In the BITX20 yahoo group, Farhan responded to a question about the BFO and the crystal filter in the BITX 40 Modules. In his response we learn more about the work of the members of the women's collective. Uma is the name of the lady who has the job of sorting the crystals. She sorts 1000 a day:
There is a longer explanation to this. The crystals we use in the BITX are microprocessor grade. This means that they are not very precise. Uma is the crystal sorter. She takes a bunch of 1000 crystals a day and sorts them by their frequency into different bags. Each board takes its five crystals from any one bag. Hence, the central frequency of each board will be slightly different from another board.
The central frequency of the ladder filter moves down from the oscillating frequency of the same crystal. That is why, in a set of matched crystals, one can be directly used as the carrier/beat frequency oscillator without needing any trimmer : it directly sits on the higher side skirt of the filter.
The BITX40 board's BFO usually ranges from 11.990.0 to 11.997.0, though in almost all cases it is very close to 11.998.0. The best way to determine the BFO frequency is to take it another ham's shack, tune in the BFO to zero beat on USB or LSB and note the frequency. I use the Rigol scope's built-in frequency counter to measure it off the modulator transformer's primary.
So, the IF offset should ideally set to the measured BFO's frequency. These are however, very subjective choices. Given that 2.2 KHz is not an ideal passband, 3 KHz is more like it, setting the BFO will determine how you would like to hear the receiver. Setting it close to the passband will make it bassy, setting it away will make it tinny. Setting it a few hundred hertz away will make it hollow. You get to choose which way you want to make your radio sound bad (harr! harr!)