Heaters management optimization

[Stefano]

I was commissioned to retrofit an old granulate mixer, the automation is all electromechanical and must be made compatible with Industry 4.0, I will use one slimline compact with some iE series expansion and terminal modules. This machine mixes at a controlled temperature, has a rotating drum with an axis inclined by about 45 degrees. The drum is surrounded by resistors, these are dense in the lower part and thin out as you reach the upper part, there are two thermocouples: one where the resistances are dense and one in the upper part of the drum where there are no resistors.

The thermocouple in the other controls the process temperature, connects to a thermostat that the operator regulates between 70 and 120 degrees, the thermostat drives a remote control switch that controls the resistances. The bottom thermocouple is connected to a fixed calibrated thermostat at 230 degrees (temperature not to be exceeded, otherwise the granules will be destroyed) which switches off the previous thermostat.

I will replace the contactor with a static zero crossing and I will choke the power with a sort of PWM, surely the temperature of the thermocouple at the top will enter a PID block to manage the heating, but I don't know how to manage the second. I could replicate the current operation by disabling the previous PID block, or, with a further PID calibrated on the maximum temperature, continuously vary the maximum power applied to the resistors. Has anyone ever had a similar need?

[Sergio Bertana]

Mine are free-minded assessments having never faced this type of adjustment. The temperature to be controlled is the process temperature so on that you have to make the control PID, but certainly when the system is switched on before the temperature spreads to the process thermocouple you will have a phase in which all the resistances will give 100% power and in this phase it could be that the lower temperature reaches its threshold value. This event, if possible, already happens today is the only thing that today the electromechanical system can do is turn off the resistances. In this way, with continuous switching on / off of the resistances, the temperature propagates in the structure and reaches the upper thermocouple.

Now you can keep the same situation, or provide a heating ramp in which you activate the resistors not at 100% power to give the whole mass time to warm up. Of course, if you use a low percentage of power it will take you longer to warm up and in any case with a predefined percentage you do not have a feedback of the real behavior.

Another possibility is to create, as you say, a heating PID on the control thermocouple, whose set point is fixed by the process PID (With a maximum of 200 degrees, not to exceed the maximum 230). Then the process PID will vary the heating PID set point to adjust the process temperature.

[Stefano]

They are struggling with the sizing of the equipment. I think I use the two PID blocks, one which varies the set point of the other which then controls the 'PWM'. The CPU will therefore have to manage the two PID blocks, the generation of the CPU static out (period of 2 seconds), the reading of a couple of 4-20ma inputs, the reading of a dozen encoders (the environment is very dirty and in some phases it becomes difficult to use the touch screen. The operators will then turn the rotary encoders from the panel to modify the setpoints. I will use the block IOEncoder). It will manage a handful of I / O (a few buttons, some lights), it will communicate as a modbus slave with the panel and as a master on 485 with a motor driver.

The logic is quite simple but the various managements are carried by the CPU. I don't know whether to try with a compact CPU or switch to one Cortex. At worst, would it be possible to request a modified input module to have various encoder hardware operators in quadrature?

[Sergio Bertana]

Given the price difference (which is not substantial) I suggest you use the CPU Cortex M7 which is also the model on which all the functions we are developing are supported. Among other things, its 2 analog inputs have a resolution of 12 bits so I think more than enough for the acquisition of your 2 4-20 mA signals.

If you only have to manage panel encoders for setting process values, it is not necessary that you use cards with hardware quadrature, the IOEncoder FB is certainly able to manage them. You can perform the acquisition of the digital inputs and the instances of the FB in a Task Fast (1 mS), in this way it is possible to acquire encoders with a frequency up to 500 Hz.

But be careful when turning the encoder by hand, even if on average the frequency is very low, you can have very high frequency peaks and therefore with the FB you can lose some counts. But since the encoders are only for setting and not for positioning I don't think this can be a problem.

[Stefano]

I would say that even if it loses some count, the operator can turn the knob a few more clicks. It is not a critical thing. If I am not mistaken, I extend the encoders with IOEncoder in a fast task, the inputs must be acquired with SysGetPhrDI because the input image mapped by the system is updated with less frequency.

[Author]

Post