Running Pico-microPyton with non-standard XTAL - 12.8MHz vs 12.0MHz

[mungewell]

I have a project which requires an accurate clock, so I am investigating changing the XTAL for a high precision TCXO. Some of the parts which might work are a slightly different frequency of 12.8MHz.

These would appear to be able to give the CPU freq (120MHz) that I want, with different PLL settings.

$ python3 vcocalc.py 120 -i 12.8
Requested: 120.0 MHz
Achieved: 120.0 MHz
REFDIV: 1
FBDIV: 75 (VCO = 960.0 MHz)
PD1: 4
PD2: 2

I modified the first board to 12.8MHz, and the code I previously uploaded is still running... however (as suspected) the device will no longer enumerate on USB

2024-03-04T19:54:37.511766-07:00 the-void kernel: [429719.363398] usb 1-1: new full-speed USB device number 83 using xhci_hcd
2024-03-04T19:54:37.639668-07:00 the-void kernel: [429719.491425] usb 1-1: device descriptor read/64, error -71
2024-03-04T19:54:37.875762-07:00 the-void kernel: [429719.727420] usb 1-1: device descriptor read/64, error -71
2024-03-04T19:54:38.111575-07:00 the-void kernel: [429719.963390] usb 1-1: new full-speed USB device number 84 using xhci_hcd
2024-03-04T19:54:38.239766-07:00 the-void kernel: [429720.091424] usb 1-1: device descriptor read/64, error -71
2024-03-04T19:54:38.475771-07:00 the-void kernel: [429720.327410] usb 1-1: device descriptor read/64, error -71
2024-03-04T19:54:38.583776-07:00 the-void kernel: [429720.435432] usb usb1-port1: attempt power cycle
2024-03-04T19:54:38.999774-07:00 the-void kernel: [429720.851350] usb 1-1: new full-speed USB device number 85 using xhci_hcd
2024-03-04T19:54:38.999821-07:00 the-void kernel: [429720.851620] usb 1-1: Device not responding to setup address.
2024-03-04T19:54:39.207778-07:00 the-void kernel: [429721.059624] usb 1-1: Device not responding to setup address.
2024-03-04T19:54:39.415769-07:00 the-void kernel: [429721.267226] usb 1-1: device not accepting address 85, error -71
2024-03-04T19:54:39.543624-07:00 the-void kernel: [429721.395201] usb 1-1: new full-speed USB device number 86 using xhci_hcd
2024-03-04T19:54:39.543677-07:00 the-void kernel: [429721.395428] usb 1-1: Device not responding to setup address.
2024-03-04T19:54:39.751773-07:00 the-void kernel: [429721.603593] usb 1-1: Device not responding to setup address.
2024-03-04T19:54:39.959776-07:00 the-void kernel: [429721.811330] usb 1-1: device not accepting address 86, error -71
2024-03-04T19:54:39.959834-07:00 the-void kernel: [429721.811551] usb usb1-port1: unable to enumerate USB device

Q. I have an SWD interface, but not really experienced with it. Can I 'poke' the registers from there to reconfigure the board in a temporary manner?

Q. How can I make these changes more permanent? Would that require a custom build of microPython??

[mungewell]

Got a connection with GDB, and can read previous CPU PLL..

$ sudo openocd -f interface/cmsis-dap.cfg -f target/rp2040.cfg -c "init;"

and

$ gdb-multiarch
(gdb) target remote localhost:3333
(gdb)  x 0x40028008
0x40028008:	0x00000078
(gdb)  x 0x4002800C
0x4002800c:	0x00062000

Which corresponds to

$ python3 vcocalc.py 120
Requested: 120.0 MHz
Achieved: 120.0 MHz
REFDIV: 1
FBDIV: 120 (VCO = 1440.0 MHz)
PD1: 6
PD2: 2

So, what happens if I just re-write?

(gdb) set *((unsigned int)0x40028008) = 0x48
(gdb) x 0x40028008
0x40028008:	0x00000048
(gdb) set *((unsigned int)0x4002800C) = 0x00042000
(gdb) x 0x4002800C
0x4002800c:	0x00042000

I don't know if this actually changes anything...

[mungewell]

And these would be the USB PLL....

(gdb) x 0x4002c008
0x4002c008:	0x00000064
(gdb) x 0x4002c00C
0x4002c00c:	0x00055000
(gdb) set *((unsigned int)0x4002c00C) = 0x00074000
(gdb) set *((unsigned int)0x4002c008) = 0x00000069
(gdb) x 0x4002c008
0x4002c008:	0x00000069
(gdb) x 0x4002c00C
0x4002c00c:	0x00074000

$ python3 vcocalc.py 48  -i 12.8
Requested: 48.0 MHz
Achieved: 48.0 MHz
REFDIV: 1
FBDIV: 105 (VCO = 1344.0 MHz)
PD1: 7
PD2: 4

Update: YESSSSS... that works. If I keep the device powered, make the change and re-enable the hub it enumerates and can be seen by Thonny.

