Design of heater circuit using resistors

[astroasma]

Hello everyone!

I am a space engineering student at TU Berlin and I am stuck on something.
I am working on designing a small (10cm x 10cm x 10cm) microbiology incubator that is intended to be sent to the ISS for future plantation of fungi.
The thing is, the incubator needs to maintain a temperature of 30 degrees at all times, the temperature on board the ISS is 20-24 degrees C. To increase the T, I am planning to design a circuit composed of solely resistors.
Resistors lose heat as a form of energy and this is how I calculated how much heat is lost:

Given a certain voltage V, a current I which can be obtained depending on the resistance in use, and for a heating time t:
Heat Produced =(PotentialDifference)×(Current)×(time)
The heat produced is thermal energy in joules, which can be converted to degrees C.

Now, the problem is the following:
I only need a circuit of resistors to increase the T by a maximum of 10 degrees C. I set my heating time to one minute.
But I am having a problem figuring out how this is related to the volume, I need this increase of 10 degrees to cover all the cube of 10^{3}cm.
Please help if you know how this is solved.

Kind regards,
Asma

[wraper]

How are you supposed to calculate how much energy you need to heat a certain amount of volume at certain amount of time without knowing thermal mass and thermal conductivity to ambient?

[Benta]

You are going about this the wrong way, I think.
For a temperature rise to a precision fixed temperature, you need a closed loop system.
What does this mean?
It means you need a precision temperature sensor in your incubator that will give feedback to your controller.
Based on this feedback, the controller will regulate the power dissipated by the resistors.
The resistors just need to be powerful enough (or even more powerful) to achieve this goal. The closed-loop controller will do the rest.

Team up with someone who knows about PID loops. Getting the system stable is not trivial.

[mag_therm]

Yes, before considering the controller and power needed etc   you will need to know :
Convection to air from surface   [ Watt/m^2.K ] expect 15 to 30 W/m^2.K
Surface Emissivity [dimensionless ref black body =1 ] , for shiny metal like aluminum ~ 0.1, for black coated metal ~ 0.9
Heat flow via any mounting attachments etc.

With those the steady state temperatures and heat flow  can be calculated.

To do the transient heating, the specific heat, dimensions and positions of all masses and the resistors will be needed.
If I were to do this I would certainly use 2D transient heat analysis.
It might seem overkill, but it adds discipline and necessitates all parameters to be entered, saves approximations on the box corners,
 moreover you get results and archives in a formal format.
And do a transient at the outset because steady state is in there too saving another model.

[astroasma]

Thank you so much for your reply!
Sorry I forgot to mention that I do have a closed loop system!
The whole circuit will be controlled via a microcontroller that will supply power to the circuit once the T sensor indicated a drop below 30, and supplies power to the resistors until the incubator's ambiant T is 30 again.
So this is not the problem.

[astroasma]

Yes this is my question! I don't know how to get there..
I just know how to calculate the thermal energy dissipated for a certain amount of resistance and I couldn't do the rest.

[Benta]

Just heat the chamber experimentally with your resistors, it doesn't have to be exact, it just needs to have "enough" power.
Trying to do all the thermal maths on this is a fool's errand.
The resistor power dissipation is just that: P = U x I.

[astroasma]

Thank you for your reply!
The material used to build the incubator will be plexiglass, the circuit will be mounted on one of the panels along with a small fan to help distribute the T equally on all sides.
So the heat flow will be really from air to air as there will be no structure to cover the circuit. The microcontroller along with the T sensor will help regulate the T once it drops below 30.
Is there any tool that you recommend to do the 2D transient heat analysis? thanks

[mag_therm]

The plexiglass might have lower convection to external air than  my post #3 because I assumed metal.
And  the radiation will now depend on emissivities of objects inside, through the (assumed) transparent plexiglass.

Does your Institution have FEM? If so there should be data available for the materials, and maybe someone familiar to assist
 with setting up thermal FEM. Maybe there is  already FEM package with transient heat coupled to fluid flow to model air inside the container.

That will help to identify / avoid stagnation of air flow, hot spots etc which might upset the experiment.

Workflow for a job like this:
Draw (solid) all the parts they way they need to be. Collect all properties by search if necessary
Model
Optimise
Refine model
....

