You can't or shouldn't take raw 14bit value straight into an equation, since the bit 0= .25C, bit 1 = .5C, bit 2 =1C, bit 3=2C. You could easily divide that raw value by 4 and get the correct temp, maybe losing the lowest 2 bits but that's ok. Is the Thermo direct to the PCB or are there wires inserted(gauge?). There are a few caveats on the datasheet.
You can't or shouldn't take raw 14bit value straight into an equation, since the LSB = .25C, bit 2 = .5C, bit 3 =1C, bit 4=2C.
I figured that out. I use excel for large equations so I can see each variable. I will just only worry about the upper 12bits. That would give me
raw MAX31855 12bit value = 100 (100°C)
Corrected MAX31855 value = 131
Is the Thermo direct to the PCB or are there wires inserted(gauge?). There are a few caveats on the datasheet.
The board I have is made by Maxim. It includes a Omega thermocouple connector
. The thermocouple uses the mating connector, with no extension wire. Online, I was seeing MAX31855 breakout boards for around $15. These boards were around $20. I figured Maxim would design A better product then the other breakout boards I was seeing. That is why I bought them.
No problem. A lot of instruments use curve fitting equations like cubic spline, quadratic, etc. to convert readings from some sort of sensor to the units they want. That maxim chip has some type of processor in it to convert the current reading to degrees using an equation somewhat like the one I sent you.
At work, we have different kinds of thermocouples, (probes, thick wire, thin wire, high speed, etc..) I will grab a few and test them out and see if they make any difference. At work we only use Omega K type thermocouples.
Thank you for the "third order cubic spline" I had no idea equations like that existed. Like I said, math is not one of my strong points. I am assuming the equation is the raw (14bit) value from the MAX31855. I will make a "DAT" table that will convert the temps.
No problem. A lot of instruments use curve fitting equations like cubic spline, quadratic, etc. to convert readings from some sort of sensor to the units they want. That maxim chip has some type of processor in it to convert the current reading to degrees using an equation somewhat like the one I sent you.
Thank you for the PDF, there is a lot of good information in there.
kwinn - where did those cubic fit values come from?
I second that question. I thought of redoing the equation for the coefficients from NIST, and see what happens. I dont know if it would work or not, but cant hurt to try.
Tomorrow I am going to see if I can barrow our Omega thermocouple tester, to confirm the exact temperature.
To not create cycling due to temperature overshoots due to the time the temperature goes through air to reach the sensor + delay you have in between reading sensor etc,
When you are getting close to targeted temperature I think you have to start using a table of pwm values to use based on current temperature and how much you want to change it.
It's like turning off the garden hose before the bucket is full because you know how much water will still come while you turning the knob,
something a human learns after a few try but computer's are not so good at unless your program in fussy logic so it learns by its mistake.
To not create cycling due to temperature overshoots due to the time the temperature goes through air to reach the sensor + delay you have in between reading sensor etc,
When you are getting close to targeted temperature I think you have to start using a table of pwm values to use based on current temperature and how much you want to change it.
I was thinking of doing that for the heating slope (2°C/s). But I am still trying to get the temperature to be correct. The thermocouples are less than 1in from each other.
I could deal with that error. But I still cant find out what my problem is yet. I am still thinking it is the cold junction in the chip. lets say (just an example) the Fluke is reading 150°C, and my oven is reading a lower temp. If I place my finger on the chip (add heat) I can get the oven to match the Fluke. But that's not the way it should work.
Kwinn's calculation is off. I had to subtract the cold junction temp from the thermocouple temp, then add 4 just to get close
Is the chip near the oven to have it's own temperature affected when the oven heats up?
It is about 10in away from the oven. I have not seen the cold junction temp go up. That is why I did not think it was a concern. Plus, I will have to account for it somehow, everything will be placed inside a plastic box that is hooked to the side of the oven. But that is after all the bugs are worked out.
If the temperatures is affected by the proximity to the oven, that is not ideal.
