the best overall.
>
> ALC:
>
> The K3 one of very few high-end transceivers that use open-loop
> application of keying envelope shaping. This ensures that the applied
> shape is not compromised by ALC action. Most transceivers, including
> some recently discussed, use fast ALC to control power output level.
> Even if they start with DSP or analog shaping, the ALC jumps in to
> limit the peak amplitude of the transmit waveform, resulting in
> envelope clipping and thus wideband clicks.
>
> So how does the K3's CW ALC work? First, we use a TX gain calibration
> procedure to store per-band gain constants. This information is used
> to preset transmit gain as you rotate the POWER control. When you hit
> the key, we start off just below this target level (about 0.5 to 1
> dB), then use a slow ALC loop to adjust gain to hit the exact level
> requested. Generally the power stabilizes in one or two dits. Shaping
> is excellent on every code element.
>
> We use similar techniques to ensure virtually perfect envelope shaping
> in data modes. In voice modes, we use two-stage ALC -- fast pre-
> crystal-filter ALC in the DSP, slow firmware-based ALC *after* the
> crystal filter -- to ensure that speech signals are completely free of
> splatter. This technique also results in extremely clean and effective
> speech compression.
>
> 73,
> Wayne
> N6KR
>