[planet12]

The whole circuit will be controlled via a microcontroller that will supply power to the circuit once the T sensor indicated a drop below 30, and supplies power to the resistors until the incubator's ambiant T is 30 again.

If you're just doing simple on/off control, you're going to have overshoot and/or oscillation around the setpoint.

Overshoot will be due to thermal inertia - you'll pump a bunch of energy into the heating element, and it will take a while to heat the air - and will continue heating the air for a while after it's switched off due to the heat still in the element.

Oscillation will happen if you have a single setpoint with no hysteresis.

Another user suggested PID control - this is by far the best option if you need a stable temperature, using PWM to do proportional control of the heater, temperature sensor to measure, and the PID to close the loop.

For thermal FEM analysis, if your organisation doesn't already have a package for it, FreeCAD can do it - although you're probably better off to just build it and measure.

[james_s]

This is exactly what I was going to suggest, you need a PID controller, that is a solved problem, there is already code out there to do what you want, or analog circuits that do the same. It sounds like what you are proposing is bang-bang control and that is not going to be accurate at all. Also instead of ordinary resistors I would suggest a film style heating element. They come in a wide variety of sizes and shapes, many with adhesive backing so they will stick on whatever you are trying to heat. IIRC they were one of the developments that came out of the space race.

[Benta]

It seems we're all on the same page.
But don't ignore my suggestion about getting someone on the team who knows how to tune a PID system. It's not trivial.

[wraper]

Thank you so much for your reply!
Sorry I forgot to mention that I do have a closed loop system!
The whole circuit will be controlled via a microcontroller that will supply power to the circuit once the T sensor indicated a drop below 30, and supplies power to the resistors until the incubator's ambiant T is 30 again.
So this is not the problem.

That in not a quite correct way to do it unless you don't really care about temperature stability or have extremely low thermal mass. You will have significant temperature cycling since temperature will continue to rise after you switch off heating and will not raise immediately once you switch it on.

[Anders Petersson]

I made an 1.3 gram analog temperature controller with adjustment pots. It controls the PWM duty cycle to an external resistive heater, for high efficiency. It uses proportional control only so it's not as exact as a full PID controller, but certainly better than the suggested on-off scheme.
The manual can be helpful also for other implementations: https://sparv.io/Heating%20controller%20manual%20v1.pdf
PM me if you want to use the controller itself.

A general note: If the heating is uneven throughout the volume, the location of the feedback temperature sensor is crucial to the accuracy of the temperature control. Your plan to use a fan is good.

General note #2: I think a simulation is overkill. Just build the box and measure where the power draw settles in equilibrium. Make sure the power supply has some margin. If the power draw is too high, add insulation around the box.

[Wallace Gasiewicz]

Here is a suggestion;
Use a mosfet on a heat sink as the heater.
Use an op. amp as the drive to the gate of the mosfet.
Regulate the drive from the op amp with the temp sensor, probably a temp sensitive resistor that is in a proper spot. Or your Controller.....
The resistance of the source to drain of the mosfet will create heat when the mosfet is turned on. You can add resistance from drain to ground in a different spot if you find it necessary to spread out the heat more than the heat sink will do.
Turning it on and off smoothly will be the biggest challenge.

You might actually want to measure the ON source to drain resistance of the mosfet you use rather than take the specifications for granted.

Do they actually want a fan??? Fungi are sensitive to humidity, what will be humidity control???

[mag_therm]

It might seem over kill, but an elementary model consists of drawing in about 12 lines and loading 9 properties !
https://app.box.com/s/r2z8z89x0htem83l00b2mwdlnk4ryud6
I can not do fluid air here and recommend that for the real job because some moving air may be necessary.

I thought there must be soil or something for the fungus.
So this model consists of a 100 by 100 by 100 mm plexiglass box with an irregular triangular pile of soil at the bottom.
The static air above the soil has a thermal conductivity of 0.030 W/m.K from a paper by NIST.

I embedded a wide copper plate in the soil and applied power to it.
At 2.6 Watt the soil and air immediately over it reaches about 35 C with air outside the plexi box  ( boundary condition) at 20 C.
The thermal time constant of the soil near the wide heating plate is aproximately 220 second.

The whole box aproximates steady state after about 2100 second.

