Classical temperature control involves monitoring a temperature sensor and adjusting the power to the heater via numerous classical methods.

I've tossed and turned at night for a week or so now trying to figure out why I haven't seen in the real world or online, why the resistance vs. temperature properties of the heating element itself is not used in a closed loop fashion to control power to said heater.

Nichrome wire for example: the resistance of the wire changes with temperature. If you characterize this aspect, why not indirectly control the temperature by varying the power such that the temperature of the wire achieves a desired resistance / temperature. The wire gets 'cool', the resistance drops, increase the duty cycle, wire gets hotter, resistance gets higher, once the resistance gets to the target, use various classical means to maintain the temperature via the resistance. The nichrome wire is in essence the RTD??? What am I missing? The advantage seems to be the elimination of a sensing device by leveraging the resistance vs temperature characteristics of the heating element itself. Ofcourse, closed loop control using the heater resistance may have it's on set of challenges. AC is typically used for these heaters. Not to mention, the heating element itself may be a huge source of noise, although I suppose this can be dealt with, it might be at an expense that far exceeds anything gained by the associated removal of the RTD.

My main interest was in reducing costs by removing the temperature sensor (e.g. RTD, thermocouple). Along with reducing system components and sources of failure. It is my experience that in industrial applications the sensing element, such as an RTD, is the common mode of failure. Eliminating the RTD would eliminate this failure mode. When a system no longer heats, there would no longer be a need to trouble shoot the RTD under the premise I've described.

In a nutshell, why can't the resistance of the heating element itself be used in a closed loop system to turn the heater on and off?

Thanks in advance,

C

 

If your heater is a light bulb, that's exactly what's happening. Heat output is self controlled by the increasing resistance of the filament with heat.

I think you could construct a "constant power" control scheme, but ultimately the temperature depends on factors other than the power INPUT. It depends on radiation, conduction, convection away from the element. Most folks are more interested in the temperature being controlled than the power.

 

If it's the temperature of the heater you care about, you can do a fast stop-and-ohm with a micro-controller. A dozen milliseconds out of a heaters life won't bother anything, but is a small circuit board full of parts more dependable than an RTD?

You could measure the voltage and current without stopping the heater. I do that all the time with a DVM and a clamp meter. Again, you are replacing one sensing device with several transistors.

"Why can't it be done" seems like a foolish question from here. It can be done. How much are you willing to spend to eliminate the RTD?

 

...why can't the resistance of the heating element itself be used in a closed loop system to turn the heater on and off?

I can be done, and I've done it myself.

I've used a PIC micro to drive a platinum wire to glowing red at an adjustable temperature within 1 degree C.

The PWM drove the power (4-6A) into the wire (freq. about 20kHz, on-time ~4uS). During the "off" time (~46uS), a precision current source engaged to measure the resistance. That was fed back into a *very* fast PID to compute a new PWM value for the next cycle.

Worked like a charm, and off 4 'AA' NiCads!

 

In heating, you're usually more concerned with the temperature of the matter you're heating than the temperature of the element itself. In your nichrome wire example, if using for hot wire foam cutting or something then sure, element resistance is probably a viable feedback. But if you're heating a room full of air or a block of steel then the resistance of the heating element is about as useful feedback as the temperature in the ladies restroom across the street.

 

The PWM drove the power (4-6A) into the wire (freq. about 20kHz, on-time ~4uS). During the "off" time (~46uS), a precision current source engaged to measure the resistance. That was fed back into a *very* fast PID to compute a new PWM value for the next cycle.

Excellent, and a lot faster than a millisecond!

 

Excellent, and a lot faster than a millisecond!

Had to be fast. The wire was a four turn coil of 0.001" platinum with a nominal resistance of ~1 ohm IIRC. Heat too long and it melts. Not fast enough and it dissipates the heat too quickly.

I did this about 13 years ago with a PIC16C774. Ancient by today's standards, but it had a really nice 12 bit A/D vs. the 10 bit A/D's most PICs come with today. The PWM was 10 bits, but I "overdrove" it to 14 bits with a neat trick that I developed. That's how I got the 1 degree C resolution/accuracy.

BTW, this was a project that simply could not be accomplished in C. The processor, running at 20Mhz, spent 90% of it's time in interrupts managing the A/D, PID, and PWM. An extra instruction here or there for C overhead would have killed the timing.

 

In heating, you're usually more concerned with the temperature of the matter you're heating than the temperature of the element itself. In your nichrome wire example, if using for hot wire foam cutting or something then sure, element resistance is probably a viable feedback. But if you're heating a room full of air or a block of steel then the resistance of the heating element is about as useful feedback as the temperature in the ladies restroom across the street.

Thanks all for the posts. It sounds as though it is possible purely from a controls standpoint. Whether the temperature control is any good or not is another matter. I'll have to do some tinkering in the lab. The problem is I'm not a controls guy and there will be a huge learning curve with the circuitry. Any guidance or suggestions here would be appreciated.

Perhaps sampling between pulse widths, as one gentleman suggested in a PWM scheme, is not necessary. Today we turn on and off the A/C mains via an SSR and PID logic to the heater wire until the desired temperature is reached. Perhaps I measure the the voltage across a resistor in series with the heater, take an RMS value to derive the heater's resistance. Turn the heater on and off until the desired RMS value is reached. Thinking out loud....

thanks again,

C

 

Thanks all for the posts. It sounds as though it is possible purely from a controls standpoint. Whether the temperature control is any good or not is another matter.
...

Well I'm sure with some nice electronics you could get the temperature at the WIRE controlled quite well. But for 99% of heating applications the wire temp is meaningless (as Strantor said).

An oven/dryer/incubator etc might be regulating at 50'C, but the nichrome wire of its heater element runs at 300'C for about a 50% duty cycle. Now the duty depends on the temperature at the sensor in the oven, which is at a distance from the heater with high thermal resistance and lags etc.

Even in a normal earthed type heating element the wire might be 300'C, the outer steel jacket of the element at 150'C and the oven at 50'C. How is your system going to keep the oven at 50'C?