Ping John Larkin setting raspberry pi pico clockspeed...

[Jan Panteltje]

just found this:
How to Overclock and Underclock a Raspberry Pi Pico
https://www.tomshardware.com/how-to/overclock-raspberry-pi-pico

 

[John Larkin]

On Sun, 19 Mar 2023 14:16:25 GMT, Jan Panteltje <alien@comet.invalid>
wrote:

just found this:
How to Overclock and Underclock a Raspberry Pi Pico
https://www.tomshardware.com/how-to/overclock-raspberry-pi-pico

Thanks for the link.

I doubt we\'ll need to overclock. We do similar instrumentation
functions on single-core ARMs at 100 MHz, and Pico is dual-core 133
MHz. Dual-core is a nice division of labor.

It sounds like we should program in plain bare c. Python is big and
slow.

A few of our planned projects will need tight clock locking, between
boxes or from a 10 MHz source. The cheap xo on the Pi won\'t be good
enough. I think the Pi can pll its cpu clock to a port pin, so we can
have a 10 MHz i/o connector and a VCXO pll and clock the Pi and/or an
FPGA from that.

I was thinking about inventing POE++, a CAT6 cable with ethernet,
power, and a 10 MHz clock on one of the unused pairs. Which still
leaves a pair for something else.

 

[whit3rd]

On Sunday, March 19, 2023 at 10:36:45 AM UTC-7, John Larkin wrote:

I was thinking about inventing POE++, a CAT6 cable with ethernet,
power, and a 10 MHz clock on one of the unused pairs. Which still
leaves a pair for something else.

No need to have any \'unused pairs\', if you run normal gigabit Ethernet
protocols; the 1-2 pairs\' common mode can be 10 MHz while the
3-4 pairs\' common mode can supply POE.

The 10 MHz bit, though, won\'t survive transmission through a normal switch.
Probably the goal is more-stable-clock? Local crystal oscillator isn\'t hard to
wire in, or expensive.

 

[John Larkin]

On Mon, 20 Mar 2023 00:51:41 -0700 (PDT), whit3rd <whit3rd@gmail.com>
wrote:

On Sunday, March 19, 2023 at 10:36:45?AM UTC-7, John Larkin wrote:

I was thinking about inventing POE++, a CAT6 cable with ethernet,
power, and a 10 MHz clock on one of the unused pairs. Which still
leaves a pair for something else.

No need to have any \'unused pairs\', if you run normal gigabit Ethernet
protocols; the 1-2 pairs\' common mode can be 10 MHz while the
3-4 pairs\' common mode can supply POE.

The 10 MHz bit, though, won\'t survive transmission through a normal switch.

Tha last things in a system would have to be a PoE switch followed by
the clock injector. I need to research PoE power injectors and see
exactly what they do.

Probably the goal is more-stable-clock? Local crystal oscillator isn\'t hard to
wire in, or expensive.

If several boxes have to generate, for example, KHz sine waves with
controlled phase, they need to lock to a common clock.

 

[whit3rd]

On Monday, March 20, 2023 at 7:11:44 AM UTC-7, John Larkin wrote:

On Mon, 20 Mar 2023 00:51:41 -0700 (PDT), whit3rd <whi...@gmail.com
wrote:
On Sunday, March 19, 2023 at 10:36:45?AM UTC-7, John Larkin wrote:

I was thinking about inventing POE++, a CAT6 cable with ethernet,
power, and a 10 MHz clock on one of the unused pairs. Which still
leaves a pair for something else.

No need to have any \'unused pairs\', if you run normal gigabit Ethernet
protocols; the 1-2 pairs\' common mode can be 10 MHz while the
3-4 pairs\' common mode can supply POE.

The 10 MHz bit, though, won\'t survive transmission through a normal switch.

Tha last things in a system would have to be a PoE switch followed by
the clock injector. I need to research PoE power injectors and see
exactly what they do.

Probably the goal is more-stable-clock? Local crystal oscillator isn\'t hard to
wire in, or expensive.

If several boxes have to generate, for example, KHz sine waves with
controlled phase, they need to lock to a common clock.

Every Ethernet packet has built-in clock synchronizing transitions; in
the original spec, a burst of sixty-three cycles,-1-0-1-0-1..... which,
from any given switch, means there\'s a common clock being distributed,
not just data. On 10base500 and 10bases200, every talker had its own
clock to distribute.