Hello AAC community,
I am trying to build a PID controlled heater & stuck with the following situation. Any feedback in this regard is appreciated.

Purpose of the system: Maintaining the set temperature of a heatbed for 1 minute to 99 hours.

Materials:

• Arduino UNO
• 12V 20A DC supply
• Q1- TIP122
• Q2- 2N3055
• 40A DC Solid State Relay

System Scheme: DSS-current1_schem.png (attached)

Problem statement:
Want to incorporate PID control & PWM for heating. The heater needs 9A continuous current to function properly.
Looking for suggestions regarding selection of proper current controlling components/ system design. (Image attached)

Question 1:
is SSR suitable for this operating PWM frequency.

Question 2:
is 2N3055 suitable for this type of operation where continuous 10A current is drawn for 1minute to 99 hours?

Question 3:
What is the standard practice for designing such applications?

======================
Experiments done:
======================
TIP122 (image attached: DSS itr-1.png)
inference: Able to get 5.7 Amps max. which is not sufficient for the heater to function properly in therms of heating time.

SSR AC 40A: Not useful since it cannot make or break the DC contact in the AC terminals. https://www.amazon.in/Robodo-Module-SSR-40DA-3-32VDC-24-380VAC/dp/B07DDKRYPP/

References:

• https://forum.allaboutcircuits.com/...s-in-solenoid-with-arduino-and-2n3055.135084/
• https://forum.allaboutcircuits.com/threads/darlington-pair-10a-12vdc-transistor-spec.7653/

Please mention if any other information is required. Any help shall be highly appreciated.
Thank you.

 

Is this a school assignment?

Do you have to use these parts?

A power MOSFET would be a much easier choice to drive the heater.
You could use the 2n3055, but it's a pain. I would not use it, were it my choice.

 

Hi Sensacell,
Thanks for your response.

Is this a school assignment?

No, it is not. This is a barebone prototype for heating pathology slides at a certain set temperature. Stable temperature management for a set duration of time is the highest concern. Trying to implement the working principle of a high cost device into a low cost device with readily available industrial grade components/ modules without affecting the heating performance.

Do you have to use these parts?

No, components are not locked down. It can be altered.

I would appreciate if anyone explains how the "Specification to Design" cycle works for product development. Any good resources for getting started with useful electronic systems design industrially.

Device that I want to develop:

 

Thanks for the additional information.

What temp range do you need to maintain?
How do you intend to measure the temperature?
Do you need a fancy display like that shown?

The design cycle starts with a clear description of exactly what you want, with as many details thought through as possible.

The basic project is fairly simple, a PID controller for this application is not hard, especially because it can run slow.
For example, motor control PID needs sample rates in the KHz range.

I would find a 20A 30V N-channel MOSFET with a logic level gate, that and maybe a pull 5K down resistor on the gate is all you need for the heater driver.

 

Unless you have a reason to make your own design, you may be able to buy all or most of what you need already in product form. For example, here is a PID controller on Amazon for $25US. In the "Frequently bought together" section below is a thermocouple and SSR. If you can find parts like this that meet your requirements, it might be cheaper and faster to buy them than make it all yourself.

 

Hello Sensacell,
Thanks for your response.

What temp range do you need to maintain?

Operating temperature range is 30*C to 100*C.

How do you intend to measure the temperature?

I have used arduino and K-type thermocouple with MAX6675 module to translate the electrical signal to readable temperature (image attached). Fluke 59 max is used for validating the thermocouple value. Thermocouple is checked with ice and boiling water, in both the cases error is lesser than 0.50*C

Do you need a fancy display like that shown?

Yes, display is required. Either 7 segment display with buttons or touchscreen GLCD. Are there any other suitable alternatives?

For PID control, this resource is used:

https://create.arduino.cc/editor/LogMaker360/86a4099c-8dff-48cd-93be-e8ed52c5a898/preview
using SSR for the heater control & looking for MOSFETs suitable for this operation. Is SSR really a good option for this purpose?

While checking the temp. of bare aluminium heatbed using Fluke IR thermometer with both preset (0.95) and proper (0.30) emissivity settings in the device, it shows wrong temperature (way lower than actual, 35*C in IR reading when actually is 60*C in thermocouple reading etc.). This issue is solved by putting Kapton tape on top of the heatbed, now the temperatures are in line with each other. Can somebody please explain why is this happening? why cant we directly measure the correct temp. of bare aluminium heatbed?

