Stupid Question-- Why not Open-loop?

The transfer function of the plant is not often known accurately enough for the desired control accuracy, and it's not in reality linear nor is it time-invariant. It's also impractical to account for all the disturbances.

Could you you adequately control the speed of your car merely from the accelerator pedal position, without feedback from the speedometer?

Best regards, Spehro Pefhany

Dnia Fri, 04 Nov 2005 07:20:30 +0100, sunnewton napisa³:

Consider the structure of your C' compensator. It's much more complicated then C (consists of both C and P dynamic). It means that your compensator can be more expensive then your plant. So maybe build a better plant? It would be simplier. Consider also that you can't control unstable plant in this way (serial connection C'P). Most significant is that you have no idea what is happening on the plant output because you have no sensor. How to set plant output exactly in required state if you can't check it? No sensor means no feedback. However stepping motors are sometimes control without feedback :). But it is another story.

In the real world we never know the transfer function for a process. Any knowledge of the process is an approximation, and includes variables that can't be measured.

It is more typical to have almost no knowledge of the transfer function of the plant when designing the controls.

sunnewt> Consider a unity feedback system consists of a plant P(s) and a

The key word here is **sometimes**. There are, however, situations where the stepper motor control can continue to provide step pulses while the motor has jammed stationary. Without some form of feedback you are totally lost about what is happening on plant.

You know the way from home to work as well as you know anything-- you've done it so many times. To gain a deep appreciation of feedback, try traveling the route with your eyes closed.

Jerry

funny I had the exact same question way back in my college years. Aside from inaccurate P(s) others have mentioned, inject some disturbances into the system and you will appreciate that open-loop will not be equivalent to closed-loop.

In fact, where you have a plant that's well behaved enough to drive open-loop you do, and you avoid all the stability problems with closed-loop control. For instance this is why stepper motors are so popular, even though the generally suck in every other way when compared to DC motors.

Dnia Fri, 04 Nov 2005 12:54:23 +0100, Paul E. Bennett napisa³:

I know about that of course. It was kind of curious tale.

Hello Spehro,

In some countries they can. It's either full up or full down. Avoiding too many dents is handled in a similar fashion, mostly by using the horn which is a digital device. Using the brakes a lot is for sissies...

Regards, Joerg

On Fri, 04 Nov 2005 02:01:02 -0500, Spehro Pefhany proclaimed to the world:

This is a good analogy. When I was a kid I used to watch people drive, observing their foot on the pedal and the speed. I noted how some would not react until the speed had dropped after mounting a hill. Others would anticipate the hill and speed up for it. My mom always had a cycle. She would accelerate to speed, then coast, then do it all over again. She hated for me to point this out.

Have you ever noticed that cruise controlled cars and manually controlled cars do not interact well. People tend to react slower than the machine which never wavers from it's attention on the task.

I've noted here how many tend to define a problem mathematically and have a lesser intuitive understanding of the process while I tend to be the opposite. I've decided that this has to do with how you learn this field. I started doing it, learning my the seat of my pants and the math came afterwards. People that go to school for controls learn the math first and the intuitive part comes later after doing it over and over. Both are important. It's just interesting to observe.

This does make me want to ask a question. How many of us had an early fascination with controls?

Be well,

HoP

The preceding message represents personal opinions and/or advice that may prove incorrect or harmful. But then maybe not. Feel free to disregard.

------- Words have no Warranty ------ ------- No View without Merit ------

Jerry Avins wrote in news:COadnfikpLMb8fbenZ2dnUVZ snipped-for-privacy@rcn.net:

Interestingly, there are some very precise open loop physiological control systems, because for some things, feedback would be much too slow, and introduce oscillations.

Very often, though, such systems have monitoring for long-term errors, and the parameters of the open loop control system are adjusted if errors persist.

Indeed: that's what reflexes are about. "Reflex" literally means "bend back" and a reflex is a pathway from afferent (sensory) nerve to spinal column to efferent (motor) nerve, without traveling the round to and through the brain. As far as I and my internist know, I deduced one that can be traces but hadn't been noted before.

Given the speed of nerve impulses, one jumping from a six-foot with knees slightly bent to avoid a jarring impact would knee himself in the chin if the brain had to signal when to stiffen the leg muscles relying on the sensation of contact for input. Instead, a reflex is primed by the ankle being bent downward and an inhibitory signal being removed from the priming synapse. The ankle straightens when contact is made, and the signal from it goes no further than the base of the spine. The reflex there briefly tenses the leg muscles open loop until the brain has time to take over.

Sometimes the pathway is inappropriately activated, and the result is usually a calf cramp (sometimes thigh). Forcing the ankle to a right angle by standing immediately releases a cramp of this origin.

Jerry

Jerry, Be careful what you suggest. I was driving home tonight and the person in the lane next to me apparently read your post and was trying it. :-}

John

In processes there are, on occassion, loops that are sufficiently self regulating, and with no requirements for tight control, so no feedback is needed. Sometime a manual loader will allow the operator to position the valve, or in some cases no valve is needed.

However, these are rare. Most of the time feedback and closed loop control is required

John

Ouch!

Jerry

With me it supposedly started when I began migrating from the horizontal floor crawl to vertical surfaces. Like the TV or the coal stove. So ... what's this little thingie do if I wiggle it ... usually followed by a loud scream from mom "Nooooo!".

One control where I realized early on that it wasn't supposed to be messed with was the 110V/220V selector, if you are in a 220V country. Phfffssssst ... POOF.

Regards, Joerg

On Sat, 05 Nov 2005 02:16:25 GMT, Joerg proclaimed to the world:

I had a cousin that came over and took disassembled all my toys leaving them for me to figure out how to reassemble them. Perhaps this contributed a little for me.

A common service call I get now is to fix some factory's machine that has already been picked apart by a half dozen others in an attempt to fix it. Deja vu?

Be well,

HoP

The preceding message represents personal opinions and/or advice that may prove incorrect or harmful. But then maybe not. Feel free to disregard.

------- Words have no Warranty ------ ------- No View without Merit ------

Of course, sometimes it is harder to see what the feedback mechanism actuially is. It comes in a lot of guises.

An example:- Signal into amplifier, amplifier to loudspeaker audio to ear. This may seem like a very open loop system but there are a number of feedback paths that are attempting to keep matters under some sort of control.

Each amplifier stage has its own feedback loop to provide just the level of gain set by the "gain demand" controls (so that is at lerast two). Then there is the loop that includes the listener. He will adjust the level to a point that he desires to hear the signal. Beyond that there may be the neighbours who begin to bang on the walls, ceiling or floor when the level of sound becomes too much for them to bear. So, as you see, you may have to consider expanding the boundaries of the system to see what may be the overall feedback mechanisms in play.

Naturally, you will select a system boundary so that you can proiperly consider the peice of the overall system you are really interested in.

If there is perfect and complete knowledge of the plant and any disturbances at all times, you could use open loop control. Unfortunately, that isn't a realistic situation. So you need some form of feedback. It's reality that gets in the way.