Birdie topic comes up, but no cookbook resolution as far as I am aware.
Here is what I have found so far.? Feedback and direction encouraged.
This has been an on again /off again background project, not through completion.
I am running a linear regulated power supply, but have also run on battery with no observed difference.
Dummy load attached to be as quiet as possible.
First started out with the stock Nano on the V6 with Reed¡¯s firmware as baseline.
I am classifying the birdies I observed into 2 general categories:
1.?????? Cat 1: Single frequency or small range (<< 1 kHz wide), potential solution with FIRST_IF shift
2.?????? Cat 2: Wide (~3-4 kHz)
Cat 1: All the song birdies I have observed so far can be removed from the tuned frequency of interest by shifting the FIRST_IF.
The 45 MHz roofing filter is pretty wide.?? I set up code to shift the? FIRST_IF up or down by 1 kHz increments.? Looking at 40 m, most Cat 1 birdies can be removed from the frequency of interest by a 1 kHz FIRST_IF shift. I found one that required a 2 kHz shift.? If this holds true across all frequencies, I am thinking about trying to set FIRST_IF to 45004000 and adding a simple IF shift routine/button to toggle between 45004000 and 45006000 as a first order patch (assuming 45005000 is actual center)? Improvements, learning from there.? E.g. if it can be systematically determined, handle in code.? Or hash table, etc.? The step doesn¡¯t need to be so large of an increment, but I didn¡¯t want to just move to the adjacent frequencies. Whatever the final best way, I think this annoyance can be solved.
Cat2:? Wide birdies are grackles. Not a cute little song bird.
I have found two Cat2 sub classes so far.
A.? ? ? 1.?Can be shifted out of the way by shifting the BFO 1 click of 50 Hz.
a.?????? This maybe consistent with comments like ¡°you need to set your BFO properly¡±
b.?????? But changing the BFO just moves the birdy, akin to the Cat 1.
B.? ? ? 2.?No software/si5351 ¡®moving birdy¡¯ solution, issue because of ¡®fixed¡¯ oscillators.? My Nano Oscillator ~16.01 MHz has a very strong even and odd harmonics.?? Si 5351 is ~25 MHz and CH340 is ~12 MHz.
a.?????? 40m: Nano - Si5351: 2*16.01 ¨C 1*25 = 7.02 #Very strong
b.?????? 20m: Nano - Si5351: 4*16.01 -? 2*25 = 14.04 #weaker, but still annoyance
The Nano oscillator can be shifted slightly by changing the operating voltage.? A wide enough shift can be obtained and still run the Nano in voltage spec for 16 MHz operation.? E.g. from 5.5V to 4.5V.? Where the birdie range does not overlap, for one test case.
Over the hunt,? the 5V Nano has been converted to 3.3V (yes I know it¡¯s out of spec at 16 MHz) and the display modified to run at 3.3V (which now is in spec)? If the Nano fails to operate, noted it is out of spec.? Running at a lower voltage, it will have less self-heating plus the 5V regulator on the board is now handling less current and thus generates less heat and raises the ambient less. ?Regulator heat sink is now at 44C at 24C ambient, and the current draw in RX for the whole uBITX is 0.20A at 13.70V input.? Reduced from 0.32A.? I have measured the heat sink temperature in the past, but couldn¡¯t find any notes.? I remember that it was hot enough to burn skin per the thermal couple measurement.
With some additional work and simple circuits, this Nano could be run at 3.3V in normal mode, and changeover to 5.0V to shift the birdies.? Would need to level shift down the Nano outputs as not to overdrive the IL9341. ?This solution is not ideal to shift this birdie class out of the ham bands, but maybe a nice method for a more full range of birdie free RX.
The si5351 is specified for XTAL of 25 MHz to 27 MHz.?? Changing to 27 MHz may be a good choice for those more oriented at 40m,? but that pushes the problem to 15m.? trade offs
What would be the downside of running the si5351 at 24 MHz (besides out of spec for XTAL per datasheet)?
If it can¡¯t operate a XTAL <5% lower than spec, a driven clock can be forced on XA and float XB. It does ¡®work¡¯ for a proof test case, driving a 24 MHz clock to XA.? My breadboarded clock was clean, just not so much so when attaching to Si5351. Just a 5V canned oscillator (no additional driver) to a voltage divider to supply 1V pp through a 100 nF coupling cap, and not so great twisted pair for clock into si5351. (I have mini coax somewhere, so I thought.)? This wasn¡¯t a 100% success, and I have not yet proceeded further.? A better experiment attempt would require sourcing additional supplies.
Also note that the si5351 can also run at a lower core voltage.? Most min frequency circuit limits are lower at lower voltage.
The solution to run the si5351 at 24 MHz seems too obvious at the moment that I must be myopic.?
What am I missing? I tell myself that it must be something, otherwise those working with this since the first Raduino would have made that change.? If it has to do with 24 MHz into the si5351 not being exactly 2X the 12 MHz USB clock and producing a signal that gets into the audio chain, maybe the 24 MHz clock could be divided by 2 and feed into the CH340.? You can see feature creep along this line of thinking¡.? Only 1 Master clock and the rest derived, but that doesn¡¯t fit the simplicity model of the uBITX¡¯s Raduino using an off the shelf Nano. Or any replacement Raduino with the controller on a pre-kitted daughter card. e.g. a Teensy or ST board.
Of course there will still be birdies with 12,16,24 MHz clocks, just not any in the ham bands per my calculations.
I will continue pursue as time and desire allows, but if someone here can point out an error in my thinking, it would be appreciated.
Rgds,
Gary
HA3HZ