SPI no longer works after update from v0.30 -> v0.37
#79
Comments
|
I put the SPI under an oscilloscope, and found that the two versions produce very different signals. Each transaction with the TMC5160 should consist of 8 + 32 bits. Where the first 8 corresponds to a register at the slave device and a R/W bit, and the 32 remaining bits are data. For each transaction, the slave's behaviour is to send back 8 + 32 bits, where the first 8 is a status byte, while the remaining 32 is an echo of the data it received in the previous transaction. In all of the pictures (sorry for the picture quality) below, we have:
Under While Zoomed in view of the first and second exchange, respectively: A minimal sample to reproduce for fn main() -> Result<()> {
esp_idf_sys::link_patches();
// Bind the log crate to the ESP Logging facilities
esp_idf_svc::log::EspLogger::initialize_default();
let peripherals = Peripherals::take().unwrap();
let pins = peripherals.pins;
// SPI pins
let sclk = pins.gpio1.into_output()?;
let sdo = pins.gpio2.into_input()?;
let sdi = Some(pins.gpio42.into_output()?);
let cs = Option::<gpio::Gpio2<gpio::Output>>::None;
// Note: CS pins have been toggled manually, since I needed > 1 (which esp-idf-hal had (has?) no support for)
let mut cs1 = pins.gpio48.into_output()?.degrade();
let spi_pins = spi::Pins { sclk, sdo, sdi, cs };
let mut spi = Master::<SPI2, _, _, _, _>::new(
peripherals.spi2,
spi_pins,
esp_idf_hal::spi::config::Config {
baudrate: Hertz(4_000_000),
data_mode: V02Type(embedded_hal::spi::MODE_3).into(),
// <- add BitOrder here if v0.30
},
)?;
loop {
cs1.set_low()?;
// This just happens to be the message that is sent in the pictures above
spi.transfer(&mut [
33 | 0b1000_0000u8,
0b1011_0011,
0b1000_1111,
0b0000_1110,
0b0011_0111,
]);
cs1.set_high()?;
}
} |
|
As a data point, I had to roll back to here to get the display on my Adafruit Magtag to work again |
|
@Dominaezzz Any ideas? |
|
From staring at the code I can't tell what's wrong. @daagaak Does I didn't touch any of the e-hal 0.2 code paths yet at this point so I'm puzzled as to how it suddenly stopped working then. @akselbor Any chance you could do a binary search to see which exact version introduces the behaviour? |
|
@Dominaezzz However, if I allow it to select |
|
@daagaak Which device are you running that on? ESP32, esp32-s3, etc. |
|
@Dominaezzz That's an on ESP32-S2. I just tried the same source on ESP32 with |
|
Yeah I figured that might be the case. This code snippet to de-assert CS doesn't work on the ESP32-S2. Lines 314 to 333 in 2c3741a Not sure how to work around it besides just using software CS. |
|
I discovered the similar issue. Let me add few notes. The following configuration does not work on ESP32-S3, the display shows just random noise: Working configuration for ESP32-S3: How to simulate with ESP32-S3-USB-OTG: |
|
Does #86 fix the issue for you? |
|
Yep, that branch is working on my ESP32-S2. |
I've been using an ESP32-S3 to control a couple of TMC5160s over SPI using
esp-idf-hal v0.30. Yesterday, I updated the project toesp-idf-hal v0.37and found that the SPI communication no longer works. The code is unchanged betweenv0.30andv0.37, except for the necessary changes tospi::config::Configduring initialization.Any idea what might be causing this? I would guess that it is related to 2d0554d? I'll try to debug the issue by myself in the meantime.
The relevant section of code looks roughly like this, after the abstractions have been removed:
Versions used:
The text was updated successfully, but these errors were encountered: