我今天在看讨论组的信的时候看到了这个话题，想了一下还是放在dx版比较合适，dxer尤其是contester，更加应该注意自己的信号情况。
如何调整自己的机器，使之正确地工作是我们每个人都应该学会的。
抱歉是英文...
-------------------------------------------------
>i suspect that over-driving an external amplifier is a major cause of
> excessive clicks on the bands today. many hams think that cw doesn't
> require
> the final amplifier be linear and drive their rigs into saturation. but a
> "linear" amplifier needs to be linear for cw just as with any amplitude
> modulation (ssb, am, cw, etc.)
key clicks also occur when bias is suddenly applied to an amp where the
transceiver is already producing an rf envelope. i recently experienced
this when i purchased the spe 1k-fa solid-state amp. the amp is advertised
as being "full qsk" compatible. however, it uses general purpose frame
relays at the rf input and output. as such, the firmware is programmed to
apply bias at a very long 15 msec after the amp's relay line is pulled low
by the transceiver. the problem isn't terribly bad with a transceiver like
the k3 as its rf envelope can be delayed in the k3 menu. however, rigs like
the icom 7700 and 7800 produce rf at ~ 9 msec. after key closure and there's
no menu to adjust it. on the scope, wysiwyg.
in the k3's default setting of 8 msec., my spe amp was hot switching as a
result of bias being applied at 15 msec when the k3 was already producing
its rf envelope at exactly 12 msec after key closure. and so, the rf
envelope was being truncated by 3 msec in full qsk mode. spe chose that
bias time to give the general-purpose frame relays an opportunity to settle
without contact bounce. sure, the amp's relays were not being hot-switched,
but the effect of key clicks is the same when the sudden onset of bias
causes the fet output transistors to instantly conduct, producing a sharp
vertical rise time of the rf envelope.
i then delayed the k3 to 16 msec and the hot-switch problem disappeared.
but in doing so, the keyed elements were also being shortened, albeit
without rf hot-switching. that fix requires use of pulse-stretching key
compensation using an external keyer. next, i tried using the k3's tx inh
handshake with the spe amp. more dit shortening problems. was the problem
with the k3 or the spe amp. i honestly don't know and don't care. i only
care that there's a viable solution is available. really, the only way to
cleanly fix these problems is at the source of the timing issue.
the solution: i ended up modifying a beat-up ameritron qsk-5, keeping the
logic interface between the qsk-5 and the spe amp100% optically-isolated.
http://72.52.250.47/images/qsk-5-interface.pdfnow, instead of the amp's firmware setting the bias activation time, the
qsk-5 tells the spe amp when to apply bias. qsk is pretty awesome with this
set-up and even the first leading 'dit' switches cleanly with no keyed
element shortening. long-term, the fix calls for using a gigavac gh-1
vacuum relay at the amp's output and an aromat rsd-12v rf reed relay at the
amp's input. spe has sent me the revised firmware that reduces bias time to
8 msec, but that requires an eeprom programmer and a spare chip that i need
to order. but with either solution, bias key clicks won't be a problem.
paul, w9ac
