After receiving some helpful data from several users, I now believe
this is the complete list of issues with WriteLog's WinKey CW
transmission:
1. WriteLog's initialization sequence for the WinKey was "out of spec".
It needed to not have a "power up reset" command and it needed a
pause after its "open host" command. Fixing those appears to fix the
worst problems. Certain winkey/comm port/OS combinations had a problem
with the init sequence because the WinKeyer never successfully
executed its "open host" command and therefore never transmitted
any status to WriteLog. Their symptoms were catastrophic in the sense
that CW transmission just stopped after transmitting a few characters.
It turns out that older versions of WriteLog have had this
initialialization problem and would consistently display "Timeout" instead
of WPM in WL's status line, but only recent versions of WL would also fail
to send properly CW in this situation. Note that this problem
was repeatable on a given hardware/OS configuration but extremely
sensistive to such a configuration: moving the WinKey from one
COM port to another on the same machine could make the
probelm disappear, especially if the other port had different hardware.
The vast majority of user configurations don't see this problem at all.
2. When WL 10.84 introduced the [winkey] BufferLevelSetting=2 INI entry,
the default value (the value a user gets when there is no entry in
the INI file) was mistakenly set to maintain far fewer
characters in the WinKey than prior versions. (Only 3 characters
ahead rather than 16 to 32). This caused some configurations with slower
CPUs and with lots of WL features enabled to experience short pauses (or
extra spaces) in CW transmission when the WinKey got to its last character
before WL gave it a new one.
3. With certain Multiplier modules loaded (GAQP and MNQP have been
identified as likely ones with others possible), and with packet
connected, WL spends far-and-away too much CPU doing multiplier
checks on incoming packet spots, especially when the user
presses SPACE after entering a CALL. This is a problem in and
of itself, but it also causes WinKey transmission to get extra
spaces for a low BufferLevelSetting. Note that WriteL32.exe should
almost never show more than a few percent CPU usage on Task Manager.
The symptom of this problem is that it goes to 100% (or 1/N of 100%
on a machine with N CPUs).
#1 and #2 will be addressed in the next 10.85 beta.
#3 is still under investigation.
Thanks,
Wayne, W5XD
_______________________________________________
WriteLog mailing list
WriteLog@contesting.com
http://lists.contesting.com/mailman/listinfo/writelog
WriteLog on the web: http://www.writelog.com/
|