TheBackShed.com

Forum Index : Microcontroller and PC projects : STEP-DIR decoder for picomite PIO

Page 4 of 4

Author

Message

Mixtel90

Guru

Joined: 05/10/2019
Location: United Kingdom
Posts: 9021

Posted: 03:02pm 15 Jul 2026

Copy link to clipboard

Print this post

At what point does it become cheaper to use a bank of Picomites instead of a single Propeller P2 and get *true* asynchronous multitasking?

Mick

Zilog Inside! nascom.info for Nascom & Gemini
Preliminary MMBasic docs & my PCB designs

PhenixRising
Guru

Joined: 07/11/2023
Location: United Kingdom
Posts: 2049

Posted: 05:24pm 15 Jul 2026

Copy link to clipboard

Print this post

Mixtel90 said At what point does it become cheaper to use a bank of Picomites instead of a single Propeller P2 and get *true* asynchronous multitasking?

For me....different animals;

The ultimate MCU is the MMBasic family but the Prop is like an FPGA that mere mortals can program as easily as a mite.

Apart from the eight CPUs, the 64 pins are processors in their own right.

They can be 16bit DACs or 16bit ADCs. One of the eight CPUs can handle 16 high speed UARTs simultaneously.

I prefer writing BASIC code but if I need a C function, I just make a class out of it and include it in my BASIC source and the same thing goes for a SPIN language function. Inline asm, etc.

No interrupts required and so no jitter/latency. You can't get better real time.

Mite + Prop is a killer combo that even grunts like me can do amazing things with.

I read the other day that the P2 is something like 14 Euros. I have the "KISS" module which is very similar to Pete's RP2350 DIL. Looks like I can get them produced by JLCPCB because LCSC now has the P2 chip.

PhenixRising
Guru

Joined: 07/11/2023
Location: United Kingdom
Posts: 2049

Posted: 02:03pm 27 Jul 2026

Copy link to clipboard

Print this post

Perfect slope on the scope for the 16-bit DAC

Check-out the spec:

Quote

The classic problem:
A naive single-PWM 16-bit DAC needs its full period to be 2^16 = 65536 clock counts long, so it can represent every possible duty-cycle step. At your 300MHz clock, that's:

300,000,000 / 65536 ≈ 4.58 kHz update rate for genuine 16-bit resolution. That's the classic tradeoff — more bits of resolution means a slower base frequency, since you need that many clock ticks to count through all the possible duty-cycle values.

How the P2's hybrid mode avoids this:
Instead of PWM-ing all 16 bits, it splits the word:
Upper 8 bits → fed straight into the pin's actual onboard resistor-ladder DAC hardware. This is instantaneous — no averaging, no duty cycle, no waiting. It just sets an analog voltage directly, updated every clock if you want.
Lower 8 bits → dithered/PWM'd, but only across 256 clocks (2^8, not 2^16), since that's all the resolution those bits need.

So the "duty cycle" mechanism only has to represent 256 states, not 65536. That's why the dither carrier lands at:

300,000,000 / 256 ≈ 1.172 MHz

Roughly 256× faster than the naive full-16-bit approach would allow, for the same 16 bits of total resolution. The real hardware DAC on the upper bits is what buys you that speedup.

Real-world bandwidth after your LPF:
The 1.17MHz dither carrier is what your low-pass filter needs to reject. Since you're building a ±10V control-type output (not RF), you'll likely want your filter's passband well below that — a decent 2-pole active filter with a corner in the low kHz to tens-of-kHz range will reject that carrier comfortably while passing whatever signal bandwidth you actually need, giving you a lot of headroom.

Volhout
Guru

Joined: 05/03/2018
Location: Netherlands
Posts: 6022

Posted: 02:11pm 27 Jul 2026

Copy link to clipboard

Print this post

Mixtel90 said At what point does it become cheaper to use a bank of Picomites instead of a single Propeller P2 and get *true* asynchronous multitasking?

No, each pico can do what a prpeller core can, but the intercore communication in the P2 is magnitudes faster, until the moment Peter invents memmory sharing between 8 pico's

Volhout

P.s. a P2 at 14 euro may in the end be cheaper than 8 pico,s
Edited 2026-07-28 00:13 by Volhout

PicomiteVGA PETSCII ROBOTS

Mixtel90

Guru

Joined: 05/10/2019
Location: United Kingdom
Posts: 9021

Posted: 02:27pm 27 Jul 2026

Copy link to clipboard

Print this post

It does depend on whether that's what's needed for a particular task though. A simple signal to start the next process, like passing a token, is easily done on a Pico using GPIO pins or, if you want something more complex, multi-drop COM because it's buffered. A Prop appears to run as a round-robin, with each processor running in turn (or at least that's what it looks like to me). Ok, the loop is very fast, but it's still synchronous. The Picos are (or can be) truly asynchronous.

PS
The only P2 I can find as a bare chip is about £22 and is only available to order. You can get the modules but they cost *a lot* more - the minimal Edge module is $79.
Edited 2026-07-28 00:34 by Mixtel90

Mick

Zilog Inside! nascom.info for Nascom & Gemini
Preliminary MMBasic docs & my PCB designs

PhenixRising
Guru

Joined: 07/11/2023
Location: United Kingdom
Posts: 2049

Posted: 02:55pm 27 Jul 2026

Copy link to clipboard

Print this post

Mixtel90 said A Prop appears to run as a round-robin, with each processor running in turn (or at least that's what it looks like to me).

All cogs run concurrently and have shared resources.

Quote
PS
The only P2 I can find as a bare chip is about £22 and is only available to order. You can get the modules but they cost *a lot* more - the minimal Edge module is $79.

Depends on the quantity:

Mixtel90

Guru

Joined: 05/10/2019
Location: United Kingdom
Posts: 9021

Posted: 03:45pm 27 Jul 2026

Copy link to clipboard

Print this post

That makes sense. The RP2350 is barely more than $1 each in any reasonable quantity, I believe. Even down as low as 80c in large quantities.

Mick

Zilog Inside! nascom.info for Nascom & Gemini
Preliminary MMBasic docs & my PCB designs

Page 4 of 4

Print this page

To reply to this topic, you need to log in.