Posted by
Kok Chen on
Dec 02, 2010; 4:45am
URL: http://elecraft.85.s1.nabble.com/Fwd-New-Sherwood-report-tp5793377p5794811.html
On Dec 1, 2010, at 5:59 PM, Barry N1EU wrote:
> It's absolutely amazing, after years of Yaesu being called out and doing nothing about key clicks in their rigs, that they would bring out a radio
> (FT-5000) and provide the user the ability to reduce the cw rise-time to 1 msec (menu mode, cw group, 063 A1A Shape). Just incredible.
The rise time by itself is not the important factor -- what is much more important are first and second order discontinuities, and even higher order discontinuities.
If you were to generate a keyed signal that turns on with a constant slope (thus has a large second order discontinuity), you are going to generate very wide keying sidebands even if that "risetime" lasts for 10 msec.
For a modern view at CW keying, take a look at Alex VE3NEA's article in the May/June 2006 issue of QEX that is titled "CW Shaping in DSP Software." Alex is of course the author of the CW Skimmer, among other things.
This is not just theoretical stuff. cocoaModem on Mac OS X is one program that generates a CW signal using the J2A Emission mode by using a Blackman window whose keying sidebands you can see in Figure 5 here
http://homepage.mac.com/chen/w7ay/cocoaModem/UsersManual/cwManual/index.html#filterA Blackman window is initially wider than say, a Hamming window (and certainly much wider than an unshaped pulse), but then it plunges down towards -100 dB with a very steep fall off. There is nothing like it in the analog world :-).
In his article, Alex had compared the Blackman-Harris window with Gaussian, raised Cosine, and other windows.
Wikipedia has a very nice page on filter windows (many people use the windowed method to design FIR filters) here
http://en.wikipedia.org/wiki/Window_functionand this plot from Wiki is especially useful if you want to homebrew your own "CW shaper":
http://en.wikipedia.org/wiki/File:Window_function_(comparsion).pngNotice the asymptotic slopes of the curves towards the right of that plot. The steeper it is, the less you will QRM stations that are far away. The shape towards the left tells you how much close-in bandwidth you are using.
For example, the unshaped pulse (black line) has the narrowest close-in bandwidth, but it is also the worst when it comes to far away bandwidth. At 100 times the normalized bandwidth of the filter, pulses that are waveshaped by the Blackman and Blackman-Harris windows are a whopping 80 dB quieter than un-waveshaped pulses. You can think of the normalized bandwidth as what is needed to pass the fundamental N-words-per-minute keying sequence without sounding too soft.
For what its worth, cocoaModem lets you dial in an equivalent risetime of 2 msec all the way to 10 msec (for QRS slowpokes like myself who want to cause even less QRM).
You need a reasonably good transmit IMD to take the most advantage of good waveshaping. All said and done, it is not the DSP part that is the limiting factor of what you can do with waveshaping CW pulses today, but the transmit IMD.
73
Chen, W7AY
______________________________________________________________
Elecraft mailing list
Home:
http://mailman.qth.net/mailman/listinfo/elecraftHelp:
http://mailman.qth.net/mmfaq.htmPost: mailto:
[hidden email]
This list hosted by:
http://www.qsl.netPlease help support this email list:
http://www.qsl.net/donate.html