nRF52840 and SX1262 BUSY signal

[RobertDeHate]

I have a LilyGo T-Echo that I am getting setup.
Because I have 3 SPI devices(Display/Radio/SDcard) and the nRF52840 has 4 SPI ports each device has it's own port.
I have stripped down the code to just this testing.ino file for Arduino IDE.
I made two functions displayGo(); radioGo();
If I run displayGo first the radioGo function hangs with:

Start
dispPort->begin done
display.init done
rfPort->begin done

If I run radioGo first then displayGo it completes with:

Start
rfPort->begin done
radio.begin done
dispPort->begin done
display.init done
Initialized

When I asked my ChatGPT buddy about this his response was interesting:
The LilyGO T-Echo:
Uses an SX1262 reference design variant
Has BUSY active-HIGH when idle
Has power gating and level translation
Violates RadioLib’s assumption that:
BUSY HIGH = chip busy
So RadioLib waits forever.

Then it proceeded to ask me to locate the section of code that handles the BUSY signal and invert it in code.

As we all know, the internet is always right, I was looking for more advice on this.
On the circuit board the module is labeled LoRa_89, SX1262 TCXO, V1.4
Ideas?

#include <GxEPD2_BW.h> //E-Ink library
#include <Fonts/FreeMonoBold12pt7b.h>
#include <GxEPD.h>
#include <GxIO/GxIO_SPI/GxIO_SPI.h>
#include <RadioLib.h> //latest V7.4.0
#include "utilities.h"

#define sdCard_Miso _PINNUM(0,8)
#define sdCard_Sclk _PINNUM(0,6)
#define sdCard_Mosi _PINNUM(0,15)
#define sdCard_CS _PINNUM(1,7)

SPIClass        *dispPort  = nullptr;
GxIO_Class      *io        = nullptr;
SPIClass        *rfPort    = nullptr;

// For Waveshare 1.54" E-Ink display:
GxEPD2_BW<GxEPD2_154_D67, GxEPD2_154_D67::HEIGHT> display(GxEPD2_154_D67(/*CS=*/ ePaper_Cs, /*DC=*/ ePaper_Dc, /*RST=*/ ePaper_Rst, /*BUSY=*/ ePaper_Busy));

void setup() {
  digitalWrite(Power_Enable_Pin, HIGH); pinMode(Power_Enable_Pin, OUTPUT);
  //digitalWrite(sdCard_CS, HIGH); pinMode(sdCard_CS, OUTPUT);    // idle HIGH (don’t hold it LOW)
  digitalWrite(LoRa_Cs, HIGH); pinMode(LoRa_Cs, OUTPUT); 
  digitalWrite(ePaper_Cs, HIGH); pinMode(ePaper_Cs, OUTPUT);
  digitalWrite(LoRa_Rst, HIGH); pinMode(LoRa_Rst, OUTPUT);

  SerialMon.begin(115200);
    while (!SerialMon);
  SerialMon.println("Start\n");
  delay(500);

  displayGo();
  radioGo();

  SerialMon.println("Initialized");
  delay(100);

}
void displayGo( void){

  //Display begin
  dispPort = new SPIClass(NRF_SPIM1,ePaper_Miso,ePaper_Sclk,ePaper_Mosi);
  io = new GxIO_Class(*dispPort,ePaper_Cs,ePaper_Dc,ePaper_Rst);
  dispPort->begin();
  SerialMon.println("dispPort->begin done");

  //Display
  display.epd2.selectSPI(*dispPort, SPISettings(4000000, MSBFIRST, SPI_MODE0));
  
  //Start display
  display.init(0, true, 2, false);
  SerialMon.println("display.init done");
}

void radioGo(void){
  //Radio begin
  rfPort = new SPIClass(NRF_SPIM2,LoRa_Miso,LoRa_Sclk,LoRa_Mosi);
  rfPort->begin();
  SerialMon.println("rfPort->begin done");
  //Radio
  SPISettings spiSettings;
  SX1262 radio = new Module(LoRa_Cs, LoRa_Dio1, LoRa_Rst, LoRa_Busy, *rfPort, RADIOLIB_DEFAULT_SPI_SETTINGS);

  //Start Radio
    // manual reset (keep this)
  digitalWrite(LoRa_Rst, LOW);  delay(10);
  digitalWrite(LoRa_Rst, HIGH); delay(10);
  int radioState = radio.begin();
  if (radioState != ERROR_NONE) {
      SerialMon.print(("failed, code "));
      SerialMon.println(radioState);
  }
  else SerialMon.println("radio.begin done");
}

void loop() {
}

[jgromes]

When I asked my ChatGPT buddy about this his response was interesting:

This is not an interesting response - this is an incorrect response (at multiple points, in fact). As can be verified with a quick google search, which I invite you to perform. And even if there was an inverted logic of the busy pin, why would changing the order of initialization fix it?

RadioLib has built-in tools that will help you debug this. And it can be made to work with this device, as evident by T-Echo being used in Meshtastic, which (surprise!) uses RadioLib. So instead of asking the confident hallucination machine for advice, I'm afraid you will have to do the actual work ;)