[Elecraft] Sequencing a chain of equipment with a K2 at the head.

[Guy Olinger, K2AV]

This is an option request, and a warning about driving QRO amps with a
K2. It also bears tangentially on several other recent timing threads.

There are very few rigs on the market that address the sequencing
needs that follow keying a CW baud into a rig that is driving other
devices.

The K2 is fine by itself. The K2/100 or K2 with the separate 100 watt
amp is fine because the timing of the K2 and the amp is controlled by
the K2 CPU.

After that, things get dicey.

When you hit a dit, the *RF* delay through various amplifiers is
almost non-existent, measuring a few cycles at *RF* frequencies. All
the delays are electronic, measured in MICROseconds.

The overall OPERATIONAL delay, going from receive state to transmit
state is another matter, taking MILLIseconds, up to 25 milliseconds,
and sometimes more depending on the type of switching devices, or
whether mechanical devices are chained.

In the good old days, there was no QSK. The operator hit a switch
which threw relays, and by the time the hand got from switch to key
all was in transmit state.

In a manner of thinking, that hasn't changed. We just want the RIG to
throw the switch for us, and we want it to throw it fast enough that
we can hear between the elements of our sent CW.

That means that the paddle dit or dah, is the SAME signal both to key
the RF *AND* transfer state from receive to transmit.

Without specific attention to the timing issues, the RF will propagate
in microseconds and be active BEFORE the mechanical changes can be
completed in milliseconds.

Problems with this are terrible clicks on beginnings of bauds,
destruction of transfer relay contacts or switching diodes, and in
some cases arcing in high power amplifiers and destruction of
bandswitches and expensive final tubes. A lot of fuse-blowing episodes
blamed on parasitics are really RF-applied-before-state-change arcing
episodes.

That is not to mention chopping off the first part of a baud or
eliminating it altogether by the time the amplifier switches and is
amplifying RF, and all the accompanying CW misinformation perceived by
the receiving station. K0 becomes U0, W0 becomes M0, etc, etc,

Many attempts at QSK with these problems merely REDUCE the LENGTH of
overlap and clipping, and do not SOLVE it.

Some contest logging programs that drive separate key and relay lines
use program delays to apply the relay signal an adjustable interval
before the key signal. The program user determines the delay needed
and sets one of the program startup variables to that value.

The FT1000MP has the delay feature built into its firmware. It is
adjustable in its internal menu from zero to 30 milliseconds. It is a
bucket brigade delay, meaning that the entire string of code is sent
delayed by that amount, without any chopping anywhere. Properly set to
the delays needed, RF is not applied until the RX to TX changeover is
complete. No chopping, arcing, no point, bandswitch or tube
destruction. This works for either QSK or VOX style operation.

Whether this operation is considered "CW VOX" or QSK, depends on
another delay, most like SSB VOX delay, which delays a return to RX
until a certain interval has passed since the last transmitted baud.
If this "hold" plus the startup delay is less than the time between
bauds, it's called QSK.

At 30-35 WPM in contests, QSK and practical delay settings (that don't
have RF already applied to closing contacts or diodes in the middle of
state change) require very fast electronic non-mechanical switching
that has been carefully engineered to SETTLE quickly.

At 35 WPM practically, mechanical changeover (including most so-called
QSK relay switching) can be set only tight enough to hear between
letters, but not between bauds. This has become my preferred QRO
contesting mode. I find that I can always hear enough between letters
that true QSK does not have much of an operational advantage. It also
reduces the amount of mechanical switching significantly.

The K2 does not have a bucket brigade delay built in. It relies on a
known starting delay and finishing delay to properly time an
individual baud through it's (and the matching 100W amp's) circuitry.

The nearest thing to a pass-along timing signal from a K2 would be:

Dot paddle low *OR* dash paddle low *OR* 8R low gives pass-long low.
The OR function can be done with diodes, but it would need to be
buffered because nearly all of the downstream devices require you to
GROUND *their* voltage.

And this will only work with devices capable of "QSK" state switching
speeds.

If you add a variable VOX delay style "hang", more devices can be
switched, but hardly all.

To be universal you need the functions implemented in the MP, 1) a
bucket brigade variable delay AND 2) a variable "hang" delay.

This COULD easily be an option for the K2, but it's not at the moment.
It's hardly a universal or wide spread need among K2 owners.
Transverter and QRO amp users only, it would seem.

73, Guy.