FX2N PID TUNING

[Colin Carpenter]

For what it's worth, this is what I've learnt this week. I had to tune a system that used a high spec pneumatic control valve to close againts a water flow generated by a centrifugal pump. The flow rate was monitored by a water meter giving a 4-20 ma output. This was the Process Variable. The Set Point was an operator specified flow rate. The valve took 10 secs to fully close (from 100% open), when working against the water flow pressure. The same valve took 5 secs to fully open (from fully closed) when working with the water flow pressure. Disabled the Derivative Constant and tried to set up the Proportional and Integral values from guesswork ..... no joy ..... just a wildly unstable system. Read the manual and tried the Autotune function ..... no joy, in fact I have't a clue how that's supposed to work, as it seems to hold the MV value constant. Anyone know? Spoke to a Guru, who said ... a) Set Proportion Constant to 100% b) Set Integral Value to 60% - 80% of the system response time (in other words how long it takes the valve to move from fully open to fully closed and vice versa) c) Disable the Derivative Constant. The Integral Value was then set to 6 secs (60 x 100 ms) and immediately the system got better. 100% on the Proportional was too much and resulted in much bigger changes than I needed .... ended up at 10% which gave me a slow, steady landing on the setpoint. Calibrated the flow rates against valve position bits, and use that value as the manual PID mv. After a few seconds, witch to auto and PID works like a charm. Fow what it's worth ......

[Goody]

Thankyou - maybe one day it will be worth a lot to me :) I am just about to do a project that I may use PID on. It must be a good project cos I am scared - thats good innit ???

[thomas1000]

Thanks for the info. I too have had trouble tuning the Mitsu PLC Loops. I tried tuning them as as explained in the following link http://www.clabberhead.com/pidtutorial.html without success. The only difference I would expect between the two is that one uses gain and the other uses proportional control. I ended up tuning them by changing the PID values and watching the response. However the Siemens PID loops tuned well with the above link. What percentage value of integral time did the guru recomend for derivative. If anyone can explain what the difference between the above link and mitsu tuning is I would be very greatful.

[Colin Carpenter]

It seems to me that the constants used by different manufactures mean different things, and that very few people truly understand the meanings, me included. The FX2N manual defines the P and I constants as shown below: Certainly, in my case, getting the "I" value right was the best piece of advice I could get. Note that in the FX2nN, the units of the I value are seconds, and it seems that this value should always be around "60-80% of the system response time" (the time it takes for the valve to move 100%). Once this value was entered, I only needed to experiment with the P value to get the system exactly how I wanted it The guru didn't really mention the D term .... seems that PI is the preferred method as Derivative values can introduce massive mood swings!! P - proportional change When a proportional factor is applied, it calculates the difference between the Current Error Value, EVn, and the Previous Error Value, EVn-1. The Proportional Change is based upon how fast the Process Value is moving closer to (or further away from) the Set Point Value NOT upon the actual difference between the PVnf and SV. Note: Other PID systems might operate using an equation that calculates the Proportional change based upon the size of the Current Error Value only. I - integral change Once a proportional change has been applied to an error situation, ‘fine tuning’ the correction can be performed with the I or integral element. Initially only a small change is applied but as time increases and the error is not corrected the integral effect is increased. It is important to note how TI actually effects how fast the total integral correction is applied. The smaller TI is, the bigger effect the integral will have. Note: The TI value is set in data register S3+4. Setting zero for this variable disables the Integral effect.

[justwhy2003]

Hi Im a bit of a novice on PID loops but understand the basic principal. Im trying to set one up in an FX3U. By following your simple steps above and reading the analog edition of the control manual, do i need to worry about Limit cycle method or Step response method? Justin