Regarding design:
System framework is attached for reference. Presently only focusing on the controlled heating part. The rest part will follow. In this regard seeking for suggestions.
I have compared three different heaters for heating time & profile. Data, plots & inferences attached in zip file. If anybody wants to build together or all by themselves, the shared data might be useful.

Unless you have a reason to make your own design, you may be able to buy all or most of what you need already in product form. For example, here is a PID controller on Amazon for $25US. In the "Frequently bought together" section below is a thermocouple and SSR. If you can find parts like this that meet your requirements, it might be cheaper and faster to buy them than make it all yourself.

Hello MrSoftware,

Thanks.
I appreciate your response. The PID controller is a great option but it will not serve the purpose of the desired functionality with respect to the design criteria. I want to learn and make it myself using sensors, actuators & peripherals.

Should the entire system framework and components be rethinked to make it a robust & reliable device?

 

While checking the temp. of bare aluminium heatbed using Fluke IR thermometer with both preset (0.95) and proper (0.30) emissivity settings in the device, it shows wrong temperature (way lower than actual, 35*C in IR reading when actually is 60*C in thermocouple reading etc.). This issue is solved by putting Kapton tape on top of the heatbed, now the temperatures are in line with each other. Can somebody please explain why is this happening? why cant we directly measure the correct temp. of bare aluminium heatbed?

IR thermometers depend on assumptions about the emissivity (emission of IR at a given temperature) of the surface. Most non-reflective surfaces have similar emmisivity which is quite high. Shiny metal surfaces have far lower emissivity and so the thermometer must be calibrated to the surface. Better instruments allow calibration to agree with direct measurements, or, you can do as you’ve done and reduce reflectivity with tape or paint.

 

IR thermometers depend on assumptions about the emissivity (emission of IR at a given temperature) of the surface. Most non-reflective surfaces have similar emmisivity which is quite high. Shiny metal surfaces have far lower emissivity and so the thermometer must be calibrated to the surface. Better instruments allow calibration to agree with direct measurements, or, you can do as you’ve done and reduce reflectivity with tape or paint.

Thank you Yaakov for the response.
Are there any better alternatival for the heater driver?

 

The most efficient heater control scheme switched the heater off and on cycle by cycle as needed. This avoids power waste in a linear control device, and for most applications the temperature does not fluctuate with the rapid cycling. Another thing that helps the response is only using the modulating control, off/on, when the temperature is within some small area near the temperature setpoint. The size of that "small area" depends on the response time of the system to heat application.
You mentioned a slide heater and since I am not familiar with such, would you please explain the heating requirements as far as temperature and the accuracy of control required. AND, if you already have a heating system in place, let us know how fast it approaches the required temperature.

 

Hello MisterBill2,
Thanks for your response.

You mentioned a slide heater and since I am not familiar with such, would you please explain the heating requirements as far as temperature and the accuracy of control required. AND, if you already have a heating system in place, let us know how fast it approaches the required temperature.

Slide heater works as follows: It needs to maintain a fixed temperature for a certain period of time (Denaturaion & Hybridization of pathology slides). It can vary anywhere between 30*C to 100*C & 1 minute to 100 hours. Accuracy is +-2*C. Heating time is 3 to 4 minutes for rising from 30*C to 100*C.

Experiments done with 2.75W/sq. inch capacity flexible kapton heaters(images attached). Heating time is somehow satisfactory, around 4 minutes or more depending on ambient temperature etc. Thinking of replacing the heater with 5.5W/sq. inch capacity heaters for a significant reduction in heating time.

Heater control is under testing using SSR, Darlington & MOSFET.
Observations:
TIP122: Drawback is max. continuous current, which is limited to 5.5A max. but the heater requires around 9A max. & if higher watt density heater is used, the current may increase upto 20 Amps.
SSR: It delivers maximum current of 7.8Amps which is satisfactory but the deduction of 1Amps current leads to increased heating time. The SSR only operates if the temperature set point is above the heatbed temperature. (SSR is in either ON or OFF state, no PWM used). After reaching the set temp., the SSR switches its state in every 2 to 4 seconds to maintain the set temp.
MOSFET: IRFZ44N is used (circuit attached). In this experiment setup the MOSFET turns considerably HOT & with 5V G-S voltage the D-S current is only 600 mA. with 12V G-S voltage, the drain current reaches upto 1.5A only. Any suggestions to fully turn ON the mosfet? In this setup only saturation & cut-off mode is used, No PWM involved with MOSFET.

As far as the results are concerned, it seems that SSR is found to be the best possible way.