2024-03-04T22:18:55.311783-07:00 the-void kernel: [438377.118652] usb 1-1.3: new full-speed USB device number 47 using xhci_hcd
2024-03-04T22:18:55.311833-07:00 the-void kernel: [438377.118888] usb 1-1.3: Device not responding to setup address.
2024-03-04T22:18:55.519778-07:00 the-void kernel: [438377.326824] usb 1-1.3: Device not responding to setup address.
2024-03-04T22:18:55.727607-07:00 the-void kernel: [438377.534579] usb 1-1.3: device not accepting address 47, error -71
2024-03-04T22:18:55.807672-07:00 the-void kernel: [438377.614397] usb 1-1.3: new full-speed USB device number 48 using xhci_hcd
2024-03-04T22:18:55.807717-07:00 the-void kernel: [438377.614558] usb 1-1.3: Device not responding to setup address.
2024-03-04T22:18:56.015591-07:00 the-void kernel: [438377.822451] usb 1-1.3: Device not responding to setup address.
2024-03-04T22:18:56.223744-07:00 the-void kernel: [438378.030300] usb 1-1.3: device not accepting address 48, error -71
2024-03-04T22:18:56.223790-07:00 the-void kernel: [438378.030530] usb 1-1-port3: unable to enumerate USB device


2024-03-04T22:20:44.739768-07:00 the-void kernel: [438486.531938] usb 1-1.3: new full-speed USB device number 49 using xhci_hcd
2024-03-04T22:20:44.863624-07:00 the-void kernel: [438486.659448] usb 1-1.3: New USB device found, idVendor=2e8a, idProduct=0005, bcdDevice= 1.00
2024-03-04T22:20:44.863686-07:00 the-void kernel: [438486.659475] usb 1-1.3: New USB device strings: Mfr=1, Product=2, SerialNumber=3
2024-03-04T22:20:44.863698-07:00 the-void kernel: [438486.659488] usb 1-1.3: Product: Board in FS mode
2024-03-04T22:20:44.863708-07:00 the-void kernel: [438486.659499] usb 1-1.3: Manufacturer: MicroPython
2024-03-04T22:20:44.863718-07:00 the-void kernel: [438486.659508] usb 1-1.3: SerialNumber: e66034f71f714a36

[mungewell]

My code outputs a 1pps output, it changes but not to the correct value... don't know what I am doing wrong :-(

These settings are 'better' match (12.0 running similar to 12.8), but still not good.

$ python3 vcocalc.py 120 -i 12.8 --vco-min 1000
Requested: 120.0 MHz
Achieved: 120.0 MHz
REFDIV: 2
FBDIV: 225 (VCO = 1440.0 MHz)
PD1: 6
PD2: 2

$ python3 vcocalc.py 120 -i 12 --vco-min 1000
Requested: 120.0 MHz
Achieved: 120.0 MHz
REFDIV: 1
FBDIV: 120 (VCO = 1440.0 MHz)
PD1: 6
PD2: 2

Before change 1pps is 0.93749ms (which makes sense), but after it is 1.04166s (which doesn't).

Update:
This is me manually adjusting PLL...

0x10017e38 in ?? ()
(gdb) x 0x40028000
0x40028000:	0x80000001
(gdb)  set *((unsigned int)0x40028000) = 0x0000011
(gdb)  set *((unsigned int)0x40028000) = 0x0000012
(gdb)  set *((unsigned int)0x40028008) = 0x00000e1
(gdb) x 0x40028000
0x40028000:	0x80000012
(gdb)  set *((unsigned int)0x40028000) = 0x0000002
(gdb) x 0x40028000
0x40028000:	0x80000002
(gdb) x 0x40028004
0x40028004:	0x00000004
(gdb) x 0x40028008
0x40028008:	0x000000e1
(gdb) x 0x4002800c
0x4002800c:	0x00062000
(gdb) c

[mungewell]

Still don't know why the end result is not clocking at the correct speed, but I was able to script the process to change the clocks - and thus be able to connect the microPython session. May be helpful to someone....
https://github.com/mungewell/pico-timecode/blob/main/test_scripts/clocks/fix_clocks_for_12-8MHz.sh

Unfortunately this does not appear to work when trying to 'fix' the boot-loader, which I think is my next part of the quest - to replace the uf2 with something different.

[mungewell]

Found the last part of the puzzle, I couldn't re-clock the boot loader, but then realized that I can flash a recompiled '12.8MHz' version of microPython via the SWD interface. D'oh.

ports/rp2/build-RPI_PICO$ sudo openocd -f interface/cmsis-dap.cfg -f target/rp2040.cfg -c "program firmware.elf verify reset exit"

And now my 1pps timing is correct too. Closing ticket.