1
\$\begingroup\$

I am really confused with the difference between steady state error and actuating error. My teacher gave the following notes on this topic and explained using 1 example. He said that there will be 2 types of problems

  1. Case of non measuring and E(s) not specified in the problem
  2. Case of measuring or when E(s) is labelled in the diagram image1 image2 image3

Other teacher explained saying that

  1. Case of steady state error -> E(s) is specified in block diagram
  2. Case of actuating error -> E(s) is not specified

He showed this below problem and said it is a case of Actuating error.

image4 image5 I am also attaching his youtube link here https://www.youtube.com/watch?v=IsCtCgNE6S0&t=235s

Please explain this topic clearly and guide me on how to attempt problems like this.

\$\endgroup\$
  • \$\begingroup\$ The static error is when \$t\$ approaches infinity or when \$s=0\$: you crank the control system, wait, and you measure the difference between the setpoint and the corresponding output : you read 4.995 V on a power supply output while you expected 5 V. The static error is 5 mV. You reduce the error to theoretically 0 when you place a pole at the origin (divide by \$s\$). Practically, the dc gain is not infinite but bounded by the compensator max gain. The dynamic error is more during a transient in the setpoint: the error momentarily deviates from the target until it catches up later. \$\endgroup\$ – Verbal Kint Dec 7 '17 at 12:07
1
\$\begingroup\$

I don't like the two definitions, but let me try to explain them. I could be wrong or others may make improvements or suggestions or not.

1) Non-measured error If you do measure and compare it, then it is not possible to control it with feedback and measure the error

  • But it may be possible to constrain the variables as in Feed-forward designs with an output regulated by some constrained ratios by regulating the input. ( However external production testers verify the outputs to meet certain specs.)

    • In fact this is how PC ATX PSU's work on secondary output's by having tight ratios on all secondary outputs with only a primary voltage feedback design on the one with the highest steady power load. ( Usually 5V)
    • In fact every electronic component is a result of 3 control systems with specs to indicate the limits or variables or fine print footnotes for conditions. Most are tested and some are not and constrained by design tolerances.

The 3 control systems are : Design, Process and Quality systems each with measurable inputs/outputs and verified outputs with error limits. Hint: all the best designs start with good specs. Good statistics are useful and unavoidable.

2) Measured errors (Steady State or otherwise)

You might be able to measure the error, but unable to control or eliminate it to some acceptance criteria due to undue constraints or you can measure it and have well defined targets for ramp, step, impulse (finite) or sinusoid.

Steady State error assumes there is no input change and the response does not change after an extended period of time and no measurement errors.

Actuator Error is "not specified" because it is to vague and can be caused by anything, measurable or not but presumed to be operating within its design environment and load.

Examples

This could be; Position error, velocity Error, voltage error, acceleration error, pressure error, impulse error, timing error, jitter error, phase error.

What is an actuator error?

I say it is the difference between an input and the measured feedback response + undetectable error ( but exists) + uncertainty in the feedback measurement. ( which can be anything from dynamic to steady-state)

My error signal list includes;

  • dynamic, steady-state, unstable ( oscillating, or noisy)
  • measurable or not measurable but known by effects and deduction
  • from measurement error or actual measured error
  • a result of a change in desired input or an initial condition.
  • a result of an undesirable disturbance , transient, step, impulse or steady-state or initial condition.
  • a result of sub-optimal or optimal system design for loop gain and phase response.
  • a result of unknown causes or known causes from failures in component(s), software code errors, system failure, design error, cost reduction, compromises in performance, aging, degradation,
  • caused by environmental stress, e.g.; ground-benign, automotive, airborne, seaborne or aerospace ( climatic, mechanical, electrical, electromagnetic etc)
    • environmental stress such as a pothole, an arc welder, a geomagnetic storm, thermal shock, earthquake or transport vibration, low air pressure from high altitude, radar pulse and ESD discharge, a mobile phone RF sync packet , etc.etc.etc.

It can also be nominal or within error limits with x-sigma deviation or guaranteed within limits under specific conditions like 25'C and other limits for specific environmental limits, or have some error sensitivity with temperature or any other variable.

What did I leave out?

Cost control errors? Quality errors (escapes), MTBF errors?

\$\endgroup\$
  • \$\begingroup\$ Thanks for answering @Tony Stewart. I am not able to understand what you wrote. Things seem to be beyond my level. I have just completed my EEE graduation. Can you please explain in simpler terms. \$\endgroup\$ – user170930 Dec 7 '17 at 6:33
  • \$\begingroup\$ Which part did you understand? Didn't you take any Control Systems courses? \$\endgroup\$ – Sunnyskyguy EE75 Dec 7 '17 at 6:53
  • \$\begingroup\$ Here are some academic ( incomplete from real world) examples ee.usyd.edu.au/tutorials_online/matlab/extras/ess/ess.html \$\endgroup\$ – Sunnyskyguy EE75 Dec 7 '17 at 6:56
  • \$\begingroup\$ No I didn't. I was only taught to solve problems like the 2 examples I mentioned. If E(s) was given then do R(s)-B(s), if not given then do R(s)-C(s). \$\endgroup\$ – user170930 Dec 7 '17 at 7:30
  • \$\begingroup\$ You know Op Amp is a simple control system with gain based on impedance ratios. with an internal integrator compensation so that and steady state error depends on feedback ratio * forward gain, now apply the same principle with (normally) unmeasured errors coming from CMRR PSRR input offset bias etc. because they are not shown in the boxes, but cannot be ignored. \$\endgroup\$ – Sunnyskyguy EE75 Dec 7 '17 at 20:37

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy