I have a question about the AD8495. Please take a look at the attachment.
I think it is not out of spec, but I would like to know the cause of these variation.
Yes, that equation is used to correct the nonlinearity and implementing it within the software will get the error down to +/-1.5mV, or less than +/-0.5degC over the 0 to 220degC temp range.
To get this equation, we took the data from the NIST thermocouple table, calculated the expected output voltage of the AD8495 for each temperature over the 0 to 220degC range, and took a 6th order polynomial fit of the data. The fit is not perfect, so there will be some error between the calculated output voltage from the equation and the output voltage from the table - in this case it is +/-1.5mV.
If I'm understanding correctly, it looks like you are using a precision source to simulate the thermocouple voltage at different temperatures.
Is your setup exactly how it is pictured with testing only one part at a time? Or are you testing all 8 units simultaneously? Are you able to replicate the 25degC data by just grounding the inputs of the AD8495? Have you tried switching out the DC input source?
Do you have a photo of your setup that you could upload?
Thank you for supporting this caes.
I tested all 8 units simultanously and have a question to you. Please refer to the followings.
I'll try to replicate the 25degC data by gounding the both inputs.
I can give you the polynomial coefficients to calculate the expected output voltage of the AD8495 for different temperatures and give you an estimate of the expected error. Can you tell me what kind of thermocouple/thermocouples you plan on using? Also, what is the temperature range that you are trying measure? From the first picture I can see the low end is about -25C and the high end is around 300C but I'd like to confirm. Is your reference temperature going to be 32C?
Can you tell me what kind of thermocouple/thermocouples you plan on using?
--> YC105SCK3.2S6U65WX1B3000/Yashimasokki.Co., Ltd. (But no English website.)
Type: K, Class: 2, Temperature Range: 0~300degC
Also, what is the temperature range that you are trying measure?
Is your reference temperature going to be 32C?
--> Our product's ambient temp spec is +5~+32degC, so I tested it at the max ambient temp.
For a measurement temperature range of 0 to 220degC, and using the equation Vout(mV) = a0 + a1*T+a2*T^2+a3*T^3+a4*T^4+a5*T^5+a6*T^6 where T is the measurement temperature:
The error from using this polynomial is about +/-1.5mV over the whole measurement range.
If you are still having issues after implementing this equation, I'd be interested to see the newest data in the plot that you originally sent.
Did you try swapping out the precision source? Do you have a way of measuring what is actually coming out of that source?
First, what cause the +/-1.5mV error? Is it quantization noise?
Second, my understanding about the total error from the AD8495 under the temperature range of 0 to 220 degC
is +/-0.5 degC (CN-0354 with nonlinearity correction, CN-0354 Figure 3.) and +/-1.5mV you calculated in the last mail. Am I correct?
Last, as I asked you before, I need the AD8495 curve with nonlinearity correction under the temperature range of -25degC to 0degC. Could you send it to me?
The +/-1.5mV error is the max error you would get if you took the actual output of the AD8495 and subtracted out the polynomial that is correcting for the nonlinearity. That is the error in mV, which would correspond to less than +/-0.5degC of error over the 0 to 220degC temp range. The results from the CN-0354 will not be the same as your application - it is using a larger temperature range so the polynomial and subsequent error is different.
Are you trying to measure down to -25degC? The curve with the nonlinearity correction will depend on the temperature range used to generate the polynomial. So if you are measuring down to -25degC, I can generate a polynomial to fit the expected data over the -25degC to 220degC range. If I just fit the region between -25degC and 0degC and you aren't going to measure there, it won't be very accurate to your application.
Have you been able to swap out the power supply you are using to generate the inputs? If you sweep the source over the range you are using for the AD8495, do you have a plot of what that looks like?
The equation Vout(mV)=a0+a1*T+a2*T^2~+a6*T^6 is used to correct the nonlinearity and should be implemented in the software?
I understand +/-1.5mV is the expected total error after correcting for the nonlinearity from the AD8495.
What causes the +/-1.5mV error?
Retrieving data ...