Tom Holden wrote:
ARRL Handbook says that the BFO frequency should be at
the -20dB point down the skirt of the SSB filter. Pete,
KE9OA, says -24dB.
1. What is the reasoning behind these choices?
2. How critical is it that the BFO be so positioned?
3. Wouldn't the shape factor have some bearing on where
it should be?
4. Is linear interpolation between the -6 and -60dB
points accurate enough to determine the BFO freq?

Thanks for all the useful replies! I neglected to explain that this question
is related to upgrading a particular receiver and another design target is
that the BFO frequency must be fixed at +/-1500Hz from the filter and IF
centre frequency (1 side for USB, the other for LSB) due to the way this
double conversion receiver tunes and displays frequency. So it is more a
question of determining what filter bandwidth and shape would be most
suitable. By targetting -20 to -24dB attenuation with this offset, I
computed that example filters suited to that offset would have -6dB/-60dB
bandwidths and corresponding audio passbands of:
1) 2kHz/5.4kHz ----- 500-2500Hz
2) 2.3kHz/4.7kHz --- 350-2650Hz
3) 2.5kHz/4.2kHz --- 250-2750Hz
Do these calculations seem reasonable?

Examples 2 and 3 seem to be acceptable for communications speech but the
steeper the skirts the greater the risk that filter tolerances will place
the BFO somewhere else on the skirts well removed from the -20dB target. Are
there any economical 455kHz filters with the characteristics of examples 2
or 3 with symmetrical skirts?

TIA ,
Tom