May be aggressive is a subjective assessment ,
If CW tx , could be described as Hard/Aggressive Keyed , that would give the same results , Ie , Soft keying / envelope shaping , produces , less side bands or Click's , less envelope shaping , produces a much harsher / aggressive / wider signal
A good comparison is the original MFSK mode and, say, ROS-HF ,
In terms of the on-air presence , MFSK and ROS are ''identical'' both use single tone , both change the tone frequency and duration, as directed by the resolve of the mode coding and modem design , both are 'base band heterodyned systems ', each may transit via, none linear systems [ OLIVIA has a phase component to the modem modulation scheme , and as such , requires linear amplification ]
But, where as the MFSK mode , produces a set of near in ‘side band's’ , [ even marked on the spectrum plot in Patrick’s Multi-psk, as additional B/W ] , ROS-HF can be seen to be clean and producing only the on air tone set .
The modem in ROS is Phase continuous , may be other systems are also, but , the ROS system , embodies design features that are missing/not as developed from other systems , ''Gray noise'' is the general term
The audio level supplied to the Transmitter , remains constant ,
However , fluctuations may be observed on the TX power meter , these , appear to be a by product of the filtering in the TX . the abrupt changes in Tone , appear to be subjected to the filter characteristics , phase shift , group delay etc, which in turn , adds phase / time distortion , the resolve of which is amplitude variation
Racal listed various option fit filters , some , voice , some data , so i suspect the problem was known about some time back , two filters in the tx/rx path , 40 MHz roofing and the 1.4 MHz pass/band , using the widest possible filter , can reduce the ‘side bands’, the DSP compensates for disturbance , but , from a transmission aspect, its better to maintain the ''waveform''
That probably explains the different on-air signals , interims of additional side bands , from a single tone , or at least that’s my general observation , Tx [ and unseen RX issues] serve to degrade the signal quality , the claim, no filter is required , hold quite true , if the DSP is correctly designed ..