I think there is no constant velocity in our every-day engineering problems. Even a car moving with a seemingly constant velocity, has too many low-amplitude diminutive changes in its velocity, although it seems it is moving with a constant velocity. Hence, the sensor is not wrong in producing noise and jitter, but it is showing us the true variation in the measured signal, even if we wrongly expect it to be very smooth.
The control process, becomes unsatisfactory, because our model is inaccurate. So I think it is not due to sensor measurement noise that the controller behaves unwantedly, but degeneracy is due to our idealistic dynamic model of reality which is not accurate. Sensor usually shows the fairly true measurement although the signal seems noisy. I think the sensed signal is to some extent true, the actual signal is idealistic and wrong. The control design fumbles these inaccuracies together and ultimately gives an almost seemingly satisfactory outcome.
Please let me know other sources of noise, you think that cause jitter in the measured signal.
Thank you.
The noise generated in the sensor or picked up by it can be defined and eliminated. one of the very hard signal to remove is the DC offset and drift in the signal amplifiers.
In order to get well behaved system dynamics you need a precise and real models of the system dynamics. In order to acquire the real system dynamics with falsification of the sensors you can assume ideal sensors. Just the opposite of assuming idealized. System dynamics are adjusted by using PID controllers.
Best wishes
Sensor noise is random variations of sensor output unrelated to variations in sensor input. Ideally, the only noise coming from a sensor is thermal noise arising from thermal motions of charges within the sensor. Another low-level source of noise is shot noise related to the fact that charge is quantized. Without signal amplification greater than about 100X, these sources are unnoticeable in most cases. Another source of noise appearing in sensor outputs is electromagnetic interference from the ambient environment. Today's environment is very noisy at extremely low frequencies (ELF) from power lines, spark ignited engines, and other equipment. Of course, thunderstorms wreak havoc with sensor measurements. So, you should always electromagnetically shield sensor input circuitry (up to and including the first amplification stage), low pass filter sensor signals, and maintain good grounding systems. As part of your prototype testing for any project, use an oscilloscope to inspect sensor signals for signs of noise. It looks like "unmown grass" riding on top of your signal. You'll never get rid of it, but you can suppress it until it ceases to cause problems.
As for your control system misbehaving, note that control systems are dynamic and subject to all sorts of dynamic-system ills. Chief among them is PIO, which is related to "pilot-involved oscillation" in aircraft. It arises from a mismatch between your control system's response time and your controlled system's response time. It shows up in every active control system and manifests itself as a spurious oscillation at a (usually) low frequency that grows until it overwhelms the system's dynamic range. I do NOT know of a general prescription for avoiding it, other than minimizing the control-loop gain.
The noise generated in the sensor or picked up by it can be defined and eliminated. one of the very hard signal to remove is the DC offset and drift in the signal amplifiers.
In order to get well behaved system dynamics you need a precise and real models of the system dynamics. In order to acquire the real system dynamics with falsification of the sensors you can assume ideal sensors. Just the opposite of assuming idealized. System dynamics are adjusted by using PID controllers.
Best wishes
If the signal you want to measure is within a given frequency band, the rest can be supressed using band pass filters, low or high pass filters, depending of need. Unvanted signals is noise. In digital signal processing and sampling one can create noise, from e.g. folding down very high frequency signals giving «moire effects» with the sampling rate. So low pass, or band pass filtering can minimise that risk.
If you do not have enough signal, and it is a piezoelectric element, then it is not large enough. MEMS transducers I do not know much about, except that they porobaly are noisy, due to samll size. Best guess is that Josephson (thermal) noise is limiting the performence, like a mobile phone mic. They seldom go below 30 dBA with the noise, and it is mainly in the high frequencies they are limited. Piezoelectric sensers are also like that. The noise floor will rise towards the high frequencies.
Thank you for your comments. After reading your comments, I still think that, sensor shows a rather fairly good measurement of a sensed signal. Thunderstorms, thermal effects, and so many such other justifications for measurement noise in the sensed signal are not well-pondered, but they only justify and legitimize the source of noise because of lack of insight over it. Sensors show a fairly accurate measurement of the sensed signals, whereas the actual signal is idealistic and wrong. Hence, our dynamic model of reality is misconceived, not our sensors measurements. It is only a sophistry by the engineers to attribute noise of sensor to such effects.
Update: Other sources of noise could be our lack of insight on fluids and solids dynamic behavior. Many of the sensors and transducers, for measurement, rely on the mechanical behavior of a beam or plate, or a fluid medium. However, the mechanical analysis of these simple load carrying structures or mediums is still immature in the existing literature (even for linear mode), hence their calibration is error-prone as is their capability to quantize variation in velocity, displacement, pressure or so on.
- Badges
- Science topic
- Similar topics
- Cognitive Science
- Thinking