Thermocouple readings not correct


 
Gary Graham

Gary Graham

TVWBB Fan
I am using an external thermocouple amplifier (Adafruit AD8495 Breakout) and am reasonably sure it is working properly, but not seeing the correct temps in Heatermeter.
The thermocouple breakout board is using 3.3v as the supply voltage. I got same results using the 5v supply, and decided to try the 3.3v... same results.

Individual component testing:
a) The thermocouple sends the proper temperature when connected to a DVM style thermocouple reader, validating the thermocouple is not fried.
b) I did a minimum config test with the Rpi with a 5v usb power source and the thermocouple breakout board connected. The HM board was NOT connected...
Measured the output pins of the thermocouple breakout board to be 1.361v using a DVM.... Compared to room temperature reading using the thermocouple jack on the DVM, which read 25.2 degrees C. (25.2 * .005) = 1.376 v

I have the thermocouple in a bundle with a Thermoworks meat probe and a ET-732 pit probe next to the fan output of a laptop.
The Thermoworks and ET-732 are within a couple of degrees of each other and reading 101-102 degrees F.

Heatermeter board Setup:
Lifted the "bottom" connection (as you would look at the LCD) for the food2 and food3 10K pullup resistors. This leaves the resistor hanging, but still connected on the +vcc side.
I used the menu page to select the probe type to be Thermocouple, and used the default value of 5mV/C.
With nothing connected to food2 or food3, the HM reports 34.5 degrees

With the thermocouple amp connected to the food3 position of J2, the HM reports ~695 degrees F.

So I played around with the mV/C setting to see if I could get the TC to read like my Thermoworks probe or the ET-732 pit probe... Those are on pit and food1, and set up as such in the configuration page.

Noticed that when I set the mV/C to 100, the HM will report the thermocouple temp to be ~108.8 degrees, and when I change the mV/C to be 99.999, the HM reported temp drops drastically to 57.6 degrees, and when I go back to the configuration page, the number I see for the mV/C is actually 99.998998.

Really odd stuff. Am I doing something wrong here ?
 
The mV/C is the gain setting for your amplifier, so you need to set that to whatever gain the amp you are using has.....
 
OK, so at 5 mV/C he should be good to go...

Gary, how are you connecting the TC to the amp, and how are you connecting the amp output to the HM board?
 
(25.2 * .005) = 0.126V not 1.367V. 1.367V is 273C or 524F.

If the TC output 1.367V? If so, there's your problem. If it is actually 0.1367V then check it at the ADC input of the ATmega. The mV/C should be within a couple percent of 5. 5 should get you within a degree or two.

With food2 and food3 "floating" (not connected) they can read virtually anything so that's normal.

The 100=99.99899 is because the ATmega stores a single precision float number and that's as close as it can get to being 100. That said, put 5 in there.
 
Bryan Mayland said:
(25.2 * .005) = 0.126V not 1.367V. 1.367V is 273C or 524F.

If the TC output 1.367V? If so, there's your problem. If it is actually 0.1367V then check it at the ADC input of the ATmega. The mV/C should be within a couple percent of 5. 5 should get you within a degree or two.

With food2 and food3 "floating" (not connected) they can read virtually anything so that's normal.

The 100=99.99899 is because the ATmega stores a single precision float number and that's as close as it can get to being 100. That said, put 5 in there.
Click to expand...

Ok, so I think I missed one of the details in the calculation, I was reading 1.3v or so on the meter, and I missed the -1.25 offset they are showing on the board:
http://www.adafruit.com/images/230x173/1778-01.jpg

Here's how they get the temp out:
You can easily convert the voltage to temperature with the following equation: Temperature = (Vout - 1.25) / 0.005 V. So for example, if the voltage is 1.5VDC, the temperature is (1.5 - 1.25) / 0.005 = 50°C
Click to expand...

... so is having the 1.25 v offset the problem ?
 
Exactly. You can cut the trace to the bias input of the AD8495 and tie it to ground to remove the offset or build a circuit to subtract it back out. HeterMeter only supports +/-127 degrees of offset.

EDIT: On the datasheet it is called the reference pin (pin 2).
 
Last edited:
Thanks That was it.

The irony is that I got the Adafruit board so I would not have to do any smt soldering. Using an xacto knife, I was able to pry and lift pin 2 up in the air, and then solder a jumper form pin 2 to ground and everything seems to be working normal now.

