April 24, 1997

Updates and corrections

In the article, "Good things come in small (IC) packages" (EDN, Feb 3, 1997), the hole sizes listed in milli-meters should have been listed in mils in the box, "What gets bigger, the more you take away?" (pg 76).

In the Leading Edge article, "Near-field recording promises 10-times areal-density increase" (EDN, March 27, 1997, pg 11), the phone number listed for Tera-stor Corp is actually the company's fax number. The phone number is 1-408-324-2110. EDN apologizes for any inconvenience.

Analyzing bearing analysis

Your article "Data acquisition quietly joins the communications revolution" by Dan Strassberg (EDN, Feb 17, 1997, pg 97) assumes that, in a remote data-acquisition system for bearing analysis, the analysis must be done continuously for all points measured. If this assumption is valid, the intelligent sensor with embedded FFT makes sense. However, I don't think the assumption is valid. Bearing failures are predictable long before actual failure.

The cost-effective method for doing bearing analysis online is to use ordinary analog accelerometers for a machine or group of machines, connected to a no-frills, minimum-intelligence, data-gathering processor for the group of sensors. This device would send raw data over a digital link (Fieldbus, EIA-485, or LAN, for example) to a central PC that would do the FFT operations, comparison to stored signatures, and operator notification. Data would not be sent continuously but only on command from the central processor. Thus, the maximum network data rate would be that of one sensor only and would be well within the capabilities of ordinary desktop computers.

Such a system could do one analysis every 1 to 10 sec, depending on the depth of analysis required and the speed of the processor. With a 10-sec analysis and 6000 bearings in a plant, each bearing would be checked at least every 16.7 hours. This time is still short compared with the usual frequency of manual measurements and the time to schedule a shutdown.

If an anomalous data packet is received from a monitoring location, software can automatically recheck it. The bearing-monitoring system should have input from the plant-monitoring and -control system to determine if a machine is running or stopped or if the machine is in a portion of its operating cycle that does not require the motion of the monitored bearing. This machine-status information can arrive via a network link from the main monitoring computer or may be derived from relay contacts on the machine's own control system that are sensed by the bearing-monitoring system. If a monitored bearing is inactive when the central PC checks it, the PC can defer the reading until the bearing is active.

If a bearing rechecks as anomalous, intervention is necessary in any case. Whether it is the bearing, the sensor, or the supporting system that needs service, it still requires service, and that means on-site intervention. Remote sensing doesn't imply remote repair.

Phil Spray
Mason & Hanger Corp
via the Internet

Senior Technical Editor Dan Strassberg responds: Some of the author's points are well taken. However, I disagree that making scheduled readings (say, enough readings to get one FFT per machine every 24 hours) would sufficiently reduce the network traffic to make centralized data processing practical. Even with daily monitoring, distributed intelligence would be preferable to centralized processing. However, before deciding on the system configuration, the system designer should perform a detailed study of the traffic that sending the raw data to a central processor would generate.

Sound off

Send your letters to Signals and Noise Editor, EDN, 275 Washington St, Newton, MA 02158 or e-mail us at bmorrison@edn.cahners.com. Our fax is 1-617-558-4470. EDN reserves the right to edit letters for clarity and length.

| EDN Access | Feedback | Table of Contents |