[Elecraft] K3 Keying

[P.B. Christensen]

> "Just done a quick 'scope check here using my Microham cw keyer and the
results are 'inconclusive'. In qsk the space is actually slightly less if
measured from the end of the fall to the start of the rise, but slightly
more if the dit is measured across the 'top' ignoring the rise/fall time. I
forgot to check using the internal keyer but this seems pretty good to me. 
Having said all that I see that I have configured my Winkeyer with 3ms
'keying compensation' which was done on the basis of it sounding a little
choppy previously..."

I modified my microHam CW Keyer to produce a buffered +5 VDC pulse for
triggering a dual-trace scope.  The most recent K3 firmware shows that the
CW envelope rises ~ 13 ms after external contact closure.  And more to the
point, the RF envelope matches the duration of the external key closure when
measuring from the beginning of the envelope rise to the trailing edge. 
This closely matches other radios, including many Ten Tec models, and
precisely matches the envelope of my Icom 7700 and 7800 (with latest
firmware).

It's interesting that you use 3 ms of added external compensation as that is
my value used on all three transceivers since their envelopes happen to
perfectly match in duration.  My personal preference would be to have all
manufacturers match the envelope time to the contact closure referenced at
the time the rise ramp completes, to the beginning of where the envelope
begins to fall.  If you think about it, the combined rise/fall comprises a
full 10 ms in the K3.   For that reason, compensation of 3-5 ms sounds less
choppy to the ear and more closely matches that of the perfect 1:3 ratio
generated by an electronic keyer.  Since the required compensation is
constant as a function of keying speed, the envelope only needs to be "pulse
stretched" to one value, regardless of key speed.

Psul, W9AC


-- 
View this message in context: http://n2.nabble.com/K3-Keying-tp840393p841397.html
Sent from the Elecraft mailing list archive at Nabble.com.