''At first glance it appears crazy''

Mike,

Welcome to the crazy world of  JR  !

The  basic  explanation is the  narrow filters  increase the  noise power 
in a limited   bandwidth , ringing  due to the  Q , similar to  CW , narrow 
filters tend to  round  the  CW signal

The DSP  engine  is  better equipped to  differentiate between  carrier and 
noise and  has a  much  greater  dynamic  range , so  optimum results  are 
obtained , when the  signal is  presented  to the  interface, as close to to 
the  original as  possible , the  DSP filter profiles are  tailored to the 
mode/speed in use

This can be  noticed  with the  new  generation of   SDR  support software , 
where  audio/voice recovery  can be  superior to conventional hardware 
based systems

With hardware filtering , there  are  transit  (group) delays which  can 
alter the  amplitude / time  , either  from  on/off  keying  or  with FSK, 
this can  be  observed when  sending  wide  band  FSK , although  the  audio 
level  remains  constant and  'phase continuous'  , its possible  some times 
to  see a  'am modulation'  envelope  on the  carrier    ... $$$$   can 
solve this  problem , but  for  most  Ham  kit its  something  that 
'happens'

Physical  constraints , if  there  is  a  very large  carrier in the  pass 
band  and  this  is pushing  the  hardware into  non-linearity / A/D  to 
over  range  , giving   quantising  errors, then  , yes  filtering would 
help , but  its more  likely  its the  analogue  path  that's  causing 
intermod products .....reducing  the  rf/if gain will  provide the  solution

I think that's  about the  picture , if  Jim's  about , im sure  he  will 
fill in the  gaps

73 -G..




--------------------------------------------------
From: "Mike Dennison" <[email protected]>
Sent: Wednesday, July 11, 2012 10:18 AM
To: <[email protected]>; <[email protected]>
Subject: Re: LF:  OPERA  Question

>> Narrow   IF filters  are  not  desirable  and   reduce the
>> performance of the  demodulator  , better  simply  use  SSB  filter
>> GL ..73 -G..
>
> Graham,
>
> Why is that the case? What does the considerable extra bandwidth
> achieve? Is the SSB bandwidth optimal, or would it be even better
> with 10kHz bandwidth, or 100kHz?. Why is 3kHz better than perhaps
> 1kHz or 2kHz?
>
> At first glance it appears crazy to let in all sorts of adjacent
> channel QRM (the bandwidth is more than ten times the size of the
> entire Opera window) when using a mode that occupies a fraction of
> 1Hz.
>
> Is it simply that the 1.7kHz Tx tone is high enough for the SSB
> filter to kill its harmonics, and on receive it is difficult to get a
> 1.7kHz tone out of a CW filter, even with passband shifting.
>
> Am I missing something?
>
> 73 de Mike, G3XDV
> g3xdv.blogspot.co.uk
> ================
>
>
>