Who is here? 1 guest(s)
 Print Thread
Thermocouple Woes
exer31337
I got everything built for an Arduino Controlled roaster, and everything including the thermocouples were working until they were hooked into the system. Once that happened I started getting erratic temp readings. I am assuming it has something to do with being grounded or interference from the AC fan etc... I tried using both the MAX6675, and MAX31855, same problem. Has anyone ran into this problem if so what was their fix? Also has anyone tried using the MAX31856 on fluid bed setups with success?

It is currently usable because I had a separate temp reader, but it would be very nice to have everything displayed onto Artisan.

Thank you for any information.

FYI I did try to ground the Arduino to earth ground and I tried that with the MAX chips themselves, no difference in readings.
 
renatoa
How erratic are the readings ?
The MAX chips you used have 0.25C degree resolution, so you will not able to read the degree decimal accurately.
An usual reading sequence looks like: 150.0, 150.3, 150.5, 150.7, 151...
With such values you can't accurate compute an usable RoR without an averaging window of at least 24 seconds.
 
exer31337
They went from 26c to 800+ with the only change being hooking them into the machine. I attempted to calibrate them but that failed as well, the calibration did not actually adjust as it should. I was using -1100 to get 800 to read around 26.
 
Piotrkurak
Check and Switch the wires on the TC's? If the control device uses common ground one backwards wired set will bollox up the whole mess.
 
randytsuch
I moved from a popcorn pumper to a SCTO type roaster a few months ago, so no longer using a AC fan.

From memory, and by seeing how it works in the SCTO, my AC fan generates a fair amount of noise. I rewired a few times to try to keep the fac AC wires away from everything else, which helped. But I never completely eliminated the noise.

IIRC, I would see spikes on the temp readings sometimes. Big jump, but then back to normal. But main problem was LCD would blank out or display garbage. I had to add code to reset it, but that was just a bandaid that didn't always work.

I already had given up trying to control the fan with an arduino. Just trying to control the fan with an arduino connected to the main ESP32 would really screw things up. Would cause resets and lose of communications. And LCD problem was much worse, more frequent.

With my SCTO, I never see any of these issues.

Randy
 
exer31337

Quote

Piotrkurak wrote:

Check and Switch the wires on the TC's? If the control device uses common ground one backwards wired set will bollox up the whole mess.


I've done that several times. (Switch wires)

I will be adding a small cap to try and limit noise.

Worse case I can get it to play back a roast using my prior roast heat and fan settings, but it would be nifty to have it graph temp for roast analysis
 
renatoa
The real cure to this is to playback using a power profile, not temperature, which is the most inconsistent thing in the whole process.
I mean measured temperature, not real temperature.

I wold diagnose the measuring chain starting with the electronics, mainly the measuring chips.
Build a fake probe, using a divider circuit, that outputs 6-8 mV, depending what temperature range you want to check, 6mV for 150 C, 8 mV for 200C. This can be done using a 1:6-800 rezistive divider, having 5V at input. For example 100 Ohms and 60-80 kO
Feed the TC input with this voltage and watch if it's stable.
You can try also connecting electronics ground to machine metal chassis, where the TC is mounted. If your measurement becomes crazy when you do this grounding, you have a mains voltage leaks in the machine, and the electronics have to be powered from a floating source, like a power bank, not with a mains adapter.
Same if you power the electronics via USB, from a laptop. Use laptop on battery during roasting, disconnected from the mains.
Surprisingly, grounding the machine do not help, at least didn't for me...

Another cure would be to switch to Pt sensors, much less sensitive to such issues than the TC.
 
ETomczak
I've been dealing with a similar issue with my roaster. If possible, try to find "ungrounded" or insulated thermocouples. The TC4 does not seem to like grounded ones. I spent a long time trying to ground everything in my roaster, which didn't work and I am now using bare thermocouples and just dealing with the noise in software. That is working ok so far.

Depending on how comfortable with it you are, you could try to disassemble your current TC and use a small amount of Teflon tape to insulate the probe from the sheath.

Or if you're already using ungrounded TCs then I have no idea. Best of luck!
 
