Temperature control

I have a task to control temperature in vacuum chamber with 10 4,5kW heaters. For switching heaters ON/OFF I’m using mechanical contactors. For a control I'm using one PID loop that is modifying preset of one timer. That timer is defining heaters ON time. Another timer is measuring OFF time.

Because of mechanical contactors this ON/OFF times are quite long so accuracy is not so good. Is there any other control loop for this task, maybe two PIDs one defining ON time and other OFF.



Thanks!

If the problem are the mechanical contactors, why don't you use solid state contactors?

re:

why dont you check a propotional control alone other than PID ? keep the switching time minimum as 1 sec. still the contactor will not give accuracy. you can only go for an open ON/OFF.

best option is go for SSR, work on PWM / propotional control...

you havent mentioned the accuracy required..

Your PID is already controlling the ON and OFF time. Doing two PIDs as you mentioned is the wrong solution.

I assume (I hate that word - but you haven't told us nearly enough to know anything about your system) that you are using 1 temp input (t/c or RTD) for all 10 heaters... that you're trying to maintain a temperature in your vacuum chamber.

Where is your temp input located? How is it mounted? If you're measuring the temperature in air (which I assume you are because it's in a vac chamber), then your process is going to be very slow to react. Air is a horrible conductor of temperature and is very slow to respond. You will find your PID overshooting horribly.

SSR would be better, but a contactor is probably quite adequate. But with what you've told us, we don't know enough about YOUR system to be able to make any good suggestions.

Isn't the issue 'straight line control', that is, getting the temperature to hold close to the setpoint without lots of oscillations above or below the setpoint?

Can we assume that the problem is that the use of mechanical contactors which requires a minimum on time and minimum off time, making for a fairly long cycle time? And that this requirement is necessary because othterwise rapid cycling of the mechanical contactors will destroy them?

Is the long cycle time producing overshoot when the heaters are on for too long, and undershoot when the heaters are off for too long?

Or is 'accuracy' some other problem ?

There should be no problems with the 10 heaters in one zone. Heat treats do this all the time. The only issue here really is cycle time. I do this in similar systems with a 60 second cycle time and get control no problem. If you use a mechanical contactor, it probably will fail after a while. So, you either need to stock some backups, get something solid state, or use an SCR. Dont forget to put a second high limit contactor in for when the control fails.

matt

If you want to use PID control don't use two loops. Use the output of the PID heater control as a percent load. Then use the percent to calculate the on and off times. For example, if you have a 30 second total cycle time and your loop output is 25% you will turn on for 7.5 seconds and be off for 22.5 seconds. This is referred to as Time Proportioning Control.

I agree with Tom Jenkins. I have seen the method he suggests used succesfully.

If I understood your post correctly, you have a PID controlling the on time of one timer but the off time is fixed by another timer that is not being modified. Is that correct? If so then your heating duty cycle is not truly porportional to the PID output, and that will make tuning the PID loop difficult.

I suggest that you use a single PID to modify the on/off time of a single time period. For example, select a time period of 10 seconds. The PID determines what percentage of the 10 seconds the timer is on, the remainder of the period it is off. So 30% is 3 seconds on, 7 seconds off, 80% is 8 seconds on, 2 seconds off.

Some PLCs have in instruction that will do time proportioning for you. Which brand/model are you using?

Thanks for all the answers. I have made something like Tom and Alaric proposed.



Another question is do I need to implement some kind of deadband around setpoint,

something like 5% of setpoint value, and disable regulation if measured value is in that area?



For a weekend I'll see if that is going to work. I hope so and if not I'll get solid state contactors.

Deadbands can help. Theoretically with PID control you don't need it, and the reset portion of the PID algorithm is intended to eliminate proportional droop, which is the difference between setpoint and actual value that persists in proportional only control. If you want to eliminate hunting deadband may help, and your PID in the PLC may have this as a standard adjustment.

How tight are you trying to control? A solid state contactor will survive rapid switching better than a mechanical one, but unless you are using a very short cycle (several operations per minute) they won't be a problem. They aren't a solution to accuracy, just equipment life.

If you can tolerate a 5% error you can probably get away with longer cycle times. Remember that the heaters have inherent hysterisis. It will take some time for the heater to warm up, and they will continue to heat the process fluid after the power is turned off. That complicates tuning. The switch time of the mechanical contactors is probably milliseconds, and that won't affect control accuracy.

tgaljar said:

Because of mechanical contactors this ON/OFF times are quite long so accuracy is not so good.
Thanks!

Click to expand...

The original poster has not indicated that this might be an issue, but let me ask if this might be his problem.

My only experence with heating in a vacuum was bad. Heating solids in a vacuum is not easy. There is very little air to even out the heating energy. This was on a vacuum chamber for drying wood. The original manufacurer used plattens that were heated with water to the temperature that was desired for the specie. My customer was looking for another way to put lumber into the chamber that was not so labor intensive. His thoughts were to use electric heaters inside the chamber and heat the wood that way instead of alternating layers of wood and heated plattens. The bulk of the energy the product recieved was by infared (this only works by line of sight). Because the wood was cooling itself by evaporation of the water, all of the surfaces of the wood were being cooled and only the surface that was in the line of sight of the heater was being warmed. Then when you add the fact that the temperature sensor was not being cooled by the evaporation of the moisture the sensor never reflected the actual temperature of the product. Remember there is no such thing as "convection" in a vacuum.

To the original poster: If what I have mentioned is not part of your problem, disregard this post

Vaccum is sometimes counter intuitive for some people, but if you think about exclusively radiant energy transfer for a minute then it makes sense. Without a gas to diffuse radiant energy shadows are completely black. You can stand in the shadow of a tall building in the day and its still light all around you. This is because air molecuels diffuse the sunlight. In a vacuum it would be completely black, so its possible to have hot and cold spots very near each other in a vacuum furnace.

We recently discussed this on a different vacuum furnace temperatur problem in this thread: http://www.plctalk.net/qanda/showthread.php?t=39387