# Seeking help for component selection. Arduino Heater Control : current 10A, system voltage 12V DC

#### MisterBill2

Joined Jan 23, 2018
9,785
A new SSR with 5 volts control input and suitable for 120 volt service may run as much as $25 at a higher priced source, or closer to$15 at a decent source. The bottom TC, #4, is what I was thinking about, attached with heat conductive epoxy.

Joined Jan 15, 2015
6,204
The problem would be overshoot if the temperature continues to rise after the heat is switched off.
And the response time of the thermocouple in the photo will be far longer than the temperature rise time with a 400 watt heater.
Dampening

As MisterBill2 points out, this is how it looks on a plot of Set Point verse Time. Critically damped tuning is the goal and why choice of thermocouple for your conditions is important.

I also believe that combining PWM with PID Control may not be the best method of control. With PWM you have a fixed frequency (period) with a variable on time. The problem is the PWM frequency is fixed. That will limit your PID control.

Ron

Joined Jan 15, 2015
6,204
A new SSR with 5 volts control input and suitable for 120 volt service may run as much as $25 at a higher priced source, or closer to$15 at a decent source. The bottom TC, #4, is what I was thinking about, attached with heat conductive epoxy.
That's how I would approach it. The max temperature is really not very high allowing for a few good options. I have used Kapton Tape and as you mentioned high temp thermally conductive epoxy.

Ron

#### MisterBill2

Joined Jan 23, 2018
9,785
To put the explanation in simpler terms, underdamped operation happens when the response speed is faster than the sensing speed, overdamped happens when the response speed can not keep up with the sensing speed. This is a gross simplification so I hope none of those with a doctorate degree in feedback systems get on my case about it. But most folks will find the mathamatical explanation a bit tedious.

AND, the spellcheck for this site has a rather limited vocabulary.

#### KeepItSimpleStupid

Joined Mar 4, 2014
5,090
A couple of things:

The chosen end-element of control depends a lot on thermal mass. I'll defines some of them

bang-bang or ON/OFF - setpoint doesn't really exist, lots of overshoot depending on system.
Proportional duty cycle - generally the time on/time off is controlled proportionally (PWM long periods)
ZCP - Zero crossing Pulse - The minimum amount of heat is 1/2 a line cycle controlled proportionally.
Phase Angle Control - The box generally converts a process signal to be proportional to V^2 for resitive heating. The firing angle is divided up into increments of discrete power.
Power control - The box converts a process signal to power, 0-10V might be proportional to 1-1000 W.

The above two might have current and power limiting.

When you use DC you can use PWM at a much lower period. You an also control a power supply 0-100%. It's easy to measure V, I, P and energy for he latter.

Phase angle fired is generally the cheapest for the best control using AC. PID generally make the setpoint = measured value and prevents overshoots, but if your minimum on time is say a second and a second causes an overshoot, then that method won;t work.

So your playing with thermal masses and how fast you would like the system to be stable and if you want the setpoint = to the measured value.

Joined Jul 18, 2013
23,899
There is a Vast amount of info out there on PID control, including free books such as McMaster Univ etc.
Max.

#### MisterBill2

Joined Jan 23, 2018
9,785
The most precise temperature control system that I have seen is one that I designed for a hydraulic test machine. It used proportional duty cycle on/off control of the heat source, and while the spec called for +/- 5 degrees C it routinely held within 1 degree at 210C in a 20 gallon tank of oil that was constantly circulating. Fast sensing and somewhat slower heating served very well. So proportional duty cycle can work very well, AND it should be simple to implement. But a fast responding temperature sensor and low electrical noise will be required.

Joined Jan 15, 2015
6,204
I have no idea if it is worth consideration or not but the thermocouple, like any other sensor, is subject to limits of error. Thermocouples are made with thermocouple wire (logical ) The grade of the wire determines the allowable maximum error the thermocouple may have. Just for example a Type K Thermocouple:
LIMITS OF ERROR
(Whichever is greater)
Standard: 2.2°C or 0.75% Above 0°C
2.2°C or 2.0% Below 0°C
Special: 1.1°C or 0.4%

The average off the shelf thermocouple is a standard grade. Special limits grade are available upon request from a manufacturer. When purchasing a thermocouple from a reputable source the buyer may request a calibration curve or correction chart. The chart may show error degrees C, degrees F or even milli-volts. The buyer calls out a range and cardinal points such as 0 degrees C through 100 degrees C at 0, 20, 40, 60, 80- and 100 degrees. Other than buying certification the thermocouple should be within the above limits of error. When better is desired you have the thermocouple charted.

No clue if this should be a consideration or not for this application but it may be a consideration. That as well as the cumulative uncertainty of everything in the measurement plane.

Ron

#### MisterBill2

Joined Jan 23, 2018
9,785
The system only needs to be accurate over a narrow range, and it should be possible to calibrate it for at least one of those temperatures. I did not think that this was going to be a production design, but rather a very small set of systems. I hope that I am correct on that.

Joined Jan 15, 2015
6,204
The system only needs to be accurate over a narrow range, and it should be possible to calibrate it for at least one of those temperatures. I did not think that this was going to be a production design, but rather a very small set of systems. I hope that I am correct on that.
My guess was it wouldn't matter also. Just figured I would toss it out there.

Ron

#### KeepItSimpleStupid

Joined Mar 4, 2014
5,090
In post #19, the twisty thing is prior to welding. if used as is all of the wraps are junctions and you get the average temperature/

One way to "weld" a TC is to pass a current between it and scraping the TC end on a carbon bock that completes the "welding" circuit/

Joined Jan 15, 2015
6,204
In post #19, the twisty thing is prior to welding. if used as is all of the wraps are junctions and you get the average temperature/

One way to "weld" a TC is to pass a current between it and scraping the TC end on a carbon bock that completes the "welding" circuit/
That "twisty thing" was my home brew Type K science experiment. I couldn't weld the tip since I gave my neighbor my tanks and torch. On the bright side it did work just fine.