Bhante

Quote

exer31337 wrote: I've done that several times. (Switch wires)

Switching several times raises warning lights!! There are only two possible connections, one correct and one incorrect, so you shouldn't have to swith back and forth several times unless you have serious confusion over your wiring? If the withing is correct, warming the probe tip with your finger should increase temperature, the opposite if it is wrong.

There are several issues that can arise from thermocouple measurement. The measured voltages are very tiny, and can easily be drowned by external signals.

1) What is the quality of the probe? Cheap low-quality probes may have an unreliable and intermittent ground connection between probe and sheath, and/or a connection where they are supposed to be insulated.

2) Do you have an elderly fridge or freezer or air-con in the neighbourhood? Using a TC4 I have found they are highly susceptible to radio interference, especially from a fridge turning on or off. An old fridge is hugely worse than a new one, although certain new ones can also cause problems sometimes. I have at times had problems with sudden program crashes due to this interference, causing the monitor program to hang. If you have this problem you may find looping the thermocouple wire through a ferric suppressor bead may help somewhat (one full turn), but the most effective solution is to buy a new fridge, or at least turn it off during the roast.

3) If you have a computer connected via USB cable you may have a ground loop problem between roaster power supply and computer power supply, so it's best to use the notebook with its power supply disconnected. A notebook power supply may cause a large degree of random noise on the RoR signal. Using a good quality probe with an insulated tip you shoud not have this problem (at least with the TC4). Conrad Electronics provide some excellent 1.5mm probes made by B+B Thermo which are very reliable, a few of which are quite low cost, and very well suited to coffee roasting. A separate power supply for the Arduino is potentially also a cause; if you are using a wireless link, try and power the arduino from the same supply as the electronics powering the roaster, which should reduce the risks of ground loops.

4) Putting a large resistor or small capacitor between each of the thermal contacts and ground can reduce grounding problems. Later versions of the TC4 have contacts for installing either resistors (version 4??) or capacitors (version 5??) but these are usually left unpopulated by default - I would recommend installing them. I think the capacitor is preferable to the resistor, but I cannot say that authoritatively.
 
renatoa
They should be both, else the filtering effect will be reduced colse to not existing.
Attached is a typical TC input filtering scheme, excerpted from a TC amplifier good practice/usage notes

