Hi Tony 

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 .. 

73-Graham
g0nbd