Avoid placing heat-generating
devices or components near the MAX31855 because this
could produce cold-junction-related errors.
Because of the small signal levels involved, thermocou
-
ple temperature measurement is susceptible to power-
supply coupled noise. The effects of power-supply noise
can be minimized by placing a 0.1
F
F ceramic bypass
capacitor close to the V
CC
pin of the device and to GND.
The input amplifier is a low-noise amplifier designed to
enable high-precision input sensing. Keep the thermo
-
couple and connecting wires away from electrical noise
sources. It is strongly recommended to add a 10nF
ceramic surface-mount differential capacitor, placed
across the T+ and T- pins, in order to filter noise on the
thermocouple line.
kwinn - where did those cubic fit values come from?
Thanks...
-Russ
They were calculated using the readings from the Fluke meter and the oven controller by a curve fit program normally used to produce the calibration curves for an optical spectrometer. It does a maximum of 100 iterations and selects the curve constants with the lowest overall absolute error.
They were calculated using the readings from the Fluke meter and the oven controller by a curve fit program normally used to produce the calibration curves for an optical spectrometer. It does a maximum of 100 iterations and selects the curve constants with the lowest overall absolute error.
I am sorry kwinn, I was unaware that you would use the Fluke as a reference. I have had the meter for a couple years now, and the meter was right next to the oven. So the cold junction in the Fluke could've been skewed a little. I am going to barrow a thermocouple tester from work. The unit I want to barrow is our brand new Omega thermocouple thermometer. I know that unit is correct, it has been calibrated until next year. That way I can provide you an accurate temp comparison.
What I will also do is, start the oven at room temp. Then raise it to 50°C, leave it there for 5 or so minutes, so the temp can stabilize. Then take a reading with the Omega unit, and compare the readings. And continue with the same pattern, increasing the temp by 50°C. Or do you think I should do 25°C per test pattern? I am just trying to provide the most information I can for everyone. Because you all have been a great help with this.
When I was working with a MAX6675 (5V version) I had a hard time getting stable readings, until I added some filtering caps. There was a VFD running nearby and it probably generated a lot of noise.
Like this:
After getting things to settle down, temps matched good enough between the MAX6675 and the Omega unit - maybe 3 or 4C, if I recall correctly. I was running at about 160C.
When I was working with a MAX6675 (5V version) I had a hard time getting stable readings, until I added some filtering caps. There was a VFD running nearby and it probably generated a lot of noise.
Like this:
After getting things to settle down, temps matched good enough between the MAX6675 and the Omega unit - maybe 3 or 4C, if I recall correctly. I was running at about 160C.
-Russ
Close, Like this
I thought of noise, I have caps out the nose on my breadboard.
I have the list of temps from the oven and the temps from the Omega unit. I put another thermocouple next to the MAX31855 chips to confirm the cold junction temps. I did cycles from 25C(oven was at 26C, program didn't run) to 250C. There was a 5 minute hold for each set temp. I did that to get everything in the oven to saturate.
No worries. Start at room temp, go to 50 as you described, then go in steps of 25. I went from 27.8 to 40.6 and then in steps of approximately 20 for the next 13 steps. Steps of 25 will be more than enough for generating a good curve fit.
I am sorry kwinn, I was unaware that you would use the Fluke as a reference. I have had the meter for a couple years now, and the meter was right next to the oven. So the cold junction in the Fluke could've been skewed a little. I am going to barrow a thermocouple tester from work. The unit I want to barrow is our brand new Omega thermocouple thermometer. I know that unit is correct, it has been calibrated until next year. That way I can provide you an accurate temp comparison.
What I will also do is, start the oven at room temp. Then raise it to 50°C, leave it there for 5 or so minutes, so the temp can stabilize. Then take a reading with the Omega unit, and compare the readings. And continue with the same pattern, increasing the temp by 50°C. Or do you think I should do 25°C per test pattern? I am just trying to provide the most information I can for everyone. Because you all have been a great help with this.
