nRF54L15 SAADC Internal Reference & Gain

[pcorbett]

I see from the SAADC page of the datasheet that there is an internal 0.9V reference on the nRF54L15, but I don't how this reference is generated. Is it an active regulated reference (bandgap derived)? Or is it just a resistor ladder from VDD?

I am trying to sample a signal within the range of 0-1.8V and I am using a 1.8V VDD. I have activated the 0.5 gain block prior to sampling and using the internal reference. How susceptible is the gain block and internal reference to power supply noise? I don't see a PSRR figure listed.

We are seeing some impulse noise on a signal only after it is sampled (not visible on a scope at the ADC pin). The signal is already substantially lowpass filtered and is well below the Nyquist frequency. 

Thanks!

[hmolesworth]

The internal reference is indeed bandgap, though everything is susceptible to power supply noise. Any SAADC single-ended measurement will give poor results; better to use SAADC differential preferably with the negative input physically connected to the source low end (typically GND) such that no current can flow along that negative input trace. Quite good results can be obtained by simply alternating measurements and reversing the differential pin selection on the SAADC (no hardware component change) to compensate for the residual SAADC offset and to some extent other internal noise and offsets. I posted some examples on the nRF52833 as a guide; this will also help on the nRF54L15. Usually a longer sampling window helps, maybe 15uSec or more depending on source impedance. Small capacitance between negative and positive input pins also helps. These note apply for single-ended sensors/voltages in addition to true differential output sensors.

seeking-more-information-on-adc-reference

Look at these results as an example; ENOB 13.1 bits

// Internal 0.6v Ref 14-bit S0S1 4-Wire RTD actual value 905.0 Ohm (reference 2001.4 Ohm) - Forward current
// VDD mV Iref uA RdsOnA RdsOnB V1 mV V2 mV V3 mV V4 mV    Vrtd    Vref  Rrtd  Error
// ====== ======= ====== ====== ===== ===== ===== =====  ======  ======  ===== ======
//   2976     926    168    129  2820  1978  1979   120  2858.8  6328.3  904.1 -0.10%
//   2978     926    168    130  2822  1979  1978   121  2859.4  6329.0  904.2 -0.09%
//   2977     927    169    129  2820  1976  1979   120  2859.1  6330.4  903.9 -0.12%
//   2979     927    170    128  2821  1980  1980   119  2859.6  6331.4  904.0 -0.12%
//   2978     926    170    130  2820  1979  1978   121  2858.3  6328.3  904.0 -0.11%
//   2979     926    167    130  2824  1978  1978   121  2858.4  6329.8  903.8 -0.13%
//   2978     926    170    129  2820  1978  1978   120  2859.5  6328.0  904.4 -0.07%
//   2977     926    169    129  2820  1978  1978   120  2857.8  6329.1  903.7 -0.15%
//
// Internal 0.6v Ref 14-bit S0S1 4-Wire RTD actual value 905.0 Ohm (reference 2001.4 Ohm) - Reverse current
// VDD mV Iref uA RdsOnA RdsOnB V1 mV V2 mV V3 mV V4 mV    Vrtd    Vref   Rrtd  Error
// ====== ======= ====== ====== ===== ===== ===== =====  ======  ======  ===== ======
//   2977     926    133    169   124   962   962  2820  2863.5  6327.8  905.7  0.08%
//   2977     926    132    172   123   963   964  2817  2864.4  6324.5  906.4  0.16%
//   2974     926    132    168   123   962   963  2818  2864.0  6324.3  906.4  0.15%
//   2976     926    132    170   123   961   963  2818  2863.3  6323.8  906.2  0.13%
//   2976     926    132    170   123   961   962  2818  2863.6  6323.6  906.3  0.15%
//   2974     925    132    168   123   961   962  2818  2863.8  6321.6  906.7  0.18%
//   2977     926    131    171   122   962   962  2818  2863.9  6325.8  906.1  0.12%
//   2976     926    132    170   123   962   963  2818  2863.1  6325.8  905.9  0.10%
//
// Measured 14-bit value of 905.0 Ohm 4-Wire test resistor using 128 samples is 905.1 Ohms with 0.01% error (ENOB is 13.1)

As an aside, if using a sensor which is excited directly by VDD (such as a 4-way bridge) to get the best VDD rejection the excitation voltage should be used as the reference, not the internal bandgap. In such a case noise on VDD (and hence both AC input and reference) then cancels the noise on the ADC input.