[cellularmitosis]

A while back I built a resistor-based heater for an aluminum Hammond enclosure (the resistors were epoxied to the aluminum using thermal epoxy), driven by an Arduino running a PID loop (if I recall, the resistors were controlled via an op amp and a small transistor -- the target temp was 25C, so only just above ambient).

I kept logs of the whole process of tuning the PID, so you can replay all of my trial-and-error:

https://github.com/cellularmitosis/logs/tree/master/20180124-25c-chamber

https://github.com/cellularmitosis/logs/tree/master/20180126-25c-chamber-tuning

I eventually got it to regulate within +/- 0.01C, which I was pretty happy with.

[cellularmitosis]

The material used to build the incubator will be plexiglass, the circuit will be mounted on one of the panels along with a small fan to help distribute the T equally on all sides.
So the heat flow will be really from air to air as there will be no structure to cover the circuit. The microcontroller along with the T sensor will help regulate the T once it drops below 30.

Just to check my understanding, there will be a plexiglass box, and inside of the box will be the resistors with a fan blowing over them, and the temperature sensor will also be in the box (probably on the opposite side)?

[astroasma]

Hii, thank you so much for your input.
I checked what you did and I think it's pretty awesome, I will try to do something with similar components but I need to design a circuit board (which will be directly placed on one of the panels of the box), and I only need an increase by 10 degrees so I need to find out how many resistors will I need for that.
Yes the box will be in plexiglass. I don't know where to place the T sensor and the fan yet, I will need to simulate how the heat distributes best and go accordingly.

[cellularmitosis]

Hii, thank you so much for your input.
I checked what you did and I think it's pretty awesome, I will try to do something with similar components but I need to design a circuit board (which will be directly placed on one of the panels of the box), and I only need an increase by 10 degrees so I need to find out how many resistors will I need for that.
Yes the box will be in plexiglass. I don't know where to place the T sensor and the fan yet, I will need to simulate how the heat distributes best and go accordingly.

Great, the good thing is that plexiglass and cardboard are both insulators, so you could implement the entire project in cardboard and then the final translation to plexiglass will only require a small change to the PID constants.

If you are fast with simulations, that sounds like a good approach.  For someone like me, it would be faster to simple implement it and just see how it behaves / tweak it.  I.e. implement the control circuit, and then have a separate arduino which just monitors a few thermistors placed within the box to see how well the heat distributes.  Once the project is working satisfactorily, the second arduino would be removed.

If the resistor approach isn’t a requirement, you could potentially simplify the project just a bit by having the heater be a single mosfet attached to a heat sink / fan combo.

[astroasma]

I am not required to use resistors. I actually thought it would be the simplest idea, I didn't think MOSFETs could be a solution too (I have very basic knowledge in electronics).
I just found out I can to use a coil as well for heat induction, check this out if you want (:

But the T increase in using that circuit looks a lot higher than what I need, what do you think I need to tweak in my circuit to get a less T increase? Thank youuuuu

[exmadscientist]

It might be worth noting that it is common in professional equipment to use the flexible heater strips (widely available at Digi-Key and friends). These are just a resistor made of a copper trace printed on a polyimide backing. They bend well and spread out the heat well. They are not so easy to control as a FET gate, but it isn't too hard (especially if you go with PWM). The ease of getting the heat out that they provide may or may not be important to you.

[Wallace Gasiewicz]

I suggested the Mosfet previously.
You need something to run the higher heater current through anyway, so taking advantage of the heat the mosfet gives off seems quite reasonable.
I have seen the Mosfet approach in commercial oven controlled Xtal oscillators at much higher temperatures.
You could probably just use an aluminum sheet metal piece cut to match one side of the box as your heat sink

Here is another suggestion: Use a thermoelectric cooler in reverse.
These things are solid state heat transfer devices
These things work well enough to make small refrigerators, use a small one to transfer heat into your box
You probably only need maybe a 2x2 cm module for this project.
You can control the heating/cooling by changing the voltage.
Many modern laboratory grade incubators use this technology. Apparently it is the current teck used.

[Wallace Gasiewicz]

There was a comment about a model with soil at the bottom.
I think is space there is no gravity "bottom". I suppose the fungus will grow on the plexiglass?

[mag_therm]

There was a comment about a model with soil at the bottom.
I think is space there is no gravity "bottom". I suppose the fungus will grow on the plexiglass?

Oh Yes !!  ....I am sorry.
And the fan will become a little sand blaster.