Make oversampling work on LM20A

[hannesb]

Because the BURST control has moved to a separate register with nRF54LM20A, oversampling currently doesn't work with this chip. The patch fixed that.

Arduino sketch to test changes (tested with XIAO nRF54LM20A and XIAO nRF54L15):

#if !(defined(ARDUINO_NRF54LM20A) || defined(ARDUINO_NRF54LM20B) || defined(ARDUINO_NRF54L15))
#error "nRF54LM20A board required"
#endif

#include "nrf54l15_hal.h"

using namespace xiao_nrf54l15;

constexpr uint32_t kSaadcSpinLimit = 400000UL;

Saadc g_adc(nrf54l15::SAADC_BASE);

void setup() {
  Serial.begin(115200);
  delay(300);
  Serial.println(F("#### SAADC Oversampling Test ####"));
}

void loop() {
  // With oversampling
  int32_t u1 = 0, u2 = 0, u3 = 0;
  if (!g_adc.begin(AdcResolution::k14bit, AdcOversample::k256x, kSaadcSpinLimit)) {
    Serial.println("g_adc.begin() failed");
    while(1);
  }  
  g_adc.configureSingleEnded(0U, kPinD0, AdcGain::k2over8, 159, 4, true);
  g_adc.sampleMilliVolts(&u1, kSaadcSpinLimit);
  g_adc.configureSingleEnded(0U, AdcInternalInput::kAvdd, AdcGain::k2over8, 159, 4, true);
  g_adc.sampleMilliVolts(&u2, kSaadcSpinLimit);
  g_adc.configureDifferential(0U, kPinD0, kPinD1, AdcGain::k2over8, 159, 4, true);
  g_adc.sampleMilliVolts(&u3, kSaadcSpinLimit);

  // Without oversampling
  int32_t u4 = 0, u5 = 0, u6 = 0;
  if (!g_adc.begin(AdcResolution::k14bit, AdcOversample::kBypass, kSaadcSpinLimit)) {
    Serial.println("g_adc.begin() failed");
    while(1);
  }  
  g_adc.configureSingleEnded(0U, kPinD0, AdcGain::k2over8, 159, 4, false);
  g_adc.sampleMilliVolts(&u4, kSaadcSpinLimit);
  g_adc.configureSingleEnded(0U, AdcInternalInput::kAvdd, AdcGain::k2over8, 159, 4, false);
  g_adc.sampleMilliVolts(&u5, kSaadcSpinLimit);
  g_adc.configureDifferential(0U, kPinD0, kPinD1, AdcGain::k2over8, 159, 4, false);
  g_adc.sampleMilliVolts(&u6, kSaadcSpinLimit);

  Serial.printf("%ld,%ld,%ld,%ld,%ld,%ld\n", u1, u2, u3, u4, u5, u6);

  delay(100);
}

With Arduino serial plotter, one can easily see the less noisy signal with oversampling 😸

[lolren]

Thanks for finding this and sharing the test sketch. I've included your contribution in 1.0.21, with the burst setting applied globally on LM20A and additional coverage for channels 1 and 7.

The connected LM20A passed three complete 36-case test sweeps. Could you give 1.0.21 another run on your L15 and LM20A when you have a chance?

Release and notes

[hannesb]

The connected LM20A passed three complete 36-case test sweeps. Could you give 1.0.21 another run on your L15 and LM20A when you have a chance?

I tested with 1.0.22, and oversampling works fine now on both XIAO nRF54L15 and nRF54LM20A.

Thx for your effort!
Hannes