LM34 temp creep due to sun/selfheating?

So I just transferred my project from a bread board (where everything was working great) to a more permanent solution and I'm having a weird issue where my LM34 sensor starts creeping up from about 65F (correct) to over 100F even though the ambient temperature hasn't changed. Anyways, unless I'm mistaken, it seems really sensitive to sunlight- as the temps start dropping as soon as I put it in the shade, but doesn't seems to stabilize around 70F.

I could imagine a slight increase in temp due to morning sunlight during a cool morning, but 40F bump? I'm not placing it in an enclosure- not sure why it seems to be self heating. Ideas?

Anyways, unless I'm mistaken, it seems really sensitive to sunlight- as the temps start dropping as soon as I put it in the shade, but doesn't seems to stabilize around 70F.

My old school mercury thermometer mounted in my garden reads way way high when sunshine is directly hitting it, and only accurate when mounted in the shade. I think if you want to read true air temperature you have to have the sensors shaded from direct sunlight as that is influnced by both air temp and direct photon exposure?

As far as any real calibration difference from your breadboard version Vs your standalone version I would have to see how you are powering the circuit between the two versions. Arduino analog input accuracy is directly related to how accurate the +5vdc supply voltage is.

Lefty

Good point about the voltage. I'm not using an arduino board for this, but I do have AREF tied to ground w/ a .1uF cap. My Arduino usually measures about 5.02V while this power supply is about 4.96V.

I've been reading about the AREF pin... mostly seems used to measure values less then 5V since the default is to use the internal 5V reference voltage... I'd assume though that's just reading the voltage off of AVCC/VCC? Is there any reason to tie the AREF pin to the 5V power supply?

Here's the circuit... right now I'm not using the MAX756 for power- just a sparkfun breadboard power supply.

If you have the LM34 package itself in direct sunlight, then I can easily believe a 40ยบ increase over ambient. Black heats up considerably in direct sunlight. Black is black because it's absorbing all wavelengths of light, reflecting none of them, which causes black objects to rise in temperature when in direct sunlight. On the flip side white is white because it reflects all wavelengths of light. That's why you see exterior thermometers of nearly any variety painted white.

So, paint the sensor white and/or keep it in the shade. I'd personally do both.

Yeah, I get that... just that it's heating too fast IMHO- a couple of minutes for 40 deg change on a cool day with a slight breeze? Seems unlikely. It didn't feel warm, and if you've ever owned a black painted car you know it takes longer then that to warm up.

Honestly, I started having problems with the thermistors shortly there after as well... not sure what was going on, but all my analog readings became suspect after a while. Was too busy smoking the brisket and getting ready for the party to start debugging. Maybe I have a short or a bad ground... it's definitely a lot harder to keep all the wiring nice and clean one a perfboard then on a breadboard.

Hopefully I can figure it out and get things working properly- I've spent a lot of effort getting the thermistors to work and switching to thermocouples would not only be expensive, but end up costing a lot of time.

So, paint the sensor white and/or keep it in the shade

Most white paint is only white in visible light - it has to be 'white' across the infra red band to really do a good job. Much simpler to glue on some aluminium foil which has a higher albedo still. Direct sunlight is of the order of 100mW per square centimeter.

Forced air circulation would protect against solar and self heating by matching the package temperature to the ambient air much more aggressively.