SPI dropping least significant bit on ESP32-S2 but works on PI Pico

[CTGControls]

CircuitPython version

Adafruit CircuitPython 7.3.0 on 2022-05-23; Adafruit Metro ESP32S2 with ESP32S2
Board ID:adafruit_metro_esp32s2

Code/REPL

Not working esp32-2 repo
https://github.com/CTGControls/CTG_DPS_Python_Ad4116_Adafruit_MetroExpressEsp32S2

Working PI Pico repo:
https://github.com/CTGControls/CTG_DPS_Python_Ad4116_RaspberryPiPico

Behavior

I get:

get_register: got [ 0x80 0x00 0x54 ] from reg [ 0x30 ] using command [ 0x70 ]

but should get

get_register: got [ 0x80 0x00 0x55 ] from reg [ 0x30 ] using command [ 0x70 ]

Description

Only happens on the Metro ESP32-S2 but not on a PI Pico.

Additional information

No response

[CTGControls]

I have not test without using "adafruit_bus_device.spi_device" as you asked me to in Discord, but I will in the next few days.

[dhalbert]

The AD4116 datasheet says:

DIGITAL COMMUNICATION
The AD4116 has a 3-wire or 4-wire SPI that is compatible with
QSPI™, MICROWIRE®, and DSPs. The interface operates in SPI
Mode 3 and can be operated with CS tied low. In SPI Mode 3,
SCLK idles high, the falling edge of SCLK is the drive edge, and
the rising edge of SCLK is the sample edge. Data is clocked out
on the falling and drive edge and data is clocked in on the rising
and sample edge.

Given that it says it operates in mode 3 (opposite phase and polarity from standard SPI), and with CS tied low, not using adafruit_busdevice makes sense. Just just busio.SPI, and use .try_lock() to grab the bus. The opposite phase and polarity can cause difficulties because the "normal" states are different from the initial resting states of the pins. It may be a phase/polarity issue in the Espressif implementation. I brought this up during an internal meeting today, and was told that we have had trouble running SPI displays with mode 3 and CS pin tied down, so we in general don't support that kind of operation. Some cheap displays don't expose the CS pin.

[CTGControls]

This is rather interesting because many of the high end Analog Device chips do use mode 3 only. Such as:

• AD411x (AD4111, AD4112, AD4113, AD4114, AD4115, AD4116)
• AD717x (AD7172, AD7173, AD7175, AD7175, AD7176, AD7177)

Others use both Mode 0 and Mode 3. Such as:

• AD3552R (give different results if you use one mode over the other)

It looks like AD is using Mode 3 for most of there newest chips as well. I normally always use mode 3 if available as I have seen better results.

As this is only my second week using Python I am trying to get up to speed.

[dhalbert]

@CTGControls Did you get this to work without adafruit_bus_device?

[CTGControls]

The bug is not in the adafruit_bus_device. The bug is some where in busio.SPI. I updated my code in the:

Not working esp32-2 repo
https://github.com/CTGControls/CTG_DPS_Python_Ad4116_Adafruit_MetroExpressEsp32S2

code.py

###OLD code
# _CTG_AD4116Class = CTG_AD4116Class

# _CTG_AD4116_ = CTG_AD4116_

# _CTG_AD4116_.startup(spiClk= 50000, CTG_AD4116= _CTG_AD4116Class)

###New code
_CTG_AD4116Class = CTG_AD4116ClassNoAdafruitBusDevice

_CTG_AD4116_ = CTG_AD4116_NoAdafruitBusDevice

_CTG_AD4116_.startup(spiClk= 50000, CTG_AD4116ClassNoAdafruitBusDevice= _CTG_AD4116Class)

Still getting:

get_register: got [ 0x80 0x00 0x54 ] from reg [ 0x30 ] using command [ 0x70 ]

but should get

get_register: got [ 0x80 0x00 0x55 ] from reg [ 0x30 ] using command [ 0x70 ]

[dhalbert]

Adafruit makes some Analog Devices accelerometer breakouts that use mode 3 for SPI. I will get some of these and see if the same problem appears. The AD411x eval boards are about $100 each (!).

[CTGControls]

If you send me an address I will send you an EVAL-AD4116ASDZ (https://www.analog.com/en/design-center/evaluation-hardware-and-software/evaluation-boards-kits/EVAL-AD4116.html) I use this with the Adafruit Metro ESP32-S2 (https://www.adafruit.com/product/4775?gclid=Cj0KCQjwvZCZBhCiARIsAPXbajsAXsF60acbyEb-QYh5ax3wf3YyIqgxCwU-X37cYN88wchRbTFKNWAaAlRUEALw_wcB)

Christopher Gay
Chris@CTGControls.com

[dhalbert]

Thank you for the offer. I will test with an ADXL343 (https://www.adafruit.com/product/4097), which is an Analog Devices accelerometer that uses mode 3 SPI first, and see if I can reproduce. I would think I can.