At http://www.kandrsmith.org/RJS/Misc/hygrometers.html there is a lot of tests on various hygrometers, like dht22, si7021 etc.

A useful comment if you are measuring quite seldom can be found at;
http://www.kandrsmith.org/RJS/Misc/Hygrometers/dht22_other.html
namely, that upon "measurement" DHT22 actually returns the previous measurement from memory - then actually does a measurement and stores that in memory.
If you poll once per hour, you get one hour old data every time. Poll twice, or poll "often enough".

Please read http://forum.arduino.cc/index.php?topic=54623.0 which basically says that the 3.3v arduino can't read stuff at 115200 baud correctly, things get corrupted due to being unable to get the correct timings. In our tests, semicolon (;) gets corrupted quite often. Unfortunately, the semicolon is "quite useful" in the serial protocol of the Gateway.

Symptoms: Gateway not responding to version checks, gateway not being able to pass on auto-allocated ID's etc.

Solution: Either change clock to 12MHz (hardware mod + software + blah), or just go slower.. 19200 baud seems to work fine for us. Change baud rate in MySensors.h and recompile your serial gateway. Easy-peasy.

V_DESCRIPTION or somesuch, so the nodes can present what each sensor is.. like "DHT Humidity" "DHT temperature" "DS18x20 temperature" "Distance to garage door" "Garage door locked" ... Right now I'm using VAR5.. but..

Using a non-scientific method; looking at the angle of battery discharge graphs over a week, I've noticed that the modules I got from ITEAD are draining batteries at about half the rate compared to an unknown (but probably fake) pretty good module.
ITEAD module; slightly under 1% per day
"other modules"; about 2% per day
This is when measuring and transmitting DHT21 data every minute.

Same card, different box (slightly too small). I cut the LED after taking the picture..

What are you using the high baudrate with? Due to clock frequencies etc, the pro mini can't even generate (actually, read) 115200 baud correctly..

Without further info, it looks like a case of "Doctor, it hurts when I do this" "Well, don't do that then.."

@Pierre-P
That i2c thing is painted and works because it's capacitive (it can even sense when your finger is hovering a millimeter or two away from it, like a capacitive touch screen (your phone)). I put one of them into the soil yesterday. If I remember, I will get back regarding results.. I placed it into the same flower as a resistive fork, with the middle chip (so I have some corrosion).

@meanmrgreen See my post a few up, I'm seeing some weird effects when connecting to more than one plant..

@mfalkvidd I'm going to do some tests with changing the sampling time and sleep time, but I'm not so hopeful since the effect is seen hours/days after an "unrelated" change.. I got stuck yesterday playing with ESPEasy, but I'll get back to debugging this..

I have an arduino pro mini, connected to 2-3 flowers using A0+A1, A2+A3, A4+A5.. unfortunately, I'm seeing some weird effects between the plants.. if I water one plant, another can change a bunch at the same time (or not, depending on stuff).. Also if I disconnect one flower, another flower can get way different values too..

I haven't figured out yet if I'm getting a capacitor effect in the flowers, sending back power through the output signals or what's happening..

Here I added water to the "green" flower, then the values for the gray one dropped. The gray one was physically untouched while watering.

Code is based on mfalkvidd, with all Ax being set to output=0 and then input+pullup for one while reading..

Lower the speed from 115200 to 19200, 9600 or similar. The frequency of the arduino is not able to match 115200 baud exactly, so the arduino and rpi are mis-communicating.

From another thread; http://wormfood.net/avrbaudcalc.php?postclock=8&hidetables=1

Solved it for me.

At http://www.kandrsmith.org/RJS/Misc/hygrometers.html there is a lot of tests on various hygrometers, like dht22, si7021 etc.

A useful comment if you are measuring quite seldom can be found at;
http://www.kandrsmith.org/RJS/Misc/Hygrometers/dht22_other.html
namely, that upon "measurement" DHT22 actually returns the previous measurement from memory - then actually does a measurement and stores that in memory.
If you poll once per hour, you get one hour old data every time. Poll twice, or poll "often enough".

@Pierre-P
That i2c thing is painted and works because it's capacitive (it can even sense when your finger is hovering a millimeter or two away from it, like a capacitive touch screen (your phone)). I put one of them into the soil yesterday. If I remember, I will get back regarding results.. I placed it into the same flower as a resistive fork, with the middle chip (so I have some corrosion).

When I arrived at work today and started checking mails, I had received a bunch of them saying that the temperature in the fridge was too high, so I had to drive home and close the door. I would have gotten a notification earlier through jabber if the server software we use for jabber wasn't acting up :(.
MySensors stuff was not involved in that particular sensor (esic sensor + tellstick + perl + perl), but the need to monitor this stuff was definitely there.

.. and you can be prosecuted for putting others in danger etc.

@MikeF Maybe an idea about turning on the display.. how about adding a PIR, and turn on the display if movement is detected..? My guess is that it'll lower the power consumption, and still show info when someone is nearby.

What are you using the high baudrate with? Due to clock frequencies etc, the pro mini can't even generate (actually, read) 115200 baud correctly..

Without further info, it looks like a case of "Doctor, it hurts when I do this" "Well, don't do that then.."