Browse Prior Art Database

SMOOTHING ALGORITHM FOR VEHICLE AND ENGINE SPEED DETECTION THAT REMOVES GEARBOX AND SENSOR JITTER

IP.com Disclosure Number: IPCOM000009160D
Original Publication Date: 1999-Jun-01
Included in the Prior Art Database: 2002-Aug-12
Document File: 5 page(s) / 191K

Publishing Venue

Motorola

Related People

Matthew Emberson: AUTHOR

Abstract

Existing speed measuring systems suffer from noise in the alignment of the sensor poles and in gearbox induced cyclic variations (most people are familiar with the sudden jittery movement of the speedometer under acceleration or the gentle sway- ing of the pointer at constant speed). Modem sys- tems use multi-pole sensors that produce rectangular waveforms, the period of which is dependent on the road speed (one example produces one edge every 20cm of vehicle travel). Examples 1 and 2 below illustrate the typical output of these sensors with no jitter or cyclic variations. Example 3 below illus- trates typical 8 pole jitter. This is caused by the 8 sensors around the shaft not being evenly spaced. The result is that for constant speed, some edges are too early and some too late, however the cycle is accurate for a full rotation from any given angle round to that same angle.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 44% of the total text.

Page 1 of 5

MOTOROLA @ Technical Developments

SMOOTHING ALGORITHM FOR VEHICLE AND ENGINE SPEED DETECTION THAT REMOVES GEARBOX AND SENSOR JITTER

by Matthew Emberson

BACKGROUND

  Existing speed measuring systems suffer from noise in the alignment of the sensor poles and in gearbox induced cyclic variations (most people are familiar with the sudden jittery movement of the speedometer under acceleration or the gentle sway- ing of the pointer at constant speed). Modem sys- tems use multi-pole sensors that produce rectangular waveforms, the period of which is dependent on the

Example 1. Sensor output at constant speed

road speed (one example produces one edge every 20cm of vehicle travel). Examples 1 and 2 below illustrate the typical output of these sensors with no jitter or cyclic variations. Example 3 below illus- trates typical 8 pole jitter. This is caused by the 8 sensors around the shaft not being evenly spaced. The result is that for constant speed, some edges are too early and some too late, however the cycle is accurate for a full rotation from any given angle round to that same angle.

Example 2. Sensor output during acceleration

Example 3. Sensor output for constant speed with example 8 pole jitter

One full rotation of sensor

Q Motmola,l"c. Is99 103 June 1999

[This page contains 14 pictures or other non-text objects]

Page 2 of 5

MOTOROLA Technical Developments

PROBLEM

  The problem then, is to convert this signal into a smooth, accurate reading suitable for use by the engine management system (where both vehicle speed and engine RPM are required) and for the instrumentation where a pleasing movement of the pointers is a necessity. Errors in RPM caused by such jitter can lead to poor engine performance, and warranty on speed indication systems is high for many vehicles.

  An accurate reading can broadly be defined as one that tracks actual speed as close as possible, with minimum overshoot and lag. A mean sampling system must know the frequency of the variations to be effective. Consider sine-based variations. Provided the variations are sampled over an integer number of cycles, then the sine variations will can- cel out. However, real world variations are a com- pound of sensor alignment error, gearbox eccentrici- ties, axle eccentricities, wheel misalignments etc., therefore it is not possible to know what the input variation frequency is, and hence the sampling peri-

Example 4. Sampled sine curve with (znin+max)/Z filtering

od cannot be known. Because of this, mean sam- pling systems fail at low speeds where only part of the cycle can be sampled. Such algorithms also react less well to low frequency variations for the same reason.

SOLUTION

  One of the key properties of jitter and gearbox related variations is that they are cyclic. The key to removing jitter comes by looking not at the mean, but at the maximum and minimum periods of the waveform in the sample time. Consider the sine wave example again. For the 0 to x solution, the min. and max...