My friend is using a MAX6675 thermocouple amplifier but is complaining of a drift error in his output over time.
(12-bit output representing up to 1024 Celsius)

I checked the datasheet and it states an expected error of around 4-8 LSB's which amounts to 1-2 Celsius given the resolution.

I was wondering how you could go about reducing the usable input range to the amplifier to ensure that the 12-bit output represents 0-300 celsius (his useable temperature range) which would improve the resolution to 0.07 c and minimize the effect of this error.

Any other solutions are welcome!

Thanks AAC

 

How is the thermal bond between the chip and the input terminals done?
Can you show your schematic and some pictures of the device?

 

How is the thermal bond between the chip and the input terminals done?
Can you show your schematic and some pictures of the device?

Its actually a friend of mine from another forum, i am not sure about the thermal bond between the chip and input terminals, but given the datasheet information and resolution they match his errors.
So my question is how to reduce the input range to improve the output resolution.
THanks

 

I don´t think that it is possible to change the range. You could try getting some other better suited chip.

 

I don´t think that it is possible to change the range. You could try getting some other better suited chip.

i Though that may be the case!
Thanks

 

Can anyone answer my question please..?, can i connect max31855 with just thermocouple and volt meter without an arduino microcontroller, if there's possible, can i get a schematic for my homemade digital heater rework station?

 

My friend is using a MAX6675 thermocouple amplifier but is complaining of a drift error in his output over time.
(12-bit output representing up to 1024 Celsius)

I checked the datasheet and it states an expected error of around 4-8 LSB's which amounts to 1-2 Celsius given the resolution.

I was wondering how you could go about reducing the usable input range to the amplifier to ensure that the 12-bit output represents 0-300 celsius (his useable temperature range) which would improve the resolution to 0.07 c and minimize the effect of this error.

Any other solutions are welcome!

Thanks AAC

OK, to get the best resolution, the temperature range must equal the input range of the A/D converter. So you need amplification. Using the thermocouple tables, you can find the TC output at the range minimum and maximum, that will give you the input span to the amplifier. Next, you need to know the A/D input voltage range, that will give you the required output voltage from the amplifier. Then the hard math: divide the output span by the input span, that number will be the required amplifier gain. Now the question is if the MAX6675 thermocouple amplifier can be adjusted to provide that gain. If not, a different amplifier is needed.

 

The MAX 6675 is designed for use with a Type K thermocouple. While the Type K thermocouple does have a fairly linear region it also has limits of error:
Limits of Error for Thermocouple Wire ; K · 0 to 1250°C · 32 to 2202°F · ± 2.2°C · ± 4°F · or ± .75%. ± 1.1°C · ± 2°F ; R, S · 0 to 1450°C · 32 to 2642°F · ± 1.5°C · ± 3°F.

So you have the limits of error of the MAX 6675 along with the limits of error of the thermocouple to algebraically add for the allowable error. For accurate thermocouples they are purchased calibrated. When I was buying Type K for example I would request calibration over 5 check points of 0, 200, 400, 600, 800 and 1,000 degrees F and I would request special limits wire be used in their manufacture. They came in with charted results at my requested points. They also carry a price tag.

Your friend can make a simple ice bath slurry and see what he gets at 0 deg C and 32 deg F. Then depending on altitude and barometric pressure at his location boil some distilled water and place the thermocouple in the water bath and he should see about 100 deg C and 212 deg F depending on location. Sort of a poor man's calibration.

While it won't improve resolution what your friend can try is averaging 10 or more readings in his code. Just remember you can't make a silk purse out of a sows ear. Remember also resolution is the ability to read an instrument or of an instrument to be read, accuracy is unbiased precision.

Edit
I should also have mentioned if your friend wants accurate scalable measurements of temperature using a Type K or about any other thermocouple he or she should find a true temperature transmitter.


These allow a user to program a span for example 0 to 300 degrees C and they output 4 to 20 mA so you place a resistor in the 4 - 20 mA loop and measure the subsequent voltage drop. In this case I can use a 250 Ohm precision resistor and get 0 to 5 volts. The rest is just a matter of code for the A/D conversion.



Ron

 

Great, I just replied to a 2015 thread.

Can anyone answer my question please..?, can i connect max31855 with just thermocouple and volt meter without an arduino microcontroller, if there's possible, can i get a schematic for my homemade digital heater rework station?

Start your own thread, please do not dig up an old thread and add to it. The original thread is seven years old.

Ron

 

Great, I just replied to a 2015 thread.
Start your own thread, please do not dig up an old thread and add to it. The original thread is seven years old.

Ron

So did I, RR. "They" fooled us again! You approached from a calibration point of view while I approached from a maximizing resolution point of view. Evidently we think differently.

 

So did I, RR. "They" fooled us again! You approached from a calibration point of view while I approached from a maximizing resolution point of view. Evidently we think differently.

Can't believe I did that. Yet, did that again. Now for several post I'll note the date and then I get complacent again.

Yes, not just calibration, people need to get a handle on allowable limits of error on a thermocouple and then understand that error is cumulative starting with a sensor and then the rest of the system. I have no clue if the thermocouples off the boat even comply with what I posted. I do know that good thermocouples comply with ANSI and IEC. I really liked the little temperature transmitters I posted and have no clue what they cost today. Totally programmable for span and output. That allowed for maximum resolution based on span. About 10 ~ 15 years ago those transmitters went for about $70 each when we bought 10 or more. Been retired 10 years now so no clue what is the newest and best today.

Not so much a matter of thinking differently as a matter of covering all the bases. Go figure? What annoys me is how people fail to read about how to post questions and just jump into an old thread. Then too I never caught the date so bad on me.

Ron

 

Some times I get a warning response that tells me this post is over nnn days old, do you still want to respond. Other times I don't. A simple and useful change would be a darker font for the "previous response" date. Surely changing a font is not a big deal. Not asking to do anything different, just a louder font.

 

Can anyone answer my question please..?, can i connect max31855 with just thermocouple and volt meter without an arduino microcontroller, if there's possible, can i get a schematic for my homemade digital heater rework station?

Some times I get a warning response that tells me this post is over nnn days old, do you still want to respond. Other times I don't. A simple and useful change would be a darker font for the "previous response" date. Surely changing a font is not a big deal. Not asking to do anything different, just a louder font.

The problem is the first quote. Once someone jumps in they see the warning, after that the thread just moves to the top of the pile. Unless we look at when the thread started here we are. In this instance it was post #6 who should just have started his/her own thread.

Ron