No worries. Start at room temp, go to 50 as you described, then go in steps of 25. I went from 27.8 to 40.6 and then in steps of approximately 20 for the next 13 steps. Steps of 25 will be more than enough for generating a good curve fit.
I dont know if you seen it but here is the temp list I got.
Can you post a pic of the breadboard connected to the modules? Where is the power coming from to the module? Is the ground and power to the module routed through the breadboard?
While you can certainly analyze the "error" and compensate in software, it just doesn't feel quite right. (Excel works pretty well for curve fitting...)
Unless the readings will be stable and repeatable, then compensation will fall apart.
Can you post a pic of the breadboard connected to the modules? Where is the power coming from to the module? Is the ground and power to the module routed through the breadboard?
Sure.
With the MAX31855 modules in place.
With the modules removed
Let me know if you see anything because I cant see anything that could be creating the error.
While you can certainly analyze the "error" and compensate in software, it just doesn't feel quite right. (Excel works pretty well for curve fitting...)
I agree with you there. But I am getting tired of chasing the error.
Unless the readings will be stable and repeatable, then compensation will fall apart.
I do agree with you there. I would be so mad t do all this work, have a board made up, and find out I did something wrong now and have to fix what I have done.
I do like the signal conditioning of the PDF, Thank you
A breadboard is always suspect. Although the data seems to be transmitting fine, I would be concerned about grounding, noise, bad connections. You need .1uf caps near at the power on each row on the breadboard, where the jumpers run from the bus upwards.
A breadboard is always suspect. Although the data seems to be transmitting fine, I would be concerned about grounding, noise, bad connections. You need .1uf caps near at the power on each row on the breadboard, where the jumpers run from the bus upwards.
I didn't place .1uF caps on the breadboard because they are on the modules next to the chips. I only assumed that would be ok to help with noise for the chips.
I would solder in a direct connect from the module's VSS to the power supply GND. I would put the blame on some issue with ground. breadboard connections, power supply.
Here are the new corrections using the Omega data. Since the "d" portion of the equation is 0 you do not need to include that part of the equation in the program.
BTW I think tranger and T Chap have a point. When you go to a PCB these problems may go away. When you get to that point it would be a good time to repeat the temperature comparison.
I would solder in a direct connect from the module's VSS to the power supply GND. I would put the blame on some issue with ground. breadboard connections, power supply.
Here are the new corrections using the Omega data. Since the "d" portion of the equation is 0 you do not need to include that part of the equation in the program.
BTW I think tranger and T Chap have a point. When you go to a PCB these problems may go away. When you get to that point it would be a good time to repeat the temperature comparison.
Thank you so much kwinn, I will try them out when I get off work. After I try giving each module its own ground wire.
I guess you could say similar. I did not make the modules, I bought them just the way they are. Maxim designed and manufactured them. I bought them from DigiKey. DigiKey's order number MAX31855PMB1#-ND
have a 31855 on a sparkfun breakout board and am tempted to do some testing.... But, got to go out-of-town for a while. Maybe when I get back?
Good luck...
I first went that route, made my own board and everything
But I was not happy with it. That is when I decided just to buy the ones from DigiKey.
Yeah, I like the looks of those little boards and the price isn't bad. What appeared to be identical between the board I was able to find and the one you have is the conditioning between the reference junction (t/c connection) and the chip. However, I see that C1 and C4 are not populated. Wonder why? These are caps between each leg of the t/c and gnd. It looks like there are solder pads on the backside. Somebody could add those if they wanted.
That Analog doc I referred to in an earlier post also shows a 1Meg resistor between t/c (-) and gnd.
