but datasheet said use 47K for calibrating sensor in 10ppm, what about 25ppm?
Use the same value. Exact value is not important just so it is small compared to Rsensor.
Note their recommended 47K is twenty times smaller than lowest Rsensor (in air), probably twelve times lower than Rsensor in 200ppm ammonia.
could you explain more about your answers? i haven't worked with these sensors yet,..
Think of the sensor as a self-adjusting resistor Rs.
It adjusts itself according to ammonia concentration.
Place it in series with 5 volts and a resistor RL.
Measure VL,the voltage across RL.
By ohm's law, I= VL/RL
By Kirchoff's law, voltage across sensor is (5-VL)
and again by ohm's law Rs = (5-VL)/I, which is (5-VL)* RL/VL
So,,, Rsensor = (5-VL) * RL/VL
Now, you know RL and after you measure VL you'll know Rsensor.
You need to do that twice--
once in air and once in (some concentration of) ammonia.
The value you get in air is R0.
The value you get in ammonia is R(insert percent ammonia).
Now you have calibrated your sensor and resistor as a pair.
You know by how much the resistance changed in response to ammonia.
could you tell me the exact values?.
No I cannot. Neither can the people who build them.
The ratio R(percent)/R0 is fairly predictable.
The value of R0 is random within a range of limits.
As stated in the datasheet that range is between 900kohms and 4900kohms.
That's a pretty wide range.
You don't know where in the range you'll start, only what direction you'll move.
That's why you have to 'calibrate' each and every one of the sensors.
Whether you use 47K or 10K or 100K for RL is unimportant.
Just measure carefully.
Their datasheet figure 3, sensitivity curve(upper graph) must have some mistake in it.
The AIR line should be on the 1.0 line not 3.5, because in air RS IS R0 and R0/R0 = 1..
your theory is right.
What is wrong is to expect an exact value without taking careful measurements.
Here's a link on sensors - i've never used these particular ones either.http://www.gasalarmsystems.co.uk/NewsDetail.aspx?referrer=newsPage&id=15
For your application it might be advantageous to put the sensor in a bridge circuit and measure the unbalance with your PIC..
That way you could "null" the bridge when sensor is installed rather than storing R0 in software... Note that's what they did in the pdf drawing - sensor and R1 are one leg of bridge, R4, R5, and RP are the other.
by the way - that link to a jpg file locked up my computer solid and i had to turn power off. Have you had any trouble with that site?