3 Wire SPI clock needs to be under 550 KHX

[Nerbing]

Cannot get over 550 KHZ using same setup that allows 8 MHZ on Nano Every and 10 MHZ on ESP32. RP Pico 2, TLE5012 and 2 feet of cable between devices, Usng 3 Wire SPI TX MOSi and RX MISO tied with a 10K resistor and data connected to RX MISO .

Using Philhower core and this sketch Have tried both PGM controlling CS and SPI controlling it and that did not matter. Here is the sample I am trying to get it to work.

#include <SPI.h>

// TLE5012 commands
#define READ_ANGLE_CMD 0x8021

void setup() {
Serial.begin(115200);
while (!Serial) delay(10);
Serial.println("TLE5012 Test");

SPI.setRX(16);
SPI.setCS(17);
SPI.setSCK(18);
SPI.setTX(19);
SPI.begin(true);
delay(5000);

// Configure SPI

//pinMode(CS_PIN, OUTPUT);
//digitalWrite(CS_PIN, HIGH);
}

void loop() {
uint16_t angle = readAngle();
float degrees = (angle * 360.0) / 32768.0;

Serial.print("Raw: ");
Serial.print(angle);
Serial.print(" Angle: ");
Serial.println(degrees);

delay(100);
}

uint16_t readAngle() {
uint16_t data = 0;

//digitalWrite(CS_PIN, LOW);
SPI.beginTransaction(SPISettings(400000, MSBFIRST, SPI_MODE1));

SPI.transfer16(READ_ANGLE_CMD);
data = SPI.transfer16(0x0000);

SPI.endTransaction();
// digitalWrite(CS_PIN, HIGH);

return data & 0x7FFF; // Mask status bits
}

Fastest I could get was 550 KHZ so set it to 400 KHZ as that seems stable. But wonder why it will not go any faster, All three were setup the exact same way so no wiring length differences .

[Nerbing]

If it helps tried it powered by the board power supply, powers by the computer usb port and taken outside into the back yard and powered by battery. Here is the board designed for the RP Pico and the little sensor board designed for it. They are connected by just under 2 feet of 5 conductor cable. SPI is upper right just in front of the GPIOs so very short direct run to the JST. All 3 board were designed similarly using the same power suppy and 10K resistor. Suspect it is something I have coded wrong but not as familiar with the PICO but was hoping to use it as they are easier for me and others to buy.

[earlephilhower]

If you have a logic analyzer, I'd recommend looking at the TLE protoboard's SPI pins (instead of the Pico's). We easily do 10MHz for SDCard SPI over really janky wires. I've done (IIRC) ~25MHZ auto SPI-refreshing on an IL-based 3.5" display using "dupoont" jumpers (which were magnetic so probably iron-core(!) and not copper...).

If you could record the same sequence on a working board and compare, that would be very instructive, too.

The other thing, you call out a 5V buck converter. The Pico 2 is 3.3V CMOS so maybe there's some issue with levels at your receiver if the TLE chip is 5V. Some kind of o-scope might be needed there, though, to see if the Vout of the Pico2 meets the high/low thresholds of your device once it goes through the wiring...

[Nerbing]

Thank you for the reply. The 5v buck is only tied to VBUS so it does not affect anything else. Did that after I gave my handwired device to someone who powered it with a 12v wall wart even though i told them it needed the 5v one I supplied. Apparently they looked the same. That way it will be harder to kill.

Was going to try loading an arduino mbed core version on it but cannot seem to get around whatever kills the serial when I attempt to use it . Tried both on the windows 11 machine and Ubuntu and both did the same with with the one pictured and a new one that had never been used. Gave up on that for now maybe will try again when I figure what is wrong. Wanted to see if that made a difference .

But am suspecting it is something about the device. Guess I need to swing by my former workplace so i can access an oscilloscope.

[earlephilhower]

If you lose USB Serial port the sketch has probably crashed. If you have a PicoProbe/DebugProbe you can hook it up and find out exactly where. May not always be where the problem is, but it often is useful.

A logic analyzer (even those cheap 24MHz ones on AliExpress) can grab SPI up to a few MHZ w/o trouble and let you compare the waveforms of both ESP32 and Pico and see if something is blatantly different at the protocol level (i.e. there's 1ms extra between CS asserts, etc.).

[Nerbing]

Suspect it is the device. Borrowed a Nano RP2040 Connect and tested it. I can reliably reach 2Mhz with it sitting on a breadboard. Just to make sure wired the RP Pico into the breadboard and could not exceed 200Khz before the readings became erratic. 300 Kzh nothing it stopped working. Going to drop this idea and go back to the ESP32. Thanks for the help, I'll use this for the solar pool heater idea I've been playing with. Do not need SPI for that and the board will work for that too.

[Nerbing]

I have been experimenting This is a RP2350 issue. An older RP Pico with a RP2040 worked perfectly at 1Mhz. Tried to order a couple of the newer A4 variants RP Pico 2 to replace the A2 ones but they sent more of the buggy ones. I can partially solve it by tying MOSI to MISO with a smaller 3.3K ohm in place of the 10K resistor and tying MOSI to gnd with a 4.7K olm with data on the device connected to MISO. I can reliably raise the speed to 400Khz by doing that. It starts to corrupt at 500Khz and drops out by 600 Khz. A shame I designed a set of PCB's for an angle reader and this made them useless with the newer boards.