Q.1: what are the potential area of failure & area of improvement of the system if we proceed with the SSR?
Q.2: Why is the MOSFET not fully turned on (Not being able to conduct 9Amps current) even when 12V is applied across Gate & Source?
Q.3: As per construction, SSR is an optocoupler connected with additional circuitry & there is a triac internally connected to the Output terminals of the relay. I connected the DC heater with DC to AC SSR relay, the actuation (output) terminal (AC) was not switching. Later I used DC-DC SSR & now it works. why is this so?

Heater Spec: 12V, 100W DC heater. (image attached)
MOSFET Discussion: https://forum.allaboutcircuits.com/threads/irfz44n-mosfet-connections.41121/
IRFZ44N Datasheet: https://www.infineon.com/dgdl/irfz44n.pdf?fileId=5546d462533600a40153563b3575220b
SSR satasheet: https://cdn.sparkfun.com/datasheets/Components/General/SSR40DA.pdf
SSR teardown:

 

Usually if a mosfet is not switching on fully it is because of inadequate gate drive. So while the circuit design may provide enough voltage, the hardware implementation may have some resistance in the source connection that leads to a voltage drop that cancels part of the drive voltage. That has happened to others before you, it is not something new. And also it may be that the mosfet does indeed need some heat sinking beyond whatever it has. And you will need a higher gate drive voltage than the drain voltage to provide adequate saturation. One choice for that voltage is a small DC to DC module to provide a second 12 or 5 volts above the 12 volt system supply voltage.

For the SSR, an AC switching SSR will not work with a DC supply, as you discovered. Are you able to have an AC source for the heater supply? That would be less power required of the DC supply section, and probably the SSR would run a bit cooler. If there is a transformer in that 12 volt 20 amp supply and you can access the secondary, and it is 24 volts center tapped, then you would have 24 volts AC and an AC SSR would work and you would have plenty of power available for heating.

For that bipolar transistor, you can use them in parallel or select one with greater current capacity, which is probably simpler.

And one more option is to connect some resistance across the control device so that the heater always has some power applied, not enough to reach the desired temperature but enough to allow the control device to remain in control, switching between high and low instead of between high and off. That could also reduce temperature variations.

 

Look at the Picmicro App AN958 for heat control using burst firing of a SCR circuit.
Also Fairchild AN-3006 and AN-3003.
Max.

 

Look at the Picmicro App AN958 for heat control using burst firing of a SCR circuit.
Also Fairchild AN-3006 and AN-3003.
Max.

Mostly, SCR control arrangements work with AC powered systems, and it appears as though this system must operate from a 12 volt 20 Amp supply. It remains to be seen if any AC at all is available for heater power. So far that has not been mentioned as an option.

 

For a 120w heater, I would personally look at burst control rather than PWM, even working with DC, For this I would use a logic level Mosfet in place of a couple of Bipolar versions.

SSR AC 40A: Not useful since it cannot make or break the DC contact in the AC terminals

If thinking of a SSR, they also come in DC out as well as AC.
Max.

 

For a 120w heater, I would personally look at burst control rather than PWM, even working with DC, For this I would use a logic level Mosfet in place of a couple of Bipolar versions.

If thinking of a SSR, they also come in DC out as well as AC.
Max.

The use of an SSR was reported, along with the complaint that the internal voltage drop limited the maximum power that could be delivered. THAT was my motivation for suggesting to use 24 volts AC and a transformer, and an AC SSR.

 

Hello everyone,
Thanks for all your insights on this topic.

For the SSR, an AC switching SSR will not work with a DC supply, as you discovered. Are you able to have an AC source for the heater supply? That would be less power required of the DC supply section, and probably the SSR would run a bit cooler. If there is a transformer in that 12 volt 20 amp supply and you can access the secondary, and it is 24 volts center tapped, then you would have 24 volts AC and an AC SSR would work and you would have plenty of power available for heating.

For that bipolar transistor, you can use them in parallel or select one with greater current capacity, which is probably simpler.

And one more option is to connect some resistance across the control device so that the heater always has some power applied, not enough to reach the desired temperature but enough to allow the control device to remain in control, switching between high and low instead of between high and off. That could also reduce temperature variations.

Yes MisterBill2, AC is also an option for the heater & probably will go with that for its advantage in terms of less heat dissipation and current through the control circuit.

Look at the Picmicro App AN958 for heat control using burst firing of a SCR circuit.
Also Fairchild AN-3006 and AN-3003.

