| View previous topic :: View next topic |
| Author |
Message |
georpo
Joined: 18 Nov 2008
Posts: 281
Location: Athens, Greece.
|
| KTY84 thermometer |
 Posted: Wed Feb 24, 2010 2:17 pm |
 |
|
Hi all!
I am trying to make a digital thermometer with a KTY84 silicon sensor.
I have this from the datasheet:
and this is my hardware:
The Vref- is at 1volt and the Vref+ is at 3.3v. This way I ignore temperatures below 0 degrees and also I have better resolution.
In the pdf I have some given ohm values for known temperatures in steps of 10 as you see above. The problem is that I do not know how to calculate the temperature for these values and also for the values between them. Unfortunatelly the sensor output is not linear.
Any help appreciated 
Last edited by georpo on Fri Feb 26, 2010 10:15 am; edited 3 times in total |
|
 |
PCM programmer
Joined: 06 Sep 2003
Posts: 21708
|
|
| Posted: Wed Feb 24, 2010 2:31 pm |
|
|
Why did this require a new thread ? Why not post this question on to
the end of your existing KTY84 thread ? |
|
|
georpo
Joined: 18 Nov 2008
Posts: 281
Location: Athens, Greece.
|
|
| Posted: Wed Feb 24, 2010 3:41 pm |
|
|
there is no other KTY84 thread!
I gave more details here in case somebody could help. There is nothing wrong I guess... |
|
|
PCM programmer
Joined: 06 Sep 2003
Posts: 21708
|
|
| Posted: Wed Feb 24, 2010 3:46 pm |
|
|
| Quote: |
there is no other KTY84 thread!
|
Look at the start of your thread here:
http://www.ccsinfo.com/forum/viewtopic.php?t=41797
It says:
| Quote: |
I am using a KTY84 temperature sensor to make a thermometer.
This changes its resistance according to the temperature.
|
|
|
|
georpo
Joined: 18 Nov 2008
Posts: 281
Location: Athens, Greece.
|
|
| Posted: Wed Feb 24, 2010 4:12 pm |
|
|
mmm I guess you are right. I already mentioned this
Anyway the point here is to find the solution. |
|
|
PCM programmer
Joined: 06 Sep 2003
Posts: 21708
|
|
| Posted: Wed Feb 24, 2010 6:16 pm |
|
|
The polynomial for the KTY84/130 is given in Figure 9-19 on page 211
of this document:
http://www.boschrexroth-us.com/country_units/america/united_states/en/products/brc/documentation_downloads/ProductDocumentation/CurrentProducts/Motors/IndraDyn_A/29578103.pdf
It says this:
| Quote: |
Tw = (A * R^3) + (B * R^2) + (C * R) + D
Tw: Winding temperature of the motor in °C
R: Resistance of the temperature sensor in Ohms
A: 3.039 * 10E-8
B: -1.44 * 10E-4
C: 0.358
D: -143.78
Fig. 9-19: Polynomial used for determining the temperature with a known
sensor resistance (KTY84)
|
Take that equation, make it into a test program, and run it for a sample
value of 1000 ohms. This gives a result of 100.6 degrees C, which is
reasonably accurate.
| Code: |
#include <18F452.h>
#fuses XT,NOWDT,PUT,BROWNOUT,NOLVP
#use delay(clock=4000000)
#use rs232(baud=9600, xmit=PIN_C6, rcv=PIN_C7, ERRORS)
float get_kty_temp(float res)
{
float temp_C;
temp_C = (3.039E-8 * res * res * res) + (-1.44E-4 * res * res) + (0.358 * res) - 143.78;
return(temp_C);
}
//======================================
void main(void)
{
float result;
result = get_kty_temp(1000); // Use 1000 ohms as a test
printf("%5.2f \r", result);
while(1);
}
|
|
|
|
georpo
Joined: 18 Nov 2008
Posts: 281
Location: Athens, Greece.
|
|
| Posted: Thu Feb 25, 2010 8:56 am |
|
|
Thank you so much pcm programmer. You are the best!
But please tell me how did you find this?
I have been searching for more than a week for an example like this.
After a few tests I found that it is not accurate...
for example the result for 250 degrees (2166 ohms) is 264.88 degrees.
14.88 degrees error  |
|
|
jaimechacoff
Joined: 14 Feb 2010
Posts: 24
Location: Santiago, Chile
|
|
|
FvM
Joined: 27 Aug 2008
Posts: 2337
Location: Germany
|
|
| Posted: Thu Feb 25, 2010 9:40 am |
|
|
| Quote: | | After a few tests I found that it is not accurate... |
Yes, you only need a pocket calculator to see this. The reason is simple, the polynominal was fitted for a 0 .. 170 °C
range, as with any polynominal, the error is increasing rapidly outside the intended range.