Comments
I figured that out. I use excel for large equations so I can see each variable. I will just only worry about the upper 12bits. That would give me
raw MAX31855 12bit value = 100 (100°C)
Corrected MAX31855 value = 131
The board I have is made by Maxim. It includes a Omega thermocouple connector
Reading the Maxim datasheet, it looks like it uses a straight-up linear fit @ 41.276 microvolts / C. http://datasheets.maximintegrated.com/en/ds/MAX31855.pdf
It doesn't look too bad compared to an eighth-order fit found in this Omega doc. http://www.omega.com/temperature/z/pdf/z021-032.pdf
At 250C the difference is about 4C.
kwinn - where did those cubic fit values come from?
Thanks...
-Russ
How does that work? I thought thermocouples were not linear.
Thank you for the PDF, there is a lot of good information in there.
I second that question. I thought of redoing the equation for the coefficients from NIST, and see what happens. I dont know if it would work or not, but cant hurt to try.
Tomorrow I am going to see if I can barrow our Omega thermocouple tester, to confirm the exact temperature.
When you are getting close to targeted temperature I think you have to start using a table of pwm values to use based on current temperature and how much you want to change it.
It's like turning off the garden hose before the bucket is full because you know how much water will still come while you turning the knob,
something a human learns after a few try but computer's are not so good at unless your program in fussy logic so it learns by its mistake.
You're right, they aren't. But, a K type is really close and many times linear is close enough (and very simple in comparison).
I was playing with a spreadsheet showing the "rocket-science" polynomial (8th order) versus linear. See attached....
31855-TC-Info2.xls
-Russ
I was thinking of doing that for the heating slope (2°C/s). But I am still trying to get the temperature to be correct. The thermocouples are less than 1in from each other.
I could deal with that error. But I still cant find out what my problem is yet. I am still thinking it is the cold junction in the chip. lets say (just an example) the Fluke is reading 150°C, and my oven is reading a lower temp. If I place my finger on the chip (add heat) I can get the oven to match the Fluke. But that's not the way it should work.
Kwinn's calculation is off. I had to subtract the cold junction temp from the thermocouple temp, then add 4 just to get close
long[TempPtr][idx] := TempCorrection[ (X~>(32-12))- (X << 16) ~> (32-8)+4] DAT TempCorrection word 0, 0, 0, 0, 0, 2, 3, 4, 5, 7, 8, 9, 11, 12, 13, 15 word 16, 17, 19, 20, 21, 23, 24, 25, 27, 28, 30, 31, 32, 34, 35, 36 word 38, 39, 40, 42, 43, 44, 46, 47, 49, 50, 51, 53, 54, 55, 57, 58 word 60, 61, 62, 64, 65, 66, 68, 69, 71, 72, 73, 75, 76, 77, 79, 80 word 82, 83, 84, 86, 87, 88, 90, 91, 93, 94, 95, 97, 98, 99, 101, 102 word 104, 105, 106, 108, 109, 110, 112, 113, 115, 116, 117, 119, 120, 121, 123, 124 word 126, 127, 128, 130, 131, 132, 134, 135, 136, 138, 139, 140, 142, 143, 145, 146 word 147, 149, 150, 151, 153, 154, 155, 157, 158, 159, 161, 162, 163, 165, 166, 167 word 168, 170, 171, 172, 174, 175, 176, 178, 179, 180, 181, 183, 184, 185, 187, 188 word 189, 190, 192, 193, 194, 195, 197, 198, 199, 201, 202, 203, 204, 205, 207, 208 word 209, 210, 212, 213, 214, 215, 216, 218, 219, 220, 221, 222, 224, 225, 226, 227 word 228, 229, 231, 232, 233, 234, 235, 236, 237, 239, 240, 241, 242, 243, 244, 245 word 246, 247, 249, 250, 251, 252, 253, 254, 255, 256, 257, 258, 259, 260, 261, 262 word 263, 264, 265, 266, 267, 268, 269, 270, 271, 272, 273, 274, 275, 276, 277, 278 word 279, 280, 281, 282, 283, 284, 284, 285, 286, 287, 288, 289, 290, 291, 291, 292 word 293, 294, 295, 296, 296, 297, 298, 299, 300, 300, 301, 302, 303, 304, 304, 305 word 306, 307, 307, 308, 309, 309, 310, 311, 311, 312, 313, 313, 314, 315, 315, 316 word 317, 317, 318, 318, 319, 320, 320, 321, 321, 322, 322, 323, 324, 324, 325, 325 word 326, 326, 327, 327, 328, 328, 328, 329, 329, 330, 330, 331, 331, 331, 332, 332It is about 10in away from the oven. I have not seen the cold junction temp go up. That is why I did not think it was a concern. Plus, I will have to account for it somehow, everything will be placed inside a plastic box that is hooked to the side of the oven. But that is after all the bugs are worked out.
