My emonGLCD is showing a room temperature about 3 degrees C higher than it should. I placed a temperature probe onto the Dallas sensor and the probe read 1 degree higher than ambient (maybe heat from the backlight and Atmega chip) but not 3 degrees higher. I understand the Dallas chips are pre-calibrated so it should be correct without software adjustment.
As you say I believe the DS18B20 sensors are pre-calibrated. the higher temperature your seeing must be due to a bit of heat from the emonGLCD. What temperature probe did you use the test against?
I have taken the DS18B20 temperature sensor off the PCB and soldered it temporarily onto the end of some long wires. Now it shows a more accurate figure as shown in the photo (compared to the central heating controller which I believe is quite accurate). I was surprised that the air temperature above the Atmega chip was 2-3 degs higher than ambient, maybe the heat is being conducted from the Atmega pins, down the short copper tracks to the sensor pins.
Overall we are very chuffed with our solar monitor and so were Solar Peak who installed the 4kW system on our roof. The Fronius inverter they recommended only had blue tooth as standard which would not reach from the cellar up two floors to our flat. And that only shows PV production.
Thanks Glyn and Co.
i´ve written a blog post on how i soved this problem on my board.
Just take a look: http://blog.pcunha.org/2013/01/emonglcd-temperature-hack-correcao-para.html
A friend (Brian D on here) has just suggested turning the unit upside down and changing the display in software so that it works the other way up. A bit like an iPad.
Seems like sensible lateral thinking to me...
Thanks for the nice blog Pcunha, I've put a link to it from the emonGLCD build guide page.
I didn't link to Paulo's blog when I added the note because the pictures did not show for me!
Open-source tools for energy monitoring and analysis. This project uses the GNU General Public Licence