Ron

#### MisterBill2

Joined Jan 23, 2018
9,785
In post #19, the twisty thing is prior to welding. if used as is all of the wraps are junctions and you get the average temperature/

One way to "weld" a TC is to pass a current between it and scraping the TC end on a carbon bock that completes the "welding" circuit/
The last thermocouples that I made were #22 wire with one twist and then joined by a very brief application of an oxy-acetelyne torch flame. Quite probably the carbon block approach is better but the torch was what I had on hand at the time. And skinny wire couples have a faster response. And they are fairly cheap.

#### MisterBill2

Joined Jan 23, 2018
9,785
A non-welded TC is not likely to be stable for long-term.and mechanical disturbances in the twisted joint can disrupt the calibration as well.

Joined Jan 15, 2015
6,204
A non-welded TC is not likely to be stable for long-term.and mechanical disturbances in the twisted joint can disrupt the calibration as well.
Oh yeah, another 5 min in the flame and it would likely have become flaky. This was a kitchen table affair and photo op.

Ron

#### nightcrawler218

Joined Dec 24, 2012
26
Thanks for all the inputs. I have tried different thermocouple and wires to obtain temperature of the heat bed.

Observations:
While using siingle thermocouple, the temperature reading is accurate but when I put more than one TC (max. 9 thermocouples attached) on a 20cm x 20cm aluminium heatbed, the readings start fluctuating & ending up with a lot of negative integer values. Can this be a ground loop issue? Tried checking with the thermocouple itself. All the TC are grounded and shielded TC & MAX6675 translation modules are used in this experiment.

Qustions:
1. How to avoid ground loop or AC interference from the TC circuit?
2. Are there any other suitable alternatives for measuing the temperature of the aluminium heatbed?

Suggestions are highly appreciated. Thanks

#### MisterBill2

Joined Jan 23, 2018
9,785
Evidently there is some interaction among the theromcouples that are connected to the plate. Does that happen when the thermocouples are not touching anything conductive? Does it happen with all of them or just some of them?

#### KeepItSimpleStupid

Joined Mar 4, 2014
5,090
Use an ungrounded thermocouple probe or an isolation amplifier. The ungrounded thermocouple would have less response time than a grounded one.

What your describing sounds just like a ground loop.

Joined Jan 15, 2015
6,204
Back in post #19 I mentioned using grounded and un-grounded thermocouple types. What exactly are you paralleling? Thermocouples can be paralleled but there are considerations which need to be made. Can you provide an accurate drawing of what you have exactly? That includes any hot and cold junctions you have?

Ron

#### nightcrawler218

Joined Dec 24, 2012
26
Hello,

Evidently there is some interaction among the theromcouples that are connected to the plate. Does that happen when the thermocouples are not touching anything conductive? Does it happen with all of them or just some of them?
yes, it happens even when the thermocouple hot junctions are not electrically connected to each other via any conductive material. for example, I put Kapton tape on the bare aluminuim side & tried taking readings but that error results occur. It is not shoeing this erratic behavior persistently. Sometimes 4 TCs are working together even without the islation, flawlessly. and sometimes it is showing wrong results even with two TCs and electrical isolation.

Connected the arduino & Main power (12V DC, 350W) supply ground together and later Figured out that the arduino & power supply GNDs have potential difference. When the controller power supply (PC USB) is off, UNO was still leeching power throuh the common GND line. Fixed this by implimenting diode in the GND line to make the current flow unidirectional. But it still not fixes the issue.

Use an ungrounded thermocouple probe or an isolation amplifier. The ungrounded thermocouple would have less response time than a grounded one.

What your describing sounds just like a ground loop.
In that case should all the TC to be changed from enclosed Grounded to enclosed ungrounded? I followed this video to indicate the type of thermocouple.

Back in post #19 I mentioned using grounded and un-grounded thermocouple types. What exactly are you paralleling? Thermocouples can be paralleled but there are considerations which need to be made. Can you provide an accurate drawing of what you have exactly? That includes any hot and cold junctions you have?

Ron
I have put 9 (3x3) enclosed shielded thermocouples onto a test bed & then fixed the assembly on a heatbed. The purpose is to obtain temperatures from different parts of the single aluminium heatbed at the same time.

Single line diagram attached for reference. Cold junction temp. is checked & corected by MAX 6675 or Max31855 (only one type of sensor is used during testing).

The entire purpose of the experiment is to obtain the temperature profiling of the entire heatbed. Thermocouples are found out to be very reliable & stable when working individually but thay also turned out to be a very sensitive device when it comes to parallelling multiple TCs.

Questions:
What measures are suitable for TC signal conditioning, other than isolation amplifiers?
Are there any other sensors to do this job under these system requirements?

Found some wafer based sensing ICs from TI but I am sceptic about the accuracy, usability and mounting of these parts.
http://www.ti.com/product/TMP107
http://www.ti.com/product/TMP107-Q1
http://www.ti.com/product/LM92
http://www.ti.com/product/TMP75C-Q1

Looking for your views & feedback.

Thank you,
Roy.

#### Attachments

• 298.9 KB Views: 9
• 181.2 KB Views: 11
• 261.9 KB Views: 9
• 310.2 KB Views: 9
• 266.3 KB Views: 8
• 287.2 KB Views: 9
• 153.9 KB Views: 12
• 122.4 KB Views: 12
• 68.4 KB Views: 11
Last edited: