It was getting late when I pressed send, a couple minor errors crept in:

>  The 12mhz filter is passing audio between roughly 500 and 2500 hz in its upper sideband,
>  so the 12mhz filter is probably centered around 11996500 + (500+2500)/2 = 1199800 hz.
>
>  With a clk1 for USB of 56995000 hz, the center of that upper sideband in the 45mhz filter
>  would be 56995000-11998000 = 44997000 hz
>  With clk1 for LSB of 32995000 hz, the center of that lower sideband in the 45mhz filter
>  would be 32995000+11998000 = 44993000 hz.

When I say "upper sideband in the 45mhz filter", I'm talking about the upper sideband
of the original signal.   Should rewrite those last few lines with a few more words:

  With a clk1 for USB of 56995000 hz, the center of that sideband within the 45mhz filter
  (now below the carrier due to sideband inversion from the high side VFO)
  would be 56995000-11998000 = 44997000 hz
  With clk1 for LSB of 32995000 hz, the center of that sideband within the 45mhz filter
  (now above the carrier due to sideband inversion from the high side VFO)
  would be 32995000+11998000 = 44993000 hz.


And I flipped the arithmetic at the wrong spot in these formulas:

>  The display shows the frequency of the suppressed carrier, always easy enough to compute:
>      For USB, it's   VFO-(clk1-BFO)
>      For LSB, it's   VFO-(clk1+BFO) 

Those two formulas should read:
 
     For USB, it's   VFO-(clk1-BFO)
     For LSB, it's   VFO-(clk1+BFO) 


Jerry, KE7ER

On Sat, Mar 10, 2018 at 11:36 pm, Jerry Gaffke wrote:

I think Gerald's complaint about USB sounding different than LSB winds up being due to
a USB signal going through the 45mhz filter a few khz away from where an LSB signal does. 
Gerald was using Ian's code which I have not yet looked over.
I assume Ian borrowed numbers from the original uBitx code.
Here's my argument:

The original uBitx code has the BFO at
    #define INIT_USB_FREQ   (11996500l)
and uses these frequencies for clk1:
    #define SECOND_OSC_USB (56995000l)
    #define SECOND_OSC_LSB (32995000l)

The 12mhz filter is passing audio between roughly 500 and 2500 hz in its upper sideband,
so the 12mhz filter is probably centered around 11996500 + (500+2500)/2 = 1199800 hz.

With a clk1 for USB of 56995000 hz, the center of that upper sideband in the 45mhz filter
would be 56995000-11998000 = 44997000 hz
With clk1 for LSB of 32995000 hz, the center of that lower sideband in the 45mhz filter
would be 32995000+11998000 = 44993000 hz.

That's a 4khz difference in where the center of the sideband of interest got placed
within the 45mhz crystal filter passband. 

Assuming the the 45mhz crystal filter is truly centered at 44.995mhz as the notes in the code suggest,
then I figure that values of 56993000 hz and 32997000 hz should be optimal for clk1.

If the 45mhz crystal filter sweeps out to be somewhere else, then add the difference to those two figures.
For example, if the center is found to be at 44.997mhz (so 2khz higher) then the two figures for clk1
should be 56995000 hz and 32999000 hz.


Regarding Tim's writup:

> When you are talking about adjusting the BFO I assume you are talking about adjusting CLK1.

No, we are talking about adjusting the BFO.
When adjusting the BFO, we should move the VFO to get us back to the displayed frequency.
The display shows the frequency of the suppressed carrier, always easy enough to compute:
    For USB, it's   VFO-(clk1-BFO)
    For LSB, it's   VFO+(clk1-BFO) 
where VFO is clk2, and BFO is clk0.

RIT allows a different VFO setting on receive than transmit.
This IF-Shift thing apparently allows a different BFO setting (with attendant VFO correction) on receive than transmit.

So IF-Shift will change what part of the sideband the 12mhz filter will capture (and thus capture a different range of audio frequencies) 
but does not change the frequency of the suppressed carrier that we are tuned to.

CW is a bit different, but easy enough.

Jerry, KE7ER