Quote

RECOMMENDATIONS FOR BEST CIRCUIT PERFORMANCE
Input Filter
A low-pass filter before the input of the AD849x is strongly recommended (see Figure 29),
especially when operating in an electrically noisy environment.
Long thermocouple leads can function as an excellent antenna and pick up many unwanted signals.
The filter should be set to a low corner frequency that still allows the input signal to pass through undiminished.
The primary purpose of the filter is to remove RF signals, which, if allowed to reach the AD849x,
can be rectified and appear as temperature fluctuations.


~~~
renatoa attached the following image:
image_2023-07-10_192605734.png
 
renatoa
Useful tested values for component for you to start could be Cc = 10nF, Co = 0.1uF, R = 100 Ohms
 
exer31337

Quote

renatoa wrote:

The real cure to this is to playback using a power profile, not temperature, which is the most inconsistent thing in the whole process.


100% what I do. I just want to temp graphs to look for flaws in the roast.
 
exer31337

Quote

Bhante wrote:

Quote

exer31337 wrote: I've done that several times. (Switch wires)

Switching several times raises warning lights!! There are only two possible connections, one correct and one incorrect, so you shouldn't have to swith and forth several times unless you have serious confusion over your wiring? If the withing is correct, warming the probe tip with your finger should increase temperature, the opposite if it is wrong.

There are several issues that can arise from thermocouple measurement. The measured voltages are very tiny, and can easily be drowned by external signals.

1) What is the quality of the probe? Cheap low-quality probes may have an unreliable and intermittent ground connection between probe and sheath, and/or a connection where they are supposed to be insulated.

2) Do you have an elderly fridge or freezer or air-con in the neighbourhood? Using a TC4 I have found they are highly susceptible to radio interference, especially from a fridge turning on or off. An old fridge is hugely worse than a new one, although certain new ones can also cause problems sometimes. I have at times had problems with sudden program crashes due to this interference, causing the monitor program to hang. If you have this problem you may find looping the thermocouple wire through a ferric suppressor bead may help somewhat (one full turn), but the most effective solution is to buy a new fridge, or at least turn it off during the roast.

3) If you have a computer connected via USB cable you may have a ground loop problem between roaster power supply and computer power supply, so it's best to use the notebook with its power supply disconnected. A notebook power supply may cause a large degree of random noise on the RoR signal. Using a good quality probe with an insulated tip you shoud not have this problem (at least with the TC4). Conrad Electronics provide some excellent 1.5mm probes made by B+B Thermo which are very reliable, a few of which are quite low cost, and very well suited to coffee roasting. A separate power supply for the Arduino is potentially also a cause; if you are using a wireless link, try and power the arduino from the same supply as the electronics powering the roaster, which should reduce the risks of ground loops.

4) Putting a large resistor or small capacitor between each of the thermal contacts and ground can reduce grounding problems. Later versions of the TC4 have contacts for installing either resistors (version 4??) or capacitors (version 5??) but these are usually left unpopulated by default - I would recommend installing them. I think the capacitor is preferable to the resistor, but I cannot say that authoritatively.


Some confusion here I am not using a tc4, I am using a emulator. I run the program and thermocouples without issue outside of the machine. Put them in hot water they rise to the exact amount they should. As the issue only happens when plugged into the machine I am assuming it is a ground issue or a noise issue. Machine is 220v AC, Arduino is powered from a laptop and it happens when plugged in an unplugged.

Just purchased the MAX83556 hopeful it removes any grounds issue if the resistors don't work.
 
exer31337

Quote

renatoa wrote:

They should be both, else the filtering effect will be reduced colse to not existing.
Attached is a typical TC input filtering scheme, excerpted from a TC amplifier good practice/usage notes

Quote

RECOMMENDATIONS FOR BEST CIRCUIT PERFORMANCE
Input Filter
A low-pass filter before the input of the AD849x is strongly recommended (see Figure 29),
especially when operating in an electrically noisy environment.
Long thermocouple leads can function as an excellent antenna and pick up many unwanted signals.
The filter should be set to a low corner frequency that still allows the input signal to pass through undiminished.
The primary purpose of the filter is to remove RF signals, which, if allowed to reach the AD849x,
can be rectified and appear as temperature fluctuations.


~~~


Will do first and update.
 
Piotrkurak
Do the probes work properly when held together such that the shafts/cover touch each other?

Common grounds will make you look like my cartoon profile
 
ETomczak

Quote

Piotrkurak wrote:

Do the probes work properly when held together such that the shafts/cover touch each other?

Common grounds will make you look like my cartoon profile


Holy cow, I think you just enlightened me as to why my TC's were giving me so much trouble earlier. I could get one non-insulated TC to work at a time, but not two, and of course! It makes sense now, seems so simple.

Thank you!
 
Piotrkurak
So how many of your T/C were grounded on one leg? Hopefully your ADC uses a common return so you can take the grounded legs and combine them otherwise you have to get either not grounded TC (I forgot the exact term) or build a resistor ladder or silicone high temp grommets to isolate them
 
renatoa
TC4 ADC has differential inputs, nothing common to the four channels.
Insulated probes and not use mains adapters for electronics are the simplest ways to address a lot of issues.
 
exer31337
FFS No way. Anyways for those of you getting a lot of static on your readings in a machine and no static on your readings outside of your machine, some "ungrounded" thermocouples have grounded wire sheaths.... so once that wire sheath touches the machine in anyplace you will pick up so much noise it is unreadable. I found an exposed ungrounded thermocouple that did not present any continuity between the thermocouple leads and neither the thermocouple sheath or the stainless steel wire sheath running on the outside of the thermocouples... I am getting perfect readings now. I've bought every single board and tried tons of code to get this to work, just buy a good ungrounded thermocouple and save yourself a lot of effort.
 
Jump to Forum: