Hello Everybody,

     Intend to use this block to control the temperature in a Spray Dryer. My questions are the following:

1.- Regarding the parameters that are irrelavant for this control, namely SELECT(According to what I read in the manual. I would like to understand this parameter better if it's actually relevant for continuous control), QPULSE, QC_ACT, CYCLE_P. Is it okay just to set them to 0 or FALSE? 

2.- When the process is started, the controller should not act very fast because the heating duct could be damaged, so we intend to use a slow controller, with limits in the manipulated variable for this discrete state (in which we haven't reached a PV value to control faster). Then we can change the parameters of the PID block, including the setpoint, changing it for the one ncessary for the operation of the spray dryer. My question is, would it be better to program two controllers or just to alter the instance DB once this first set point is reached. If the latter is the best option, how would you recommend doing it? Is it necessary to stop the controller, then change the parameters and then start it again? 

3.- I'm having a problem with the parameter assignment interface, when I call the DB58 from it, it says that it's not an instance DB, when actually in its properties it's explicitly written "instance DB for FB58".

I do not see you need a PID to your application because you do not need a fast response 

the attached file explain little bit about the philosophy of your control 

Thanks for your response MAZENJER,

     I appreaciate the information; it could be very useful if for the purposes of humidity control and other measures of the output of the spray dryer having those fluctuations proves acceptable. However, it doesn't answer my questions about the FB58. 

     Having the ability to control the heating system slowly and then "fast" and compensating for the step change in the SP is something we are after, as well as reducing as much as possible the steady state error. Given that there is another control for the air flow inside the spray dryer and it also affects the humidity, we want to be as precise as possible with the temperature. If you can—or anyone—provide some information regarding my original questions I would really appreaciate it.

MAZENJER

I do not see you need a PID to your application because you do not need a fast response 

the attached file explain little bit about the philosophy of your control 

 What about this: Starting only with the P action slowly and then when the PV is getting within 90% of the SP turning on the D and I actions for a more precise control in steady state.

hi,

the FB58 has a integrated tuner. you will find tuning function by opening the Instance-DB and going online. Then go to Options-> Controller tuning. With the tuning function the FB58 will find the P,I and D parameter. 

The Online Help gives real good information how the tuning and the whole FB58 works. 

hope this helps,

regards,

As for unused options you can set false, or leave empty.

There are two possibilties for your application:

1st. Implement a ramp from actual temp to setpoint with rate of change. When you start the SP shall slowly ramp up, the controller will try to adjust that temp.

2nd. Use manual switch and set manualy the output, or what I do usauly: Limit the min/max values of PID to defined value that is manualy set, when you want to switch to auto ,just set min/max limit to default value.

mopS04

 the FB58 has a integrated tuner. you will find tuning function by opening the Instance-DB and going online. Then go to Options-> Controller tuning. With the tuning function the FB58 will find the P,I and D parameter

  Thank you very much, what about the problem with the parameter assignment interface? When I call the DB from there it says that's not an instance DB; however the DB explicitly says in its properties dialog that it's an instance DB for the FB58.

Marko Bursic

As for unused options you can set false, or leave empty.

 What about the SELECT parameter? I'm not really understanding that one, is it really irrelevant for contiuous control?

Marko Bursic

1st. Implement a ramp from actual temp to setpoint with rate of change. When you start the SP shall slowly ramp up, the controller will try to adjust that temp.

 Let me see if I understand: I make the SP slowly—at a speed close to the speed I want the PV to reach the actual operating value—ramp up and make the PV follow it, all that with a PID tuned as if I wanted to control "fast"?

Marko Bursic

2nd. Use manual switch and set manualy the output, or what I do usauly: Limit the min/max values of PID to defined value that is manualy set, when you want to switch to auto ,just set min/max limit to default value

  I didn't understand this, could you please go into a little more detail? I mean, you start with a manual setting of the Manipulated Variable which is actually already at a limit, then when switching to automatic the limits change so it can control. I don't think I understood completely.

In the manual states that SELECT has to have a value at each call. I guess the PID algorithm and PULSGEN are called in with different sample times, because the PID is more heavy on computation, also the sampling time has to be defined according process system dominant times, while PULSGEN has to be called more frequently if you want 1% PWM resolution of output (100x times faster than PID sampling time), therefore in the interrupt the PID is skipped several times and only PULSGEN is evaluated.

As you dont plan to use PULSGEN, then you should set SELECT to a value that disbles pulsegen, SELECT=3 (not sure).

As for PID tuner, I have forgot the usage, maybe someone will give you correct procedure to monitor and setup the PID with your instance DB.

Ramp: Yes, you tune the controller tight to reject the distrubance, then you need to implement a ramp function that will ramp the SP towards the final SP with specified rate of change.

Manual: I have seen problems when turning from manual to automatic, as the controller is reset and wakes up with integral part zero, this causes a bump. I can't say for this temp controller with tuner, I usualy use classic PID  FB41, but to overcome the bum I don't switch to manual mode, rather I limit hi/lo output value to a desired manual value. With such trick, the integral part of controller is always computed to match the output, then when the limits are reset to default values (like hi 100%, lo 0%) then the controller output continues from last value, bumpless transfer.

Marko Bursic

In the manual states that SELECT has to have a value at each call. I guess the PID algorithm and PULSGEN are called in with different sample times, because the PID is more heavy on computation, also the sampling time has to be defined according process system dominant times, while PULSGEN has to be called more frequently if you want 1% PWM resolution of output (100x times faster than PID sampling time), therefore in the interrupt the PID is skipped several times and only PULSGEN is evaluated.

As you dont plan to use PULSGEN, then you should set SELECT to a value that disbles pulsegen, SELECT=3 (not sure).

As for PID tuner, I have forgot the usage, maybe someone will give you correct procedure to monitor and setup the PID with your instance DB.

Ramp: Yes, you tune the controller tight to reject the distrubance, then you need to implement a ramp function that will ramp the SP towards the final SP with specified rate of change.

Manual: I have seen problems when turning from manual to automatic, as the controller is reset and wakes up with integral part zero, this causes a bump. I can't say for this temp controller with tuner, I usualy use classic PID  FB41, but to overcome the bum I don't switch to manual mode, rather I limit hi/lo output value to a desired manual value. With such trick, the integral part of controller is always computed to match the output, then when the limits are reset to default values (like hi 100%, lo 0%) then the controller output continues from last value, bumpless transfer.

 Makes good sense, thank you very much. I'll have a go and test it and probabl ask some more questions afterwards.