Thanks Max for the suggestion. I have gone through the datasheet & it seems very useful when I switch the heater typre from DC to AC, which is going to be in the next design iteration.

AC heater specification is 220V AC, 400 watts. But with AC load I have a concern of power factor. Though the heater is acting as a resistive load, we must consider the power factor below unity, i.e. 0.75. In this configuration the heater might consume (400/(220*0.75)=) around 2.5Amps AC which can be easily controlled by SSR or SCR. Is it a standard practice in the industry to consider power factor for resistive loads as well? or am I missing something?

Question regarding sensors:
Presenylt in the device K-type thermo couple with max6675 translation module is used. Previously tried with LM35-DZ also.(Images attached)
Issue:
Both thermocouple and LM35 are working but there are some design constrains I am facing. Thermocouple's tip with thermal paste takes more than 130 mm vertical space above the heatbed. For enclosure design it is a HUGE disadvantage, it will make the device unnecessarily big and wasteful. For placing thermal sensors on heatbed, the maximum space I am thinking of is 30mm x 30mm x 30mm.

LM35 has heat conductivity issue. The plastic (or similar casing) naturally prevents it from getting fixed on top of the heatbed properly & The expermental data from the LM35 also seems not so accurate due to heat transfer issues from metal heatbed to plastic casing of LM35 sensor.

There is also another way of getting the heatbed temperature, which is thermistors that come with 3D printer heatbeds. But the output readings will depend on the R2 of the thermistor voltage divider circuit that can vary depending on temperature and other factors, thus involving errors in the heatbed temperature reading which I am trying to avoid.

My question is that, are there any other ways of using these sensors in an alternate fashion to overcome the design constrains? or any other suitable sensors that can fulfil this design requirement?

Thank you all for the generous support.
Roy.

 

N.C. There are all kinds of thermocouples, including those made from really thin wire and having no housing. Not only are they much smaller, they also offer much faster response, and they cost much less than that very solidly built one in your pictures. A thermocouple made from #22 wire could easily be bonded to your heater plate and respond rapidly to temperature changes. The only serious issue will be electrical isolation, which would require that you use a thin copper foil ground plain between the heater and everything else.. It is simple and easy to produce your own thermocouples from available TC wires, if you have a welding torch. It does take a bit of practice, though.
A thermistor could also work well, but my experience is that each one must be calibrated in the application. Not hard for a single unit, but a big deal for production quantities if accuracy matters.
The LM35 conductivity concern will matter most if the heating rate is faster than the conduction following the temperature, and so with the higher wattage heater it will probably not serve your purpose.

 

In this configuration the heater might consume (400/(220*0.75)=) around 2.5Amps AC which can be easily controlled by SSR or SCR. Is it a standard practice in the industry to consider power factor for resistive loads as well? or am I missing something?

Resistive loads are not usually an issue when considering PF correction, mostly inductive loads and where high speed switching equipment is used in number.
Max..

 

Just to illustrate what MisterBill2 is getting at:


The inset upper left is AWG 22 while the main image is AWG 12. Thermocouples also come in a wide variety of sheath materials for sheath type versions ranging from stainless steel to inconel depending on environment to be used in. Here are several sheath types and note the diameters. The below are type J but the same is true of type K or any other popular type. The below also use the standard large TC Connectors but they also come with much smaller Mini Connectors. These also come in Grounded and Un-Grounded designs where the junction is or is not grounded which may matter..



As mentioned, when choosing a thermocouple two of several factors to consider are the temperature range and the response time. Obviously the smaller the sheath the faster the response. That is true of just about all types. There are also thermocouples designed for surface mount for measuring surface temperatures.

How fast your thermocouple responds will effect the control process, frequently called dampening. You may want to consider that.

I looked at the first SSR you linked to and sincerely hope you did not pay the shown Amazon price for that SSR of $399 as that has to be in error.

I won't try and suggest a commercial controller but there are commercial PID controllers which can store profiles of temperature control as well as ramp and soak features which would be difficult to acheive with an Arduino.

Ron

 

Power factor with a resistive heater in the on/off cycling mode does not usually have a power factor far from unity. PWM control could be different, probably will be. The 12 volt 20 amp switching power supply will be a bigger concern.
For the small volume being heated I am thinking 400W is too much, but if you are running a heater element rated 400W at 220 on 120 volts then the power level should be about right. Servo response, (and the system is a servo system) is easiest to keep stable when the heating rate is not faster than the sensing rate. 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.