On Thu, 01 Jan 2015 11:41:23 -0800, Jeff Liebermann
wrote:

Blundering onward...

Siting:
https://www.campbellsci.com/weather-station-siting
ftp://ftp.campbellsci.com/pub/outgoing/apnotes/siting.pdf
http://wxqa.com/resources.html

This one is quite good on siteing.
http://weather.gladstonefamily.net/CWOP_Guide.pdf

Basically, what you're missing with your unspecified model Acurite
weather station is a radiation shield for the temperature sensor. What
you're trying to do is measure the air temperature, not the
temperature of the plastic box, the nearby walls, exhaust vents,
parking lot heat island, reflections from low-e glass, foliage
transpiration cooling, and a zillion other sources of error. Here's a
fun example:
http://802.11junk.com/jeffl/crud/KSBW-WX-Station.jpg
That's the local radio station's weather sensor array on the left,
sited over the HVAC system.

More on badly located weather stations. This site is well worth
skimming:
http://www.surfacestations.org
Even the official weather stations are often badly located:
http://www.surfacestations.org/odd_sites.htm

This is one of my early attempts at a radiation shield (pagoda):
http://802.11junk.com/jeffl/pics/wx/slides/radiation-shield-01.html
It was ok, but I missed something obvious. There's a gap near the top
of the wall where the cable goes through. Inside the building are
several high power transmitters, which produce lots of hot air, that
greatly affects the indicated temperature. Here's the CWOP report on
temperature and barometric accuracy. The station was removed and is
being moved to a better location sometime in the next month.

Enough for now. Your problem will be:
1. Do some reading on proper weather station siteing and Stevenson
Screen radiation shield construction.
2. Find a better weather station if you want accuracy.
3. Register with the various amateur weather web sites (Weather
Underground, CWOP, etc and see if your numbers track those of the
stations around you.

--
Jeff Liebermann
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558

On 02.01.15 0:48, Jeff Liebermann wrote:
On Thu, 01 Jan 2015 11:41:23 -0800, Jeff Liebermann
wrote:

Blundering onward...

Siting:
https://www.campbellsci.com/weather-station-siting
ftp://ftp.campbellsci.com/pub/outgoing/apnotes/siting.pdf
http://wxqa.com/resources.html

This one is quite good on siteing.
http://weather.gladstonefamily.net/CWOP_Guide.pdf

Basically, what you're missing with your unspecified model Acurite
weather station is a radiation shield for the temperature sensor. What
you're trying to do is measure the air temperature, not the
temperature of the plastic box, the nearby walls, exhaust vents,
parking lot heat island, reflections from low-e glass, foliage
transpiration cooling, and a zillion other sources of error. Here's a
fun example:
http://802.11junk.com/jeffl/crud/KSBW-WX-Station.jpg
That's the local radio station's weather sensor array on the left,
sited over the HVAC system.

More on badly located weather stations. This site is well worth
skimming:
http://www.surfacestations.org
Even the official weather stations are often badly located:
http://www.surfacestations.org/odd_sites.htm

This is one of my early attempts at a radiation shield (pagoda):
http://802.11junk.com/jeffl/pics/wx/slides/radiation-shield-01.html
It was ok, but I missed something obvious. There's a gap near the top
of the wall where the cable goes through. Inside the building are
several high power transmitters, which produce lots of hot air, that
greatly affects the indicated temperature. Here's the CWOP report on
temperature and barometric accuracy. The station was removed and is
being moved to a better location sometime in the next month.

Enough for now. Your problem will be:
1. Do some reading on proper weather station siteing and Stevenson
Screen radiation shield construction.
2. Find a better weather station if you want accuracy.
3. Register with the various amateur weather web sites (Weather
Underground, CWOP, etc and see if your numbers track those of the
stations around you.

4,5 and 6:Put all your thermometers in one place,
to properly compare them.
8,9 and 10:Repeat that on the other side of the house.

On Fri, 02 Jan 2015 07:23:22 +0100, Sjouke Burry
wrote:

4,5 and 6:Put all your thermometers in one place,
to properly compare them.

Much depends on the initial tolerance of the thermistor sensor.
Typically, a 1% thermistor is good for about 0.03 C error for the
thermistor alone, and about 0.5 C error if you include the associated
electronics. With a cold junction reference, 0.1 C is typical. If
the thermometer uses a 5% tolerance thermistor, just multiply
everything by 5. Note that the OP was complaining about a 6-7 degree
difference.

Details:
https://learn.adafruit.com/thermistor?view=all
See bottom of article for accuracy estimates.

8,9 and 10:Repeat that on the other side of the house.

Reminder. The goal is to measure the temperature of the air, not the
room or building walls. Doing this test indoors is just asking for
complications due to stratification (it's warmer near the ceiling) and
the multitude of local heat sources found indoors. Isolating the temp
sensor from everything except the air is why real weather stations use
radiation shields.

The problem can be reduced by isolating the thermistor and using a fan
to blow air to the sensor. The design turns out to be fairly complex:
http://www.davisnet.com/news/enews/images/1210SPARS.jpg
I'm not a big fan (pun intended) of fan aspirated radiation shields,
but they do work (if you keep them clean).
https://www.google.com/search?q=fan+aspirated+radiation+shield&tbm=isch


--
Jeff Liebermann
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558