Add support for Waveshare RP2350 Can board

[NoRolex]

Hello!
Im quite new to the RP family and wanted to make an OBD2 diagnostic tool for cars but haven't had any succes flashing anythig to this board.
I tried all the waveshare as well the generic rp2350 and the rasbperry pico 2 options but none of them seemed to work. It uploaded but nothing happened on the board and it went right back to boot device mode. The flash_nuke.uf2 seemingly worked fine so i dont think that it is a faulty board. (it is the only thing that worked)
Thank you guys in advance!

[earlephilhower]

All RP2040 and RP2350 boards are really the same, with the CPU and flash plus "some other bits". That board looks like a barebones one with a couple support chips hardwired to some GP pins.

I suggest starting from basics. Set the Generic RP2350, Select 4MB flash size, Chip Variant 2350A, and then run a simple LED blinker. Assuming the documentation on the WS site is correct, it should blink the onboard led

void setup() {
  pinMode(25, OUTPUT);
}
void loop() {
  digitalWrite(25, HIGH);
  delay(1000);
  digitalWrite(25, LOW);
  delay(1000);
}

If that doesn't blink the LED then I suggest seeking out a schematic for that board (I can't find one on their site) to double-check the pinout.

If that does blink then you're fine but you might need to configure your computer if you're running Windows to actually see the serial ports it delivers. The docs here have lots of info on running Zadig/etc. as appropriate.

[Lannakin]

I'll take a look at the demo stuff WS provides for this board real quick and see what might be up, and I can post stuff to test, if you'd like.

[Lannakin]

Schematic is here; fortunately I had to learn how they structure their site. Or unfortunately? Anyway...

I'd agree that the generic RP2350 should at least work, but while seeing about whipping up a draft for a new variant, I noticed that their decision to have XL2515_INT_PIN on GPIO8 means you don't get to use SPI1 for anything but the CAN controller, unless you can have XL2515_INT_PIN (8u) at the same time as #define PIN_SPI1_MISO (8u). Idk, I had thought that it wasn't possible to do that since every other variant seemed to use different pin assignments to avoid overlaps.

Looking at waveshare_rp2350_lcd_0_96, though, there's :

#define LDC_SPI         (1u)
#define PIN_LCD_DC      (8u)
#define PIN_LCD_CS      (9u)
...
#define PIN_LCD_RST     (12u)

and then:

#define PIN_SERIAL1_RX  (1u)

#define PIN_SERIAL2_TX  (8u)
#define PIN_SERIAL2_RX  (9u)
...
#define PIN_SPI1_MISO   (12u)

[earlephilhower]

The one thing that sticks out to me is they really, really cheaped on the flash there. They're using a Puya P25Q32SH chip. Check the ESP8266 Arduino repo where we had nothing but horrible compatibility issues with Puya flash due to their "not quite like the others" behavior. It was cheaper for a good reason in our experience.

Before looking into the CAN stuff, basic functionality needs to be verified. Can you run the simple blink sketch? LED's on GP25 so it should run as-is and blink the LED.

If it doesn't, then that Puya flash is probably at issue. Either the timing doesn't work with the Pico2's boot ROM setup and a special shim needs to be added before anything at all runs from flash, or it's not even programming properly by the ROM in UF2 mode. To verify the program was actually written properly by the boot ROM you'd need to upload the blink sketch then use a DebugProbe/PicoProbe to download flash memory to a file and check against the .BIN. If those don't match, you have a serious problem and I'd say the board is not fit for purpose. If they do match, then we'd need to see if anyone else has a "stage 2" boot.S that can be shoved in somehow. We needed those on the original Pico, but on the Pico2 there's no real ROM support for them (nor is there a need in general).

If blink does work, then we can see about a setup.

Re: the CAN chip, it looks OK and yes I think only SPI1 can be used. There's no overlap, because you don't need to use 1 block of pins for the SPI1 pins, you can pick any pin you like that's marked SPI1RX (GP8 or GP12).

[Lannakin]

What if NoRolex never comes back?* I mean, I could probably manage to justify buying the board, and if my roommate expresses concern, I can ask how much his last set of key caps cost, but it'd suck if the Puya flash is super uncooperative.

If I do order it, I'm willing to do tests etc if you're willing to endure the questions of someone bumbling through C and C++ like a drunk cat on stilts.

*note: read this in a jokingly scared voice

[Lannakin]

I looked up Puya I am impressed.

[earlephilhower]

Yeah, we had real issues with any filesystem or flash writes. I don't remember any specific esptool upload problems (but no guarantee the esptool block write is the same as the RP2 boot rom write), but any filesystem/eeprom stuff was always an issue. Partial writes didn't always work and for SPIFFS/LittleFS I think we needed 256byte writes. IIRC we had to fake things and do some sort of caching and read-modify-write to get it working most of the time. We do 256 byte writes here (SDK minimum) too.

Not sure if this chip would have the same issue, but some LittleFS data integrity tests would probably catch it easy enough.

[NoRolex]

Hello again!
Im sorry for the very late reply. I tried basically everything that was humanly possible and i started to get crazy so i got to the hardware side of the story and i found out that i had a malfuntioning boot or reset button. Or around somewhere at least. Tried cleaning it as softly as i possibly can (with proper tools) and finally got to the point where it blinked for the first time in months! My luck isn't the best so i tought id leave it for a couple of minutes so i can verify that it is truly working and almost burned down my desk because when i came back the Windows went crazy popping messages with "overcurrent on USB port". The board literally cooked itself alive doing nothing and the 3.3V rail got dead short. The RP2350 got damaged in the process and it is officially dead. I rather spend a couple bucks more on the Adafruit RP2040 CAN board next time.
Than you guys for your time.

Ps.: it was working with the generic RP2350 option.
Edit: spelling.

[earlephilhower]

Crap,sorry to hear that but thanks for updating!