Occupied BW of CW
There seems to be some lingering debate about keying speed affecting occupied bandwidth of a transmitter. I'm fortunate enough to have equipment that directly measures occupied bandwidth. The FCC accepts this currently certified equipment as proof of bandwidth.
The following printout is from an unmodified IC-751A running at 100 watts carrier output on 3.5000 MHz.
The data is taken at 10, 25, and 40 dots per second with 50% duty cycle. This is normal CW weight.
The occupied bandwidths are:
We see the actual measured occupied BW of a relatively good CW transmitter is essentially the same and does not track CW speed. By the way, the small ripple in occupied BW is caused by a small envelope shape change. This is an artifact of the ALC system in the 751A.
In the continuing e-Ham thread, one person claimed the reason the IC-751A did not track bandwidth with speed was the fact an IF carrier oscillator was keyed. The fastest way to solve that argument is to test another totally different rig. The following is data from an FT1000MP MKV with keyclick reduction mods. These mods are outlined on my 1000MK V page.
The following is the occupied BW of the FT-1000MP MK V
The measured occupied BW's follow:
If you look at the chart and plots above, it might appear the 5 dps is a flyer or error. Actually it is a good reading and cross checks with a selective level meter.
Dots at 5 dps are wider because the ALC releases between dots. The ALC alters the waveform, producing a very sharp roll-over at the leading edge of every dot. The sharp edge on the attack substantially widens the bandwidth. At higher speeds the ALC stays constant and does not sag between dots. This allows the shaped waveform to flow through the system without additional sharp edges.
This is additional proof the edges are everything for determining bandwidth, and the speed only has an incidental effect as it modifies the waveshape of the rising or falling edges at radio performance speed limits.
I've measured dozens of radios and they all follow this trend.