Seeking Simple Temperature Solution

What exactly are you meaning by this word "simple"? The LM34 gives you a simple output voltage. If you are looking for a digital sensor, look at Microchip's I2C/SPI offerings. IIRC Dallas (now Maxim) also has 1- wire digital temp sensors.

If it is just gas in the line, with no air, then there is no oxygen and hence no possibility of combustion even if you were sending an arc across the pipe. If you don't want to puncture the pipe - and I probably wouldn't - then maybe you can calibrate your sensor to infer the temperature of the gas from the temperature of the pipe surface vs ambient.

Why? Too lazy to code? I have a thermistor measuring hot water between 80F and 120F. Just 5 lines of C and a lookup table for the linear mapping.

temp

deal

need.

I don't mind the programming, but I didn't want to calibrate the probe. I don't have the equipment to simulate various temperatures and find the three constants required for calculation.

I'm not sure how to implement a lookup table. I can do 256 lines of switch/case, but that seems tedious and inefficent.

what about using the measured value as index into an array holding the lookup table ?

lukasz

--
-------------------------------------------------------------------------
  Lukasz Salwinski                             PHONE:        310-825-1402
  UCLA-DOE Institute for Genomics & Proteomics   FAX:        310-206-3914
  UCLA, Los Angeles                            EMAIL: lukasz@mbi.ucla.edu
-------------------------------------------------------------------------

Yes, just an array of 10 to 20 data points should be sufficient. I haven't fine tune my data yet. But so far so good with:

lup[] {int r, t} = {{0,0},{90,80},{100,88},{110,95},{120,102}, {130,108},{140,115}};

for(i=0; i lup[i].r) return((int)lup[i].t); }

I am using a 5% thermistor. 1% is available for a little more. How accurate do you need?

You could use the SE95:

It is cheap, has a resolution of 0.03125°C and a accuracy of ±1°C from

-25°C to +100°C. Last week I've implemented a simple lowpass filter (a simple first order filter , like described here:

) for such a sensor build on a board to measure the board and ambient temperature in a case and if you use the default rate of 10 conversion per second and a lowpass filter cutoff frequency of 0.1 Hz or lower, it is very stable, much better than 0.1°C. And even blowing the sensor very lightly causes a significant change in the filtered value. Maybe this could be useful for a digital candle, which can be blown off :-)

--
Frank Buss, fb@frank-buss.de
http://www.frank-buss.de, http://www.it4-systems.de

I'll look into the SE95. I need at least 5% accuracy.

So ±1°C accuracy from -25°C to +100°C should work. Usually they are even more accurate near room temperature, like for your range from 4°C to 30°C.

This is some C code for calculating the temperature from the value returned from the chip:

float calculateSe95Temperature(int data) { const float resolution = 0.03125; const int bits = 13; const int msb = 1 = max || data < 0) return 0; if (data & msb) data = data - max; return float(data) * resolution; }

"data" is calculated like this, after reading the temperature registers in unsigned char i2cBuffer:

int data = (i2cBuffer[0] > 3);

I've used this on an embedded Linux system, could be optimized for microcontrollers and fixed point numbers, or if you don't expect negative numbers.

--
Frank Buss, fb@frank-buss.de
http://www.frank-buss.de, http://www.it4-systems.de

You don't have hot and cold running water, half a soda bottle, and a cheap kitchen thermometer?

In fairness though, when I didn't feel like going through that I used an SPI sensor - AD7814ARM I think it way.

Yes, I don't have hot enough water for the kitchen, i.e. dishwasher. So, I repipe my house with a cheap AVR temperature controller. The thermistor is small enough to fit inside a 3/8 hole of a pex pipe. Hopefully, it will hold 55 PSI. If not, I will try it in the brass fitting.

Here are pictures of the controller, relay and pump.

On Feb 13, 4:01=EF=BF=BDpm, "eeboarder" wrote= :

d.

use a pt100 resistive element, simple, cheap, accurate, available.

On Fri, 13 Feb 2009 08:01:19 -0800, eeboarder wrote (in article ):

What is your A/D like? How many bits? Any transistor or diode will work. Take a 2N3904 and run a small current, like a mA. Measure the drop across the junction with the A/D. It is directly proportional to temperature - nice and linear at these temperatures and current.

But, you have to calibrate. Icewater and boiling water is fine. Set offset with ice-water mix for zero or 32, set gain (software if you have enough bits or hardware if not) at boiling to get a convenient change like 1 mV per degree or a delta of 5 on your A/D reading. The whole thing should cost less than 5 cents if you already have the A/D. 25 cents if you need an op-amp.

-- Charlie Springer

How about Excel and the Data sheet?

Your temp. range is fairly narrow and upper/lower are not extreme. There are number of possible solutions, but I'm a DalSemi 1-wire bigot so..... ;-) Several years ago I wanted to monitor the gas water heater in my very drafty basement to see how often the burner lit. I taped a DS1820

1-wire temp. sensor to the copper pipe exiting the heater and ran the three twisted wire-wrap wires upstairs through an existing hole in a baseboard. Every 3 minutes, 24/7, a uC read the temp. and logged the time-stamped value in EEPROM. While the data did not give the temp. of the water inside the tank (which I didn't care about), spikes in a plot of that data showed how often the burner fired up.

Regarding your application, I would not put a sensor inside the gas line. I would just tape it firmly to the outside of the pipe. If you want the measured temp. to more accurately reflect that of the gas inside the pipe, insulate several feet of the pipe (sensor underneath) with readily available foam that is made for water pipe. It's very cheap.

--
Michael

number of possible

wanted to monitor the

I taped a DS1820

twisted wire-wrap wires

read the temp. and

temp. of the water inside

often the burner

What is the big deal with sensor calibrations? Thermistor can measure

-40C to 120C (-40F to 248F). Just put a thermometer next to the thermistor and adjust the difference in reading. Run it through your operating range and be done with it. You only need to do it once.

temprange

If you want good accuracy, you'll need to compensate for altitude for the temperature of the boiling water. In both the freezing and boiling cases, you want to use decent water. Here in the desert, our readily available water is at least 10% moisture.

emp

al

need.

ed

here are number of possible

l years ago I wanted to monitor the

lit. =A0I taped a DS1820

hree twisted wire-wrap wires

4/7, a uC read the temp. and

the temp. of the water inside

wed how often the burner

e quoted text -

Here is the code with linear interpolation:

#define MAX_LUP 8

struct{int r, t}lup[MAX_LUP] =3D {{0,0},{80,72},{90,80},{100,88},{110,95},{120,102},{130,108}, {140,115}};

map(int r) { int i,j,r1,t1,r2,t2;

for(i=3D0; i

... snip ...

I understand that Antarctica is the most arid desert extant. If true, it would appear you could get water with a moisture density below 10%. :-)

--
 [mail]: Chuck F (cbfalconer at maineline dot net) 
 [page]: 
            Try the download section.

Mine was a variant of a joke in the High Plains during the dirty

50s when it would sometimes rain mud. The joke was that the rain was 10% moisture. The desert reference was to the amount of chemicals in the water. I have no idea what the chemicals do to the boiling and freezing points but I'm certain it's not good.