ForthNet@willett.UUCP (ForthNet articles from GEnie) (01/05/90)
Date: 12-21-89 (09:52) Number: 53 (Echo) To: ALL Refer#: NONE From: CLIFFORD KING Read: (N/A) Subj: WHAT'S DSP Status: PUBLIC MESSAGE DSP Message 2 Reposted What is DSP? For an introduction, see previous messages in this thread. Hello again, Digital Signal Processing algorithms are used anytime you read an input and must make a decision based on that input. The input data can be 1 bit (yes or no), an integer number, or a floating point number. A trivial example of DSP might be to read a sensor and then based on whether the sensor read true or false, turn a relay or off. What does this have to do with digital signal processing, you say? Aren't I just trying to redefine real time control as digital signal processing? Yes, you could say that is what I am doing. However, it is true that you are processing a digital signal and producing a new output based on the input values you have read so far. That is what digital signal processing is all about. In the trivial case, I just mentioned, there is no need to delve further into the field of digital signal processing. However, if you wanted to control the action of the relay beyond simply turning it on or off each time you go thru the process loop, say to ensure that the relay once turned on stays on for a minimum period of time, then some additional processing is required. Usually a feedback loop must be created that looks at the current input value as well as previous input values. Once you establish a feedback loop no matter how trivial it may be, you are solidly in the realm of digital signal processing and all of the algorithms and analysis methods available. Many of you are possibly aware of "PID" (Proportional, Integral, Derivative) loops. These are a simple form of digital signal processing based on error control techniques which require the current input as well as past input values. By accumulatiing error terms, the integrals, and the derivatives of the error terms, an output function is created that can be used to control a motor, a relay, or whatever device you are operating. More sophisticated feedback loops employ filtering algorithms to accomplish the same thing as PID loops. However, the mathematics and the algorithms are much better documented for filtering algorithms than for PID loops which tend to be initialized by the preconceptions of the person programming the PID or by some general guidelines that may or may not be correct. This is not to say that there is anything wrong, with PID loops, it's just that you may not be able to get the PID loop to work properly with the coefficients you have chosen and you may not have any way to determine how to adjust your coefficients to get the loop stabilized. By using filtering techniques, which may require additional computations over the PID algorithms, it is easier to predict the actual behaviour of your control loop and to calculate the proper coefficients to obtain the desired behavior. Also changing the behavior of the control loop in the desired manner is easier when filtering methods are applied instead of PID algorithms. Some food for thought. If you have a need for PID algorithms, try investigating filtering algorithms instead. You might find the filtering algorithms better suited to your requirements, plus it will be easier to change the behavior of the control loop in a predictable way. If you have any questions about DSP leave me a message on the RCFB, Clifford King. NET/Mail : RCFB Golden, CO ... Vesta Technology & Denver FIG for Forth! ----- This message came from GEnie via willett through a semi-automated program. Report problems to: 'uunet!willett!dwp' or 'willett!dwp@gateway.sei.cmu.edu'