I suggest to use a spreadsheet program (e.g. MS Excel) and fit your own polynominal to the resistance table from
KTY84 datasheet. When you say 264°C isn't accurate, you may want to consider, that the possible error resulting
from KTY84/130 tolerance is larger than 14° at 250 °C. So if you are targetting to accuracy, you should think
about an industry standard thermocouple or Pt100/Pt1000 sensor. |
|
|
PCM programmer
Joined: 06 Sep 2003
Posts: 21708
|
|
| Posted: Thu Feb 25, 2010 12:17 pm |
|
|
| Quote: |
But please tell me how did you find this?
I have been searching for more than a week for an example like this. |
I just typed this into Google:
|
|
|
georpo
Joined: 18 Nov 2008
Posts: 281
Location: Athens, Greece.
|
|
| Posted: Thu Feb 25, 2010 12:34 pm |
|
|
The thermometer is for a home made reflow oven and the max temperature is about 250 degrees so LM35 is out of the question.
The sensor is not self heating because the test was done only in software with example resistor values as in pcm programmer's example.
The current through the sensor is about 1.65mA.
As for thermocouples, they need amplification. right?
I could do some tests with my multimeter's thermocouple but I found the KTY84 easier because the values can be measured by the ADC without extra hardware. |
|
|
jaimechacoff
Joined: 14 Feb 2010
Posts: 24
Location: Santiago, Chile
|
|
| Posted: Thu Feb 25, 2010 1:17 pm |
|
|
upload a TXT file or an Excel file with all the data you want to fit and i will do it for you.
but, I can't have it for today ... I am also in troubles with some software hahahaha : ) |
|
|
dyeatman
Joined: 06 Sep 2003
Posts: 1934
Location: Norman, OK
|
|
| Posted: Thu Feb 25, 2010 1:34 pm |
|
|
Have you considered something like a DS1820 in the TO92 case?
It goes up to 250 degrees F and would give you much better accuracy and eliminate all the error calculations.
_________________
Google and Forum Search are some of your best tools!!!! |
|
|
FvM
Joined: 27 Aug 2008
Posts: 2337
Location: Germany
|
|
| Posted: Thu Feb 25, 2010 3:27 pm |
|
|
The temperature in question was 250°C rather than 250°F. Clearly you can't use any Si based IC at these temperatures.
Si PTCs as KTY are fine, I use them since decades. I just wanted to mention their limited accuracy in wide
range operation, as indicated by the datasheet. But it's most likely sufficient for your purposes.
Up to 400 °C, platinum resistance thermometers are mostly preferred to thermocouples, except for applications,
where small size matters. They also don't necessarily need amplification, but the t.c. is lower, so have possibly an
unsufficient resolution with a 10-Bit ADC. You can say, that a KTY sensor fits the accuracy and price of a 10-Bit ADC
much better than a Pt1000 sensor, which rather demands a 12 to 16-Bit ADC.
If you use the KTY84 in an extended temperature range, you should observe datasheet Fig. 3 "Sensor resistance
as a function of ambient temperature and operating current", particularly the polarity sensitivity.
P.S.: A final remark. As said, fitting your own polynominal for a specified range is easy. But you can also calculate
a classical "rule of three" table interpolation with the exact data sheet resistance table.
P.P.S: MS. Excel calculated the below 3rd order filt for the 0..250°C range:
| Code: | | y = 0.0000000174598x3 - 0.0001046520x2 + 0.323308x - 135.9886 |
|
|
|
georpo
Joined: 18 Nov 2008
Posts: 281
Location: Athens, Greece.
|
|
| Posted: Thu Feb 25, 2010 6:18 pm |
|
|
Thanks everybody for the support!
For now I use a table with the fixed values from the datasheet as a
reference and use a "rule of three" to find the values between them, the
result is more than 5°C error for temperatures over 100 °C.
What would be the best method to turn the ADC value directly to °C?
I use this:
Vref+ =3.3v
Vref- =1v
So, the resolution is 2.3v / 1023 = 0.002248 volts / ADC step
The pullup resistor of the sensor to 3.3v is 1Kohm
since Vout= (R2/(R1+R2))*Vin
for a 1000 ohm sensor value I have :
(1000/2000)*3.3 = 1.65 Volts - 1volt (Vref-) = 0.65volts
0.65volts / 0.002248 = 289.145. This is the ADC reading for 1000 ohm sensor value.
This method works, but I am sure there is a much more simple way to do it.
FvM, please tell me, what are the X and Y values in your Excel formula?
y= resistor and x= temperature? |
|
|
|
FAQ 
Search
Register
Profile 
Log in to check your private messages 
Log in
