Please help with code for DS18B20

[fratello]

THANK YOU VERRY MUCH !!!! You make me happy ! I will try the code and I will keep You informed. Thanks again !
****** A HAPPY NEW YEAR !!!! ******
******* TO ALL !!!! *******
LE : Now is great, even still remain the 1°C difference...but "OVERALL do not ask a +/- 0.5°C precision sensor to give 1/100° C .... !!!".
Thanks again !
B.t.w : what can I use instead of DS18B20 for more accurate measurement ?

[Acetronics2]

B.t.w : what can I use instead of DS18B20 for more accurate measurement ?

Hi, Fratello

I had a look to all the sensors you can find on the market ...

The best "out of the box" accuracy ( all including ) will be roughly +/- 0.5 °C ... don't be sad !!!

going further needs multipoint CALIBRATION ... with some really expensive devices.

Note linearity is "not so bad" in our measuring range ... so, with a two points calibration ( 0 and 100°C ) , you can reach +/- 0.1 °C without too much pain.

And remember ... DS18x20 in continuous measurement mode like here, without being fixed on a heatsink, has a 1°C self heating value ... ( a look at the Datasheet will confirm !!! )

Just have a look to the values displayed from the power up ...

Alain

[Melanie]

You missed some Alain.

We have Industrial Sensors that directly return 16-bit Temperature to two Decimal Places (in the range 0.00C-100.00C).

It is however a complete waste of time... just breathing in the vicinity, or even just walking nearby create air-currents which wildly change the readings - we just end up ignoring the last digit. If you're trying to distill some moonshine for the New Year festivities, you're not really bothered in monitoring the temperature to hundredths of a degree.

[Acetronics2]

Hi, Mel

Happy to read from you !!!

Of course, those sensors exist ... ( you have forgotten to give us the price ... lol ), but are totally out of our Hobbiysts galaxy ...

Have a Happy new Year

Regards
Alain

[Melanie]

Happy New Year to you and everyone too...

Aa a rough guide about 25 Euro's in singles, falling to below about 18 Euro's at 100 pcs, and about 10 Euro's at 1000 pcs.

Yes of course, you can't beat the few cents a DS1820 costs, but then it's not a bare chip, but a fully enclosed product with attached leads and connectors, post sensor processor, integral noise immunity, etc etc with integral seals all ready to bolt straight into your pipework or manifold.

If you want cheap, then just use an NTC Thermistor and plot it's curve into a PIC. You can easilly get better than 0.1C accuracy through simple interpolation across 90% of the curve. It's the way it's done in 99.9% of the cheaper Industrial and domestic Controls. I really don't see the point of DS1820's, LM35's etc, they're a waste of money for something that can be done so much simpler if only folks stopped and thought about the problem. OK, if you've only got a single I/O and a dozen items on the interconnecting bus, then fine, but otherwise why?

[Acetronics2]

Ok, Mel

You're talking resolution while I'm talking accuracy ...

DS already give .0625°C resolution ... but that do not mean anything because you only know you are inside the +/- 0.5 °C error gap.
LM 335 have your ADC resolution ... but always an "out of the box" +/- 0.5 °C precision

That is why I say ( and repeat ... ) you can't get a better reading WITHOUT a calibration ... or comparison with a known reference.

Even the MS5534 ( 16 bits! ) which comes with its calibration coeffs ... is given for a GUARANTEED +/- .8°C accuracy ...

For NTC's the problem is the +/- 5 to 20% value tolerance ... that need an individual curve adjustment to each sensor ( only 1 point ... yes... but one point calibration ! )


The only **predictable** High accuracy sensors are Platinium RTD's ... But , ... How much to pay for a class A probe ??? ...

Temp and Thermal measurements are always funny ... and make you humble !!!

Alain

[fratello]

Thank you for replays !!!
I wish to all a happy NEW YEAR 2009 !

I think I go crazzy with this little thermostat probleme ; I write ( based on Yours great codes) four routines ; everyone display a different temperature. Which of them is -in yours opinion - the best ?

1.OWOut DQ, 1, [$CC, $BE]
OWIn DQ, 2, [Temperature.Lowbyte, Temperature.Highbyte]
Convert_Temp:
Dummy = 625 * Temperature
TempC = DIV32 10
TempC = (Temperature & $7FF) >> 4
Float = ((Temperature.Lowbyte & $0F ) * 25 )>>2
Temperature = TempC*100 + Float

2.OWOut DQ, 1, [$CC, $BE]
OWIn DQ, 2, [RAWTEMP.Lowbyte, RAWTEMP.Highbyte]
Convert_Temp:
dummy = RAWTEMP * 625
TempC = DIV32 100
Temperature = TempC

3.OWOut DQ, 1, [$CC, $BE]
OWIn DQ, 2, [Temperature.Lowbyte, Temperature.Highbyte]
Convert_Temp:
TempC = (Temperature & $7FF) >> 4
Float = ((Temperature.Lowbyte & $0F ) * 25 )>>2
Temperature = TempC*100 + Float

4.OWOut DQ, 1, [$CC, $BE]
OWIn DQ, 0, [Temperature.LowByte, Temperature.HighByte, Skip 4, Count_Remain, Count_Per_C]
Convert_Temp :
Temperature = (((Temperature >> 1) * 100) - 25) + (((Count_Per_C - Count_Remain) * 100) / Count_Per_C)
if Temperature > 32767 then '32767 check if BIT12=1
Temperature= ~Temperature
temperature=temperature>>3 ; divide by 8

Thanks in advance for your support ; I hope I don't became boring !!!!