[TenTec] Pegasus/Jupiter keying - In search of Perfect CW :-)

Tom Rauch W8JI@contesting.com
Sun, 3 Feb 2002 09:20:37 -0500


Hi Paul,

> I'm not sure of the methodology,  but I recall a graph from a Navy
> study in which keyed envelope shape rise/decay was plotted against
> minimum discernable signal strength.  Another graph depicted CW speed
> against MDS conditions as well...and a third graph combined both
> factors.  I'll look for this...

Look for it and let me know. I have the CCIR data, and at 35 WPM 
under severe fading you might start to notice a change once in a 
while during fades, but that would only be true if receiver bandwidth 
was also wide. 

We are doing a disservice to people who operate CW by giving 
operators the impression they will "get out" better by using hard 
keying. The ONLY possible improvement will be while in the fade, 
and any improvement will be unnoticeable with a 5mS rise and fall 
at 35WPM. 

As I said before, having the high order sideband does no good at all 
if we are using 500Hz or narrower filters. We give up S/N ratio in 
direct proportion to the bandwidth increase, if we use a 1kHz filter 
S/N is 6dB less than with a 250 Hz filter. If we use a 250 Hz CW 
filter, the rise and fall limit is about 5-6mS with a sine shape rise 
and fall, any sidebands wider than that simply do not make it 
though the filters.  

What that means in plain language is if we use a 250Hz filter, 
everything we hear (even if sent with a square rose and fall) sounds 
**exactly**  like it was generated with a 5mS or so rise and fall.

As a matter of fact, that is now how I generate my CW signal. I run 
a hard CW rise and fall (under 1mS) into cascaded 250Hz filters at 
the IF frequency of the transmitter. Viewed on a scope, the result 
is a 5-6 mS rise and fall.

JT1CO and a dozen JA's had no trouble copying that sound 
through noise on 160 meters today.
 
> Another interesting source is found in my 1998 ARRL Handbook on page
> 15.7:

That was all written by Bill Sabin. The waveshape he chose is NOT 
a filtered dot, it is instead the shape of a single pole R/C filter. 
Such shapes were about all we could muster in the years of 
vacuum tubes, before we had op-amps and narrow bandpass filters. 
The shapes shown as "ideal" in the Handbook are certainly not 
state-of-the-art shapes even by early 1980's technology!   

The important thing is Bill trusted the CCIR study, which is what I 
recommend we all follow. 
 
> through noise and fading...A transmitter's CW waveshaping is therefore
> usually hardwired to values appropriate for reasonabley high-speed
> sending (35 to 55 WPM or so) in the presence of fading.  As a result,
> we generally cannot vary keying hardness at will as we might vary a
> voice transmitter's modulation with a front panel control (W9AC: two
> notable exceptions include Kachina and Kenwood's menu-selectable
> DSP-based CW waveform generator).  Rise and fall times of 1ms to 5ms
> (5ms rise and fall times equate to a keying speed of 36WPM in the
> presence of fading and 60WPM if fading is absent) are common."

That part Sabin clearly extracted from CCIR data. At 5mS rise and 
fall with FILTERED dots, the improvement by using a faster rise 
and fall is immeasurable at 35 WPM with fading signals and 60 
WPM with no fading. 

Not only that, we should keep in mind the only error change would 
be during a fade while you still had some S/N ratio.

With that in mind, why should we give people the idea it is 
advisable to use UNfiltered dots and 2 or 3 mS rise and fall? Do 
you really think people are going to work 70WPM on fading 
circuits, and 140WPM on non-fading circuits? Do we really think 
missing an occasional extra dot at 35 WPM on marginal signals is 
worth making a change that causes us to occupy twice the 
bandwidth?  

A 5mS rise and fall causes a 500 Hz tone to reach full amplitude in 
2.5 cycles, a time far too fast for anyone to notice as "soft".

The last thing in the world we need to do is to keep junking up the 
bands with clicky signals. Allowing people a choice of using a CW 
rise and fall capable of 150WPM is not really necessary, and 
probably would result in the same situation we have now.

What might be acceptable is making the transmitter follow the 
receiver bandwidth, then if people like to listen to hard keying on 
and empty band they can pick a wide filter and maybe a 2mS rise 
and fall. When they pick a narrow filter on crowded bands, they 
won't be trashing other people.

Allowing transmitters to be wider than receivers wastes spectrum. 
It doesn't help a thing.
73, Tom W8JI
W8JI@contesting.com