I ran a shakedown test with a boiler on the gas stove, and found one flaky connection, and have it all tightened up for the big cook this weekend.

I am planning a 150 pound cook of brisket, butts and ribs in the cabinet for Saturday, and will try to make a nice cook log.
(only the 2 biggest meats of each type will get a probe though)

Thanks again for the help. The removal of the reference voltage did the trick.
 
Gary, sounds like a great alternative for SMD. My eyesight is terrible so i can't get it right.
how about some photo's and a short tutorial to get us goning
 
Tom Steed said:
Gary, sounds like a great alternative for SMD...
Click to expand...

Hi Tom,
The photos on the Adafruit web site show the size of the thermocouple amp board I am using... its about the size of a quarter. I noticed a more noisy signal after making the reference voltage modification to put the vref on pin 2 to ground, and the noise is pretty significant, enough to give the temps reported +-150 degrees (nearly 1v p-p)... Here's what it looked like, before RRD (the graphing tool) does the smoothing on the data points:

HOH_loins_1hr.png


Note the difference in the jagged edges on the thermocouple reading compared to the thermistors. (thermocouple is labeled as "amb", the others are thermistors)

Here is a cook log for the entire duration of the cook for comparison. I will give the play-by-play description on the Heatermeter cook logs page... ~170 lbs in my insulated cabinet shown below (no probes in meat, just monitoring the cabinet temps)

HM_cook3_HOH.png


Thermocouple related comments...
G) thermocouple temps cannot be right.. reseated TC, jiggled wires and eventually rebooted to recover.
H) no issue... just unplugged so I could pull the BBQ trailer to the event.
I-J) thermocouple temps did not look right again, so I rebooted and when HM came back they looked more normal.
 
Last edited:
You shouldn't be seeing so much swing on the thermocouple. It seems like something is definitely wrong. You sure it has a good connection? It should more stable with GND as the reference, not less stable.

Also you can click the little save icon on the top of that archive view page to turn the graph into a PNG and then right click "Save as..." it. Not sure if that's any more convenient than a screen shot though.
 
Tom Steed said:
Did you just a pit probe jack for the input to the HM from the amp board.
Click to expand...

Tom,
I have the Adafruit TC board connected to the HM board by wires connected in place of the jack. I using this channel to monitor high temperatures, and not planning to control the pit with the thermocouple, so I have it connected to the food3 probe (the connector closest to the top edge of the HM board).

I am using a case that has some room in one direction (in the "back") - the case is a 3"x5" index card holder. It is tight on the sides using the ralink WiFi dongle with antenna, but it fits... and there is plenty room to add more breakout boards behind the Rpi board, so I made a breakout board that holds the smaller TC board and the I2C-ADC board that plugs into the pins where the HM board connects to the RPi.

Bryan Mayland said:
You shouldn't be seeing so much swing on the thermocouple. It seems like something is definitely wrong. You sure it has a good connection? It should more stable with GND as the reference, not less stable.

Also you can click the little save icon on the top of that archive view page to turn the graph into a PNG and then right click "Save as..." it. Not sure if that's any more convenient than a screen shot though.
Click to expand...

Bryan,
Been taking it easy for a couple of days, and now packing up for a road trip to Alabama... I am doing an interstate move of the smoker, etc... and bringing my old school Tektronics scope. I will have to do some poking around and see what I can find. I should be able to pinpoint the source of the noise when I get my "rec room" set up again.

I like being able to load the graphs from a previous point in time because I can mouse over the time line and review the actual time and temps, although the probe names do not match what is stored in the Linkmeter configuration at the time the stash was made. The RRD package is an extremely cool and powerful aspect of HM, and I have been reading about how to put in data into a database for my "more meat probes" project.

Thanks again for the help.
 
Last edited:
Gary Graham said:
Tom,
I like being able to load the graphs from a previous point in time because I can mouse over the time line and review the actual time and temps, although the probe names do not match what is stored in the Linkmeter configuration at the time the stash was made.
Click to expand...
Yeah that's been something on the TODO list for a while but I don't know how many people actually use the archive so it hasn't even been a big priority to add. The probe names are only stored in the HeaterMeter itself so once you change them you lose the old values. I was planning on adding a metadata file that goes with the stash that allows you to store the probe names and some description or something but it seemed like something that wouldn't get a lot of use.
 
+1 for being able to save the probe names with the archives. Having the PID settings shown would also be handy.
 
You must log in or register to reply here.

 

Back
Top