Practical Solutions to Machinery and Maintenance Vibration Problems
Chapter 12, Low Frequency Vibration
Section 2, Measuring Low Frequency Vibration with Accelerometers
There are two potential problems that must be considered when attempting to measure low frequency vibration using modern digital instruments that typically utilize accelerometers:
(a) Reduced sensitivity at low frequencies: The sensitivity of standard accelerometer/instrument combinations used today is significantly reduced at frequencies below the 120 - 300 cpm range (2 - 5 Hz). This is caused by electronic noise which becomes a significant component of the total low frequency signal as the very small acceleration signals diminish with frequency. Many instruments use filters to minimize the noise, but this can mask the true signals as well. The instrument supplier should be able to supply calibration charts that accurately indicate the frequency range of the particular instrument or transducer, but remember that the actual working range will also be affected by operating conditions such as mounting method, temperature and stabilization time. Special accelerometers and instruments are capable of measuring down to 6 - 10 cpm (approximately 0.1 Hz), but these are usually reserved for special applications due to their expense.
(b) Instrument integration errors: As previously mentioned, it is usually recommended that displacement units be used for measuring low frequency vibration. However, a potential problem arises when very low frequencies are received by accelerometers and converted from velocity through single integration, or displacement through double integration. Integration can cause errors on the vibration spectrum. These errors are evident as large "spikes" that occur at the lowest frequencies usually within the first three lines of resolution (sometimes called "spectral bins"). Sometimes the amplitude is high enough to force the instrument to autorange to a higher scale to the detriment of other higher frequencies on the spectrum. The effect is usually worse when using displacement units due to the "double integration error."
To verify whether the indicated peak is real or not, retake the spectrum in acceleration units (which disconnects the integration circuits), and the questionable peak should disappear.
Another way to measure low frequency vibration is to use a dial indicator to measure displacement directly. This is quite accurate up to about 120 cpm, after which indicator "bounce" may become a factor. If required, velocity can be determined from an indicator's displacement reading by using a nomograph, slide rule converter or calculations. The runout of the shaft being measured has to be considered, but this can be vectorially subtracted from the dial indicator reading.
A much better way to measure very low frequency displacement is to use proximity pickups. Proximity pickups and their associated instrumentation do accurately measure displacement down to zero frequency. Unfortunately, most plants do not, as yet, have such pickups and instruments readily available on a portable basis. However, more and more, later models of vibration analysis equipment include means for using proximity pickups along with other types of transducers. If not readily available, a dial indicator, though improvised, may be the only way to determine true displacement amplitude.
Whether using a proximity pickup or a dial indicator, shaft runout will be part of the total reading. Runout can be automatically compensated with appropriate instruments or be vectorially subtracted. Determine the runout "high spot" with the dial indicator and vectorially subtract it from the high spot when running at operating speed. As this may be inconvenient to accomplish, it helps to know that for most very small speed machines, the slow roll runout is usually very low compared to the amplitude measured at running speed and, therefore, in most situations, can be ignored.
This textbook contains only part of the information in our Practical Solutions seminar.