PID

Learn the NEC with Mike Holt now!

gadfly56

Senior Member
Location
New Jersey
Occupation
Professional Engineer, Fire & Life Safety
What is 'stable'?
The feedback is via 4-20 transducer.
PSI is very stable at += .05 but Hz varies, by 10 in the same time frame.
Can we have a little more background? If the PSI is the output, and holding that close, then I'd say it's very stable and the PID controller is doing its job. Is the Hz variation 10 orders of magnitude or 10 Hz? What is the average value for Hz?
 

petersonra

Senior Member
Location
Northern illinois
Occupation
engineer
Is this graph showing the output to the drive or some kind of feedback from the drive?

As a practical matter since it is stable and doing what you want not messing with it is probably your best option.

This looks a little like excess derivative action. I rarely use D action. It just causes all kinds of things like this in slow loops. But it does not seem to be hurting anything.

Keep in mind that if this is a centrifugal pump the pressure generated will be based on the RPM and the loop probably is not even necessary as the pump will put out whatever pressure the pump curve says it will at whatever RPM it is running at.
 

Jraef

Moderator, OTD
Staff member
Location
San Francisco Bay Area, CA, USA
Occupation
Electrical Engineer
If the pressure is the Setpoint, then how it gets there is somewhat irrelevant. And is that the actual frequency, or the commanded frequency? If that’s just the commanded frequency, remember that the actual frequency involves the ramp time too.
 

gadfly56

Senior Member
Location
New Jersey
Occupation
Professional Engineer, Fire & Life Safety
If the pressure set point is 50 psig, I'm curious how it maintains pressure with zero output from about 7:21 to 7:23. Is there a bladder tank in the system?
 

TwoBlocked

Senior Member
Location
Bradford County, PA
Occupation
Industrial Electrician
My experience is more practical than theoretical. I look at the "manipulated variable" (in this case hz) for steadiness, not the "process variable" (in this case PSI). So, what I see is too much gain. Halve the gain and see what it looks like. As far as derivative, I've seen it recommended to set the derivative at 1/8 the integral after you have obtained your best results without derivative. This has worked well for me.
 
Location
NE (9.06 miles @5.9 Degrees from Winged Horses)
Occupation
EC - retired
My experience is more practical than theoretical. I look at the "manipulated variable" (in this case hz) for steadiness, not the "process variable" (in this case PSI). So, what I see is too much gain. Halve the gain and see what it looks like. As far as derivative, I've seen it recommended to set the derivative at 1/8 the integral after you have obtained your best results without derivative. This has worked well for me.
The PSI is steady, but I don't really need it that good. Seems like I should be able to get smoother Hz line. I'll give it a try.
 

Carultch

Senior Member
Location
Massachusetts
What is 'stable'?
The feedback is via 4-20 transducer.
PSI is very stable at += .05 but Hz varies, by 10 in the same time frame.

Stable just means that it approaches a steady condition in the long run, that has the output following the input signal of the controller, where you desire it to end up. As opposed to unstable, which means errors from the input signal are exacerbated, and the output gets farther and farther away.

The middle ground to this, is marginally stable, which means it oscillates around the desired state, and the amplitude of the oscillations remains constant.

There are plenty of behaviors that all classify as stable, but some make for a better controller configuration than others. Such as rise time, overshoot, settling time, and steady state error, all of which you try to minimize, so that the output gets to the desired state as fast as possible, and stays there. But since you can't necessarily minimize all of these at once, you "pick your battles" based on what is best for your application. Sometimes, any overshoot is unacceptable, so you aim to be as close to critically damped as you can, and stay on the overdamped side within real world tolerances. Other times, a barely-noticeable overshoot is acceptable, and you choose such a design, because it helps you achieve a better rise time and settling time.
 

kwired

Electron manager
Location
NE Nebraska
I thought you retired? :sneaky: :LOL::giggle:
I have an HVAC friend that retired a couple years ago. Was a general construction guy before he went to HVAC. Decided to take on some "small projects" after retirement, but one day told me he had recently been physically working harder on those projects than he did most the time he was in HVAC.
 
Looks like the bladder is doing it's thing (keep the pressure steady), so watching the pump control becomes kind of irrelevant. Actually, since you have two separate actuators trying to affect the tank pressure (pump Hz and bladder), the fact that they're not fighting with each would make me just close the cover and say "Looks good".
 
Top