WiPy ADC

[B_Rubble]

Hi all,

I'm having some issues with the embedded ADC function, and was hoping to get some insight. I need to measure 0-10v, so I put together a simple voltage divider to scale that down to 0-1

-My ADC is not registering different values, when fed 0.2v and 0.7v I received the same value.
-Stability/Timing? I notice that my ADC values sporadically drop. I have tried with and without a 100ms pause, and with a 100nf capacitor, but this does not seem to help.

def checkpin():
    time.sleep_ms(10)
    adc = ADC()
    adc_c = adc.channel(pin= 'P13', attn=ADC.ATTN_11DB)
    for i in range(10):
        time.sleep_ms(100)
        b = adc_c.voltage()
        time.sleep_ms(100)
        v = 3.3*b/4096
        print(b,", ",v)
        time.sleep_ms(100)

An example output: (this does not change with different voltage input)

348 ,  0.2803711
3134 ,  2.524951
3134 ,  2.524951
142 ,  0.1144043
142 ,  0.1144043
3134 ,  2.524951
142 ,  0.1144043
3134 ,  2.524951
3134 ,  2.524951
3134 ,  2.524951

Any thoughts/suggestions would be much appreciated. I have a ADS7818 which I can try uart with, but I'm just trying to determine the easiest way to measure 0-10v ±0.1

Thank you

[robert-hh]

@b_rubble The call to adc_c.voltage() returns a voltage scaled in mV, depending on the attenuation factor you have selected. That is 11db, so the range is about 0-3.3V. if you want to get the raw ADC values, you have to call adc_c.value() or simply adc_c(). That will return a raw value in the range of 0-4096.
The ADC of the ESP32 is not very good. You get the best linearity at 6 dB attenuation, and the low offset is at about 70 mV.
I cannot tell whether you see this much changing values. Besides high noise and non-linearity I did never observe such changes. It may be related to external factors like your circuitry, sockets, ....
If I repeat the here on a LoPy4, I get plausible (albeit poor) values with you script, feeding 2V into P13. Note that only the first column is fine.

1993 ,  1.605688
2034 ,  1.638721
1969 ,  1.586352
1968 ,  1.585547
1881 ,  1.515454
1917 ,  1.544458
1971 ,  1.587964
1948 ,  1.569434
2007 ,  1.616968
1974 ,  1.590381

[B_Rubble]

@robert-hh thank you very much for your insight. I have actually used some of your previous posts to help troubleshoot other issues, so I definitely appreciate you taking the time to answer. I modified my script to just output value, voltage and fed in a fairly consistent 420mv (measured as 419-421mv through my fluke). Unfortunately it seems that even with calibration there would be too much noise for my uses. I continue to get very odd behavior, so I think I will use an off the shelf ADC to accomplish this task.

Do you have experience or suggestions for off the shelf converters? I don't have SMT capabilities with my current setup, but I do have a ADS7818 which I will try to use with UART.

Thank you

[robert-hh]

@b_rubble You can use the built-in ADC is the noise does not matter or if you can ignore the lower 4 bits.

I have used an ADS1115 with an ESP8266, attached with I2C. It has a SMD package, but you can get modules at any size, down to a DIL8 pin layout. But this is a rather slow device. The ADS7818 is faster, and it should be possible to read that with SPI. UART won't work.
So I do not know your application needs, speed or precision.

[robert-hh]

@b_rubble Here is a sample script using the ADS7818 with SPI:

# Drive the ADS7818 ADC using SPI
# Connections:
# xxPy | ADS7818
# -----|-------
# P10  |  CLK
# P11  |  CONV
# P14  |  DATA
#
from machine import SPI, Pin
from ubinascii import hexlify

spi = SPI(0, SPI.MASTER, baudrate=4000000, polarity=1, phase=1, bits=16)
buf = bytearray(2)
vref = 2.493 # measured at the ADS7818

while True:
    # start a conversion and get the result back
    spi.write_readinto(b"\xe0\x00", buf)

    # extract the raw ADC value and derive Voltage
    value = ((buf[0] << 8) | buf[1]) & 0xfff
    volt = 2.0 * vref * value / 4096

    print(hexlify(buf), value, volt)
    res= input("Next: ")
    if res == "q":
        break

The conversion and reading back is done within the single spi.write_readinto() call. It is then easy to embed that in a small class.
Note: Since the ADS7818 runs at 5V, insert a resistor of 3.3k to 10k into the data line back into the ESP32 to limit the current into the ESD diodes.