I have a couple soil sensors running with simple bolts in the ground. No two sensors will read the same in the same soil, but that is because they are just a relative setup not an accurate setup. Doing what is recommended above is necessary. Determine what % correlates to watering time on the plants. Once this is done it is set and forget.

Also, on your batteries
"The phenomenon itself is very real. If a NiCd battery is repeatedly charged after it has only been partially discharged it will develop a lower voltage and a lower capacity. Fortunately, this effect is reversible by conditioning NiCds."

So I vote to help your poor cheap lamp batteries out by writing logic into your arduino that varies by battery voltage. You can write it in ranges

battery > X volts report every Y minutes
battery > Z volts report every P minutes
battery > .....etc

also to keep your battery in good shape you may consider every so many days ( 2 - 5 ) writing logic that says
every Q days report every 20 seconds (or another interval that will drain battery) until battery voltage = min
then
sleep until battery voltage = max

@mfalkvidd said:

thanks! That makes a ton of sense. These forums are awesome, I really appreciate everybody

@flopp

guys seriously this is just yet another reason to write your reporting interval based on battery voltage and time and not on time alone.

You are using extremely cheap systems with cheap batteries of unknown age that are likely very prone to incur reduced capacity over shorter times and memory from discharge cycles.

See the posts above. This really is one of the greatest ideas I have seen for monitoring, just needs a tweak

With these findings I would set a floor voltage around 1-1.1 V for the arduino to go into sleep mode (find your own floor voltage by looking at your sleep discharge rate and length of night time).

Adding a capacitor to the radio is what I needed to do! Made all the difference in the world!