They were calculated using the readings from the Fluke meter and the oven controller by a curve fit program normally used to produce the calibration curves for an optical spectrometer. It does a maximum of 100 iterations and selects the curve constants with the lowest overall absolute error.
I am sorry kwinn, I was unaware that you would use the Fluke as a reference. I have had the meter for a couple years now, and the meter was right next to the oven. So the cold junction in the Fluke could've been skewed a little. I am going to barrow a thermocouple tester from work. The unit I want to barrow is our brand new Omega thermocouple thermometer. I know that unit is correct, it has been calibrated until next year. That way I can provide you an accurate temp comparison.
What I will also do is, start the oven at room temp. Then raise it to 50°C, leave it there for 5 or so minutes, so the temp can stabilize. Then take a reading with the Omega unit, and compare the readings. And continue with the same pattern, increasing the temp by 50°C. Or do you think I should do 25°C per test pattern? I am just trying to provide the most information I can for everyone. Because you all have been a great help with this.
You mean like this?
When I was working with a MAX6675 (5V version) I had a hard time getting stable readings, until I added some filtering caps. There was a VFD running nearby and it probably generated a lot of noise.
Like this:
After getting things to settle down, temps matched good enough between the MAX6675 and the Omega unit - maybe 3 or 4C, if I recall correctly. I was running at about 160C.
-Russ
Close, Like this
I thought of noise, I have caps out the nose on my breadboard.
I have the list of temps from the oven and the temps from the Omega unit. I put another thermocouple next to the MAX31855 chips to confirm the cold junction temps. I did cycles from 25C(oven was at 26C, program didn't run) to 250C. There was a 5 minute hold for each set temp. I did that to get everything in the oven to saturate.
I dont know if you seen it but here is the temp list I got.
Unless the readings will be stable and repeatable, then compensation will fall apart.
This http://www.analog.com/library/analogdialogue/archives/44-10/thermocouple.pdf is an interesting read on t/c's and signal conditioning.
-Russ
Sure.
With the MAX31855 modules in place.
With the modules removed
Let me know if you see anything because I cant see anything that could be creating the error.
I agree with you there. But I am getting tired of chasing the error.
I do agree with you there. I would be so mad t do all this work, have a board made up, and find out I did something wrong now and have to fix what I have done.
I do like the signal conditioning of the PDF, Thank you
No problem, there is no hurry.
I didn't place .1uF caps on the breadboard because they are on the modules next to the chips. I only assumed that would be ok to help with noise for the chips.
BTW I think tranger and T Chap have a point. When you go to a PCB these problems may go away. When you get to that point it would be a good time to repeat the temperature comparison.
I have a 31855 on a sparkfun breakout board and am tempted to do some testing.... But, got to go out-of-town for a while. Maybe when I get back?
Good luck...
Yea I guess that couldn't hurt to try it.
Thank you so much kwinn, I will try them out when I get off work. After I try giving each module its own ground wire.
I guess you could say similar. I did not make the modules, I bought them just the way they are. Maxim designed and manufactured them. I bought them from DigiKey. DigiKey's order number MAX31855PMB1#-ND
I first went that route, made my own board and everything
But I was not happy with it. That is when I decided just to buy the ones from DigiKey.
That Analog doc I referred to in an earlier post also shows a 1Meg resistor between t/c (-) and gnd.