For ESP32 motor driver the ADC code does not take the attenuation in to account

[norventa]

Having looked through the ESP32 code for the motor driver, attenuation is set:

adc1_config_channel_atten(pinToADC1Channel(pin),ADC_ATTEN_11db);

in DCCTimerESP.cpp

but then it is not taken into account when calculating the current.

ESP32 documentation shows:

Vout = Dout * Vmax / Dmax (1)
`where:

Vout
Digital output result, standing for the voltage.

Dout
ADC raw digital reading result.

Vmax
Maximum measurable input analog voltage, see ADC Attenuation.

Dmax
Maximum of the output ADC raw digital reading result, which is 4095 under Single Read mode, 4095 under Continuous Read mode.
`

[habazut]

So where is the problem? Are the measurements incorrect? How did you measure current? With the attenuation set to 11db it should match the factor in the motor driver. See Motordrivers.h and Motordriver.cpp:

unsigned int MotorDriver::raw2mA( int raw) {
  return (int32_t)raw * senseFactorInternal / senseScale;
}
unsigned int MotorDriver::mA2raw( unsigned int mA) {
  return (int32_t)mA * senseScale / senseFactorInternal;
}

Harald.

Fun fact: They call ADC_ATTEN_DB_11deprecated and write that you should use 12 but their table lists 11 and not 12.

[norventa]

sense_factor=0.7;
senseScale=256;
senseFactorInternal=sense_factor * senseScale;

unsigned int MotorDriver::raw2mA( int raw) {
return (int32_t)raw * senseFactorInternal / senseScale;
}

raw < 4096, so as long as senseFactorInternal < 2^32 / 4096 there will be no integer overflow. sense_factor isn't limited but I suspect a maximum of 16 would cover all bases:
senseFactorInternal <= 16 * senseScale
senseFactorInternal < 2^32 / 4096
or senseFactorInternal < 16 * 65536
thus senseScale can be 65536.

By the way, for ESP32 there is little need to use senseScale as it has a float maths unit.

The motor shield outputs 1.65V per Ampere. Taking the reciprical 1/1.65A per volt. That is 0.6A/V. So shouldn't sense_factor=0.6?

With 11dB attenuation the ESP32's ADC can measure upto 2.45V:
2.45/1.65 = 1.5A

I haven't measured the current as it fluctuates quite a bit being ac but I have had an over-current alert, which shouldn't have happened.

[habazut]

You can measure the current on the input. The ESP32 has a baseline of some hundred mA and then if you start a loco the increse is your loco. Do you have the CSB1 or any other combination of ESP32 and motor shield?

The calculation is done this way (without float) as it can be the same and efficient on AVR, ESP32 and STM32 like this. The factors are however different for the same shield on different CPUs.

Harald.

[Ash-4]

It is not a question of the attenuation/code, but that we opted for a conservative setting -- assumed that folks going the ESP32duino route would also opt for clone motor shields. If using a Genuine Arduino R3 motor shield, 0.6 might be better.

You can use the current sense factor you decide is safe to use. 0.7 is not simply a mathematical result, but the result of testing and recognizes the ADC range of the ESP32-WROOM is 0.15V-2.45V. It is also based on the use of the 5k/20k voltage divider and the benefit of the Schottky diodes present on A2,A3.
https://dcc-ex.com/mkdocs-test/reference/esp32/esp32-l298-shield/#clone-motor-shields

If you look at the 611mA in the second column, the pincount of 850 (not shown) would result a current sense factor of 0.719.
https://dcc-ex.com/mkdocs-test/reference/hardware/motorboards/l298-shields/#voltage-drop
additional testing data:

1135mA - pincount 1600 - factor 0.709
1585mA - pincount 2224 - factor 0.713
1939mA - pincount 2783 - factor 0.697

The settings are conservative if you are using the Genuine Arduino R3 motor shield.

The discord server is good for an interactive discussion.
https://dcc-ex.com/support/index.html#preferred-option-discord