Long term experience, they work a while and than they freak out. Reliable moisture sensing in soil is a difficult task.

Hi, thanks for the deep explanation, what type of epoxy do you use?

There are some problems with the capacitive moisture sensor. First and most important, there is a design mistake on thousands of these circuit boards. If you lay down the device in front of you on the bottom you see C4 and next to it on the right R4. The right side of R4 and the left side of C4 should be connected. The both contacts should have conductivity to Ground. Messsure from ground to R4 right, if you do not get conductivity you got a sensor with the design fault. Than you need to run a short wire from R4 (right side) to C4 (left side), because C4 is connected to Ground. The 1 MOhm Resistor R4 is the discharge resistor for C4, wich is the smoothing capacitor for the Puls-signal of the 555 Chip. Without the discharge resistor the cap slowly runs up higher/inconsistent values.

Next de-solder the 3 wire socket and solder a 3 wire cable straight to the board. Than connect 5 Volts and meassure the output. Should be in the range 3V-4V or 1 to 2.5V depending on chip and Voltage-regulator.

The connection between the MCP and the sensor needs to be shielded well. The 4 Data connections between the pi and the MCP needs to be level-shifted. The pi runs on 3.3V the MCP on 5V.

Last, dip the entire circuit board from head to tow with cable in epoxy. When dry take the readings of the device again ( in air and in water) with voltmeter and put this numbers into the calibration settings on the hydrosys settings page. Make small adjustments to the numbers until desired result. Done.

Other suggestion is to proceed excluding the possible issues, for example remove the sensors an put a known tension on one of the input pin of the mcp3008. You can use two resistors to make a partition. Keep it for a couple of days to see if you got the same reading issue. I also suggest to use only one input for this testing.

Thanks for the response Angelo.

i set calibration to blank and changes setting to voltage, here are the results:

All the "joints" are soldered. I dont understand why i can get consistent readings like everyone else? There must be something cause erratice behavours. And I have 7 sensors to put in my pots, but pointless at this point.

Any other suggestions?

Hi There,

It looks like an unstable contact.

The spike at around 250 after the calibration means that the sensor is reading zero volts if we apply the inverse formula (-250/100*(2.16-1.25)+2.16).

To have a clearer view about the voltage reading, I'd put all the calibrations value to zero so the reading will be the actual voltage coming out from the Sensor.

From the data I see that the MCP3008 provides the value at each interval, so the connection from the MCP to the raspberry should be ok. Seems that the problem happen at around same time for both sensors so it might be on some common pin like the reference voltage which drops to zero. I suggest then to check the Vref pin on the MCP3008.

Let me know if things gets better.

Regards,

Angelo