Atlas Ph Ezo Experience Communication Problem Solved

Mycodo Mycodo Issue/Bug
,
1

:test_tube: Debugging Atlas Scientific pH Sensor with Mycodo on Raspberry Pi (I²C & UART Troubles)

TL;DR
My Atlas Scientific pH EZO sensor worked fine with Mycodo on Raspberry Pi—until it didn’t. It suddenly stopped responding over I²C after long stable operation. Switching to UART helped temporarily, but odd behavior followed. This post documents the problem, the debugging journey, and how it was finally solved.

:toolbox: Setup Summary

  • Sensor: Atlas Scientific pH EZO
  • Controller: Raspberry Pi 4B (Bullseye)
  • Platform: Mycodo
  • First interface: I²C
  • Later interface: UART (via /dev/ttyS0)
  • Power source: Dedicated clean 5V PSU
  • Other tools: PL2303 USB to UART adapter, PuTTY, Python serial script

:exclamation: Problem Description

After a few months of stable measurements, the sensor stopped responding through I²C.

Over I²C:

  • i2cdetect no longer showed the EZO board.
  • Mycodo logs had:
    Read failed. No device detected at address.
  • Even after reboots and reconfiguration, Mycodo failed to read.

Switched to UART:

  • Initially worked.

  • But later, the following began appearing in Mycodo logs:

    • SerialException: device reports readiness to read but returned no data
    • UnicodeDecodeError: invalid start byte
    • AttributeError: 'NoneType' object has no attribute 'isalpha'

  • Sending R via PuTTY or a Python script returned:

    *OK57.11

    or

    *ER

  • The sensor responded inconsistently, especially after using continuous mode C.

  • Eventually, no data was received.

:male_detective: Suspected Root Causes

These issues seem to stem from one or more of the following:

  1. Sensor Buffer Corruption due to prolonged use of continuous mode (C)
  2. Improper interface switching between I²C and UART without full power cycle
  3. Multiple processes accessing UART (e.g. Mycodo + PuTTY at the same time)
  4. Over-polling the sensor at too short intervals (e.g. 1-second loop)
  5. Leftover configuration from Mycodo conflicting with manual serial reads
  6. PL2303 USB adapter on Windows not properly releasing the COM port
  7. Garbage data from sensor when in an unknown internal state

:white_check_mark: Solution (What Worked)

  • Factory reset the sensor using UART:

    → Command: Factory
← Response: *OK

  • After reset:

    • Normal R reads returned 6.554, etc.
    • No more errors unless continuous mode was enabled again.

  • Fully disconnected power when switching between I²C and UART modes.

  • Ensured only one interface enabled at a time—removed I²C pull-ups and jumpers when using UART.

  • Set Mycodo polling interval to 5 seconds or more.

  • Disabled continuous mode (C,0) and only used discrete reads (R).

  • Used Python + pyserial script to test commands safely outside Mycodo.

:test_tube: Python Debug Terminal (UART)

import serial

ser = serial.Serial("COM6", 9600, timeout=1)
print("Atlas pH Sensor Terminal. Type 'exit' to quit.")
print("-" * 50)

while True:
    cmd = input("→ Command: ")
    if cmd.lower() == "exit":
        break
    ser.write((cmd + "\r").encode())
    resp = ser.readline().decode(errors="ignore").strip()
    print("← Response:", resp)

ser.close()

:white_check_mark: Signs of Recovery

After re-integrating with Mycodo and properly configuring the UART input:

  • Readings like success: ['6.544', '*OK'] began appearing.
  • pH values were stored in InfluxDB successfully.
  • Temperature compensation from another sensor (?T,27.0) was confirmed working.
  • No more decoding or attribute errors.
  • Uptime remained stable over several hours.

:brain: Final Thoughts

Atlas Scientific pH sensors are precise but can be sensitive to communication state. This incident taught me a few critical lessons:

:bulb: Lessons Learned

  • Always isolate interfaces: don’t leave I²C and UART jumpers both connected.
  • Avoid continuous mode (C,1) unless necessary. It tends to cause state lockups.
  • Restart hardware when switching modes.
  • If Mycodo fails suddenly after working: check UART/I²C conflicts and reset the sensor.
  • Use manual tools (PuTTY / Python) to confirm hardware is functioning before debugging Mycodo.
  • Add logging in your scripts to trace unexpected states.

:boom: Most Important:

Even if your system was stable for hours or days, sudden crashes can and do happen. Don’t assume hardware failure. Debug layer by layer.

Tags: atlas-scientific, ph-sensor, raspberry-pi, mycodo, uart, i2c, sensor-troubleshooting, diy, pyserial, factory-reset, ezo

2

i fixed this issue with chatgpt, also this topic fully wrote by chatgpt also. im sorry if is there any wrong explanations. :grimacing: