[WT-support] CW speed

[Floyd Sense]

Well, I set up a test environment today for the mysterious CW speed problem, 
but today the speed change works as it should.  The PC is running exactly 
the same set of programs as yesterday, but with a CPU monitor on.  The 
highest observed peak CPU utilization was 12.5% on my AMD 2200 processor.  I 
cannot draw any conclusions about what might have been going on yesterday. 
During the period that Win-Test exhibited this behavior, I stopped and 
restarted it several times and used WriteLog in between to verify proper 
operation of its CW generator and saw no problems there.

While the software controlled keyers such as Winkey, microHam, etc. are 
interesting, I think they appeal primarily to users of software whose 
authors have not been able to come to grips with the difficulties of 
real-time programming.  Both WriteLog and N1MM Logger never miss a beat 
while editing previous contacts and simultaneously sending high-speed CW 
from the F keys.  I've never observed a stutter in either of those packages. 
On the other hand, many USA hams have stuck with their under-powered older 
PCs and very likely do see the problems that these semi-hardware solutions 
address.

Thanks for your assistance, and will see you in the contests.

73, Floyd - K8AC


----- Original Message ----- 
From: "Laurent HAAS - F6FVY" <f6fvy at free.fr>
To: <support at win-test.com>
Sent: Thursday, November 18, 2004 3:56 AM
Subject: Re: [WT-support] CW speed


> Hi Floyd
>
> Floyd Sense a écrit :
>> Maybe we have stumbled across something here.  I am not using a parallel 
>> port - am using a serial port.  You are right - a measurement is in 
>> order, but I assure you that I can tell the difference between 30 and 32 
>> WPM.  I will put a scope on it in the next day or so and look at a string 
>> of Es and see if it changes at all.  The other thing that I did not 
>> mention is that the speed sent is much slower than the speed that is 
>> indicated.  When using WriteLog on the same hardware, the CW speed very 
>> clearly changes and the indicated speed is accurate.
>
> The fact that the speed sent is much slower than the displayed speed may 
> be a clue that a so-called heavy load on the CPU (or OS) slows down the 
> internal timer. In fact, there are so many factors affecting the inside 
> timer accurancy. Not so easy to remotely decipher this...
>
> Anyway, with so-called modern OS, and the APIs Mr Gates is offering, it's 
> almost impossible to build "real" realtime processes.
>
> We rather focused on the ability to type ahead (which is not always 
> possible, even in WriteLog, if I remember well), and having a very 
> convenient user interface (try to modify your previous QSO for a typing 
> mistake in the middle of a 400 Q / h pile-up with other packages...), even 
> if the CW speed (which will be very occasionaly modified) isn't perfect... 
> To be honest, working a mult at 32 WPM (or was it at 36 ?) is not my 
> concern : To work him _IS_ my concern :-)  Got the picture ?
>
> BTW, Olivier already implemented the W5XD multikeyer interface, and also 
> intends to add the WinKey (K1EL) interface in a near future (by chance, it 
> will add compatibility with other hw packages like the great MicroHam 
> microKeyer and others).
>
>> I don't have a problem with screen size, but did change to the floating 
>> windows option.  I found that allows me to position the entry area of the 
>> log window where I want it on the screen (lower right corner).
>
> And in this mode, WT doesn't hide other useful applications windows. I 
> know users who keep chatting with their buddies on IRC, while running 
> contests ;-)
>
>> Do you think the CW speed problem might be related to use of the serial 
>> port instead of the parallel port?
>
> Franckly, I don't know. I always used the parallel port output (my // 
> interface is wired for years !).
>
> 73
>
> Larry - F6FVY
>
> _______________________________________________
> Support mailing list
> Support at win-test.com
> http://www.f5mzn.org/cgi-bin/mailman/listinfo/support