Continue to Site

# Muffle furnace control- PID etc?

Status
Not open for further replies.

##### Member level 5
I am building a laboratory muffle furnace for heating molds and melting.
Temperature up to 1350C, power 4kW, heating, heating wire kanthal A1 approx. 0.5 kg,, izolation 10cm AlO3+SiO2. Temperature measurement, thermocouple, for now K, later S or B.
such a classic, nothing special-
Now I'm starting to solve the furnace control.
I will use MCU (STM32) for control.
What can it do?
1. Reach the desired temperature and hold it there for the specified time.
2. They reach the desired temperature in the specified time. That is do not exceed dT/dt, for example 1C per 1 minute.
it looks simple. I measure the temperature once per second, if it is lower than the desired one, I heat it, if not, I don't heat it.
At the same time, from perhaps 20 measurements, I will calculate the speed of the temperature increase, and if it is higher than the required one, I will block the first control loop from heating even if the required temperature is not reached.
alternatively I can incorporate the PID into the control loop.¨However, I have no practical experience with PID (coefficient tuning) or furnace control.
I would welcome a practical example of furnace drive or at least advice from someone who has already dealt with a similar matter.

Thanks

You might consider Fuzzy Logic control (which is sort of like what you were explaining).
It uses If-Then-Else logic in the control loop and uses no linear PID functions.
A Fuzzy Logic program is written the way a (fast) human operator might tweak knobs or switches to control a system (e.g.
If the temperature is slightly high Then slightly turn down the heater power Else
If the temperature is slightly low Then slightly turn up the heater power
Else
If the temperature is correct Then maintain the heater power
Else
If the temperature is moderately high Then moderately turn down the heater power
Else ........................)

Here's a tutorial on it.

Ready made temperature controllers with display are available for a few 10$. They are mostly implementing PID control, optionally slow PWM for SCR full wave switching. Oven control isn't particularly demanding, I would use PID algorithm in the first place. PID parameters can be either determined with Ziegler-Nicols method or calculated from measured step response. thank you for the responses >temperature controllers with display are available for a few 10$.

Yes and no.
the cheap ones usually do not allow
- the use of different thermocouples
- cannot or it is difficult to enter the required temperature development curve
- they have one or two segment displays from which you cannot tell which phase of heating you are in
- cannot directly communicate with PC and telephone
- I can no longer monitor parameters such as energy consumption for a given "melting" list of uses, etc.
- those who can do at least part of it don't cost $10, but more like$1000 or more see for example nanodac

last but not least, I can shop, I want to learn something new and build it myself

> Fuzzy Logic
I generally know what it is and that it is used in similar cases. It's practically as mysterious to me in detail as determining the coefficients in a PID
Yes temperature control is actually simple from a PID point of view, easier than holding a PID ball on an inclined plane.
I expected that in every book about PID, the PID proposal for heating, be it a furnace, a kettle or a house, would be presented as a practical example. Well, it's not, at least in the books I've downloaded and looked at so far.
To approach the issue

In precision casting using the lost wax method, which is used for example by jewelers or jet engine manufacturers, different molds are used to make molds, and each mold must be hardened and heated to the temperature at which it is cast before use. Here is an example temperature profile for UNICAST
It is baked for 14 hours with the exact temperature profile

Hi,

precise temperature control is not a simple task.

Thus, the first you need to do is decide your requirements.

***
I`ve done several temperature regulation applications.
In one (conveyor belt furnace for curing plastics) I continously measured temperature and power and calculated the heat capacity of the parts on the belt (there could be no device on the belt .. or the belt could be full of devices) and calculated the energy for needed for the parts to heat up to the desired temperature. ..besides the power needed without any goods. It was a quite difficult algorithm. So when new good came in, the air temperature became a lot more than the expected temperature of the goods, to quickly adjust their temperture, then the air temperature dropped ... in the end it worked very well.

This way we avoided to put temperature sensors inside the goods. There were several zones with independent temperature control.

Klaus

Btw: I don´t know why you need different thermocouples. Usually you decide a temperature range and then chose the suitable thermocouple.

OK i will take it step by step and from the end.

Why more types of thermocouples?
For me, the primary reason is that I have K-type thermocouples in a drawer and up to 1250 is enough, the question is what up to 1350. If not, I replace it with S, B comes into consideration, but it does not measure well below about 300C.
Secondary reason. I want to measure the temperature also between the layers of insulation of the furnace and the temperature on the outer surface of the muffle. And that's thanks to K. The reason is the calculation of heat losses and possibly the creation of a mathematical model of the furnace. High-pressure insulations have the property that their thermal resistance changes with temperature.
I have a module fitted with a MAX31856 and a 2x4 low-ohm analog switch, so I can put practically any combination of up to 4 thermocouples on the furnace.

Calculations.

I also did some math.
We heat the oven itself.
Only thermal insulation has a weight of approx. 30 kg and a heat capacity of 1,13 kJ/kg K.
We heat the heating wire aprox 500g Kanthal A1, heat capacity 0,46 kJ/kg K
Heated stuff very variable worst case 2kg copper, heat capacity 0,4 kJ/kg K and 1kg graphite 1.4 kJ/kg K

heat loss through the wall
at a temperature of dT
200C 65W 0,065 kJ/s
600C 360W 0,36 kJ/s
1000C 1080W 1,08 kJ/s
1200C 1820W 1,82 kJ/s

at a heating output of 4kW, i.e. 4kJ
The biggest problem to maintain the temperature will be at low temperatures, when losses are minimized.

Hi,

for the measurement side I expect some specifications:
* what accuracy/precision/resolution at what temperature?
* Which is the most important temperature range?
* what is the most critical temperature for the things inside the oven?
* what dT/dt do you want to achieve/regulate?

***

So k-type thermocouples should work. Don´t you agree?

***
You say 4kW heating. Is this one single element? Or is it 3 phase? Can you use wye delta configuration?

Klaus

It's practically as mysterious to me in detail as determining the coefficients in a PID
Okay.
But is seems more obvious what the Fuzzy Logic If-Then-Else statements should be to get the profile you need, as compared to the opaque operation of a PID loop.

>applicability of unusable thermocouple type K

The literature commonly states that it can be used up to 1250 (1260) C.
If you look at the tables, you will find that they contain voltages up to 1373 C.
Around 1400C, this type of cell starts to melt.

My conclusion, the thermocouple type K can really withstand 1350 C and can measure it, the question is what is the service life at this temperature. I'm afraid she'll be miserable.

>Heating elements

Now I only have one for 4kW 230V.
If it's not enough I can add another one or two 4kW each. max 3x4kW, only Y, not delta.
I have it at home, i.e. in the workroom near the house and I am limited by a 3x25A circuit breaker.

>Rate of temperature rise

I looked at the curves of the individual masses and the fastest is 3.5 C/min, but it should be taken as the maximum allowed increase. Slower growth is not a problem, it just prolongs the "baking" time of the mold.

The plan is this.
I have to finish the doors and their closing, connect the heating to electricity and at least measure the temperature.
Then I turn it on and measure the rate of temperature rise.
According to the results, I can try to add secondary insulation, but that won't help much, I won't reduce the losses at 1200C by more than 300W. The second option is to add a second 4kW heater.

Hi,

please use the "Reply" feature of the forum software. (Mark the text you want to reply on, then press REMARK)

The benefit of 2 heaters (you need 2 phases)
* 1 heater to neutral --> 4kW
* 2 heaters to neutral (Y) --> 8kW
* 2 heaters in series to neutral --> 2kW
* 2 heaters in series between two phases (Delta) --> 6kW

Klaus

Status
Not open for further replies.