Peter Nachtwey
Member
Instead of this info getting lost in another thread I will start a new thread.
For those that don't know you can search for Advanced Control on this forum to see topics where I get serious.
I was asked to evaluate this.
https://literature.rockwellautomation.com/idc/groups/literature/documents/wp/logix-wp008_-en-p.pdf[/QUOTE]
Actually it is a good start on how to apply the PIDE. This knowledge and the PIDE is a enough for most people to get the job done.
However, I do have some nit picking on the pdf and a rant about how control is taught.
1. I have never liked the term velocity mode or position mode. The two formulas have NOTHING to do with velocity or position.
I brought this topic up on a newsgroup sci.engr.control over 20 years ago. Another person said to think of it as increment mode for velocity and full value mode for position control. I liked and have used his terms since.
I prefer to use the incremental mode when possible but that means there must be an integrator gain to sum up the incremental changes in the control output.
What isn't mentioned is that the incremental mode makes it easy to prevent integrator windup. It happens naturally as long as the total output is limited to 100% +/-.
However, with as simple trick of how the control output is limited, the integrator for the full value mode can be prevented from winding up.
It is true that incremental mode makes it easier to change gains on-the-fly.
In full value mode we change gains slowly to avoid the discontinuities that the Rockwell document complained about. Those that have been to Delta's Advanced training how how we do this.
Rant! So you can change the control gains on-the-fly. What do you change the gains to? This is not covered and we show how to find the gains. Also we can update the closed loop gains every millisecond or faster.
The multiloop selection is good.
We/our customers are often controlling applications where position/force limiting is required. The question is how did they implement this. Another question is can it be bi-directional? In other words hold a set point like a position until the force pushes too much one way or another.
The Timing Modes are good. Does anybody use anything other than the Periodic Mode? The Oversample mode should have been implemented on the older PLCs to compensate for varying scan times.
The RTS mode uses double precision floating point which is good. The trapezoidal mode help but not that much.
The Ratio Control is almost like gearing. No big deal.
Cascade Control explanation is good. It correctly shows that two feed back devices are necessary for each loop. There is nothing special here.
Split-grange Time-Proportioned Loops.
The srtp instruction looks good but does anybody use it? Maybe I am not paying attention.
Again, nice tools but how does one set this up? It seems that a lot of trial and error is required.
I would need to know more about the SRTP instruction to comment on it further. I do know it would be wasteful cycle back and forth between heating and cooling. I also know that cooling happens much faster with water depending on the temperature different. Sometimes cooling can be done with a lack of heating or heating can be done by a lack of cooling. It depends on the ambient temperature and whether passive heating/cooling is fast enough.
Actually, I would have preferred an output of +/- 100%. Positive numbers are heating and negative numbers are cooling. In our motion controller have the ability to have different gains for positive outputs and negative outputs.
This seems more natural and flexible to me.
The simulation is crude but will probably work for most people doing temperature control.
Overall the document is good and worth understanding thoroughly.
However, it is not really enough and it is basically about using the tool.
I know this is a PLC forum and the Rockwell document is perfect for this forum, but there is so much more. Read my tag line.
For those that don't know you can search for Advanced Control on this forum to see topics where I get serious.
I was asked to evaluate this.
https://literature.rockwellautomation.com/idc/groups/literature/documents/wp/logix-wp008_-en-p.pdf[/QUOTE]
Actually it is a good start on how to apply the PIDE. This knowledge and the PIDE is a enough for most people to get the job done.
However, I do have some nit picking on the pdf and a rant about how control is taught.
1. I have never liked the term velocity mode or position mode. The two formulas have NOTHING to do with velocity or position.
I brought this topic up on a newsgroup sci.engr.control over 20 years ago. Another person said to think of it as increment mode for velocity and full value mode for position control. I liked and have used his terms since.
I prefer to use the incremental mode when possible but that means there must be an integrator gain to sum up the incremental changes in the control output.
What isn't mentioned is that the incremental mode makes it easy to prevent integrator windup. It happens naturally as long as the total output is limited to 100% +/-.
However, with as simple trick of how the control output is limited, the integrator for the full value mode can be prevented from winding up.
It is true that incremental mode makes it easier to change gains on-the-fly.
In full value mode we change gains slowly to avoid the discontinuities that the Rockwell document complained about. Those that have been to Delta's Advanced training how how we do this.
Rant! So you can change the control gains on-the-fly. What do you change the gains to? This is not covered and we show how to find the gains. Also we can update the closed loop gains every millisecond or faster.
The multiloop selection is good.
We/our customers are often controlling applications where position/force limiting is required. The question is how did they implement this. Another question is can it be bi-directional? In other words hold a set point like a position until the force pushes too much one way or another.
The Timing Modes are good. Does anybody use anything other than the Periodic Mode? The Oversample mode should have been implemented on the older PLCs to compensate for varying scan times.
The RTS mode uses double precision floating point which is good. The trapezoidal mode help but not that much.
The Ratio Control is almost like gearing. No big deal.
Cascade Control explanation is good. It correctly shows that two feed back devices are necessary for each loop. There is nothing special here.
Split-grange Time-Proportioned Loops.
The srtp instruction looks good but does anybody use it? Maybe I am not paying attention.
Again, nice tools but how does one set this up? It seems that a lot of trial and error is required.
I would need to know more about the SRTP instruction to comment on it further. I do know it would be wasteful cycle back and forth between heating and cooling. I also know that cooling happens much faster with water depending on the temperature different. Sometimes cooling can be done with a lack of heating or heating can be done by a lack of cooling. It depends on the ambient temperature and whether passive heating/cooling is fast enough.
Actually, I would have preferred an output of +/- 100%. Positive numbers are heating and negative numbers are cooling. In our motion controller have the ability to have different gains for positive outputs and negative outputs.
This seems more natural and flexible to me.
The simulation is crude but will probably work for most people doing temperature control.
Overall the document is good and worth understanding thoroughly.
However, it is not really enough and it is basically about using the tool.
I know this is a PLC forum and the Rockwell document is perfect for this forum, but there is so much more. Read my tag line.