As a member of the Sound and Vibration (S&V) Committee, receiving questions from NEBB S&V firms and CPs about sound data, vibration data, and instrumentation is common. Below, I will describe three instances which raised concern regarding a basic understanding of how S&V measurements are made and instrumentation is set up.
Because everyone makes mistakes periodically and ends up with useless test data, we must repeat work. By using this article as a guide to revisit and refresh our understanding of how we should be doing our jobs, we overcome three issues:
- The proper instrumentation setting to take sound data to generate noise criteria (NC) and room criteria (RC) curves.
- Instrumentation set-up for vibration measurements so usable data can be taken.
- The issue of sound level meter calibration requirements.
Error #1: Sound Level Measurements for NC/RC Curves
Those who have taken the Sound CP or CT technical seminars know the instructors stress the proper setting on a sound level meter when taking sound data to generate NC and RC curves or ratings. Most sound specifications call for one or two measurements: either the overall sound level, which is typically an A-weighted decibel (dBA) measurement, and/or the NC/RC rating, which is an octave band measurement. The octave band measurement for NC/RC must always have the instrument set to have the frequency weighting of unweighted (dBZ).
In accordance with the 2015 NEBB Sound and Vibration Procedural Standard, a sound level meter or analyzer should be set up as follows:
- Frequency Weighting: Z, linear, flat (this designation is different for different instruments)
- Time Constant: Slow
- Full Octave Data: From 31.5 to 8,000 hertz, or Hz. Note: The labeling for the unweighted setting may vary from manufacturer to manufacturer and may be dBZ, dB(flat), or dB(linear). Read your instrument’s Owner’s Manual.
Overall, this isn’t difficult, but it brings up a nuance, which depends on the manufacturer of the sound level meter. Consider these two examples:
- Manufacturer A: instrument configuration has set-up options which apply the frequency weighting (dBA, dBC or dBZ) to both the overall sound level and the octave band sound levels.
- Manufacturer B: instrument configuration has set-up options which allow the user to set the overall sound level frequency weighting (dBA, dBC or dBZ) and then independently set the octave or third-octave band frequency weighting (dBA, dBC or dBZ).
The error arises when both overall sound levels and octave band data are to be measured. If the instrument is not configured correctly, the A-weighting may be applied to the octave band sound level readings. So, what does this error really mean when the octave band data is measured A-weighted instead of un-weighted?
In the NEBB Sound technical seminar, attendees learn that when a measured sound source level is within 10 dB of the background sound levels, a correction for background sound levels must be done before the NC or RC plots are made. For example, a correction can be made for the 125, 250, 2000 and 4000 Hz octave bands, although a correction in the 125 Hz octave band is sketchy since it is so close to the background.
What can be done with the 31.5 and 63 Hz octave bands? The answer is nothing since the A-weighted data is below the background levels, and there is no method to make a correction. The data is not usable, and the measurements must be repeated. This is analogous to jumping into a swimming pool; if the water is clear, you can see where to jump to miss the obstacle under the water’s surface, but if the water is muddy, you don’t know where to jump.
Take-home point: Set the sound level meter or analyzer up correctly for the measurement which needs to be made because you may not be able to recover the test data, which leads to having to make measurements again, increases cost, and lowers profit.
Error #2: Obtaining Useful Data from Vibration Measurements
Those who have taken the Vibration CP or CT technical seminar know one of the topics covered is the frequency resolution of the vibration measurements. The term frequency resolution, in vibration lingo, is the distance in Hz between two adjacent data points in the digital Fourier transform.
When setting up the vibration instrument for vibration measurement, the frequency range (bandwidth) is selected, and the number of Fast Fourier Transform (FFT) lines is selected. The frequency range is the highest frequency (Fmax) minus the lowest frequency (Fmin). For example, with a measurement from 0 to 1000 Hz (60 to 60,000 rpm), Fmax is 1000 Hz, and the bandwidth is 1000 Hz. The number of FFT lines is just that—a number that always comes in multiples of two, i.e., 200, 400, 800, 1600, etc.
The frequency resolution is the maximum frequency times the window factor divided by the number of lines. For this discussion, we will ignore the window factor.
Frequency Resolution (∆f) = (Fmax * Window Factor) / Number of Lines
If Fmax is 1000 Hz and the number of lines is 200, the frequency resolution is 5 Hz. We only get to see information every 5 Hz, meaning we have no information about the amplitude of the vibration at, say, 17 Hz.
If Fmax is 1000 HZ and the number of lines is 800, the frequency resolution is 1.25 Hz. Typically, for machinery vibrations, a 1 to 1.25 Hz resolution is sufficient to diagnose all vibration-related issues.
Let’s set up a scenario of making a measurement on an electric motor/fan assembly with the following parameters:
- Motor speed = 60 Hz, four poles will run at ~1800 rpm (30 Hz). Due to slippage, the motor will actually run a bit slower. A 5 hp motor will have ~3 % slip, which results in an operating speed of 1746 rpm or 29.1 Hz. Also, assume a direct line connection, no VFD.
- Motor to fan sheave ratio = motor 8 inches/fan 10 inches.
- Fan speed = motor rpm times 8/10 inches = 1,397 rpm (23 Hz)
- Number of blades on the fan = 11.
- Blade pass frequency = fan speed times the number of blades = 15,365 rpm (256 Hz)
In NEBB vibration work, the frequencies of interest are typically one times (1x) rpm and harmonics and the blade pass frequency (BPF) of a fan or the vane pass frequency (VPF) of a pump. Here, we are interested in the following frequencies: 24, 29 and 256, and maybe the 2nd harmonic of each (48, 58 and 512 Hz).
With this information regarding the frequency range of interest, we can set up our vibration instrument several different ways—all of which are acceptable for this application:
- Fmax = 800 Hz, number of lines = 800, resolution = 1 Hz.
- Fmax = 1600 Hz, number of lines = 1600, resolution = 1 Hz.
- Fmax = 1000 Hz, number of lines = 800, resolution = 1.25 Hz.
With each of these setups, the frequency resolution is sufficient to measure vibration information at the frequencies of interest. Which will then allow us to assess the severity of the measured vibration levels.
What if we made a mistake and set our meter off with a much lower resolution?
Example 1: Fmax = 1000 Hz, number of lines = 200, and resolution = 5 Hz. Here, we only measure data at frequencies which are multiples of 5: 5, 10, 15 and so on. And, we have missed the 1x operating frequency of the motor and fan and the blade pass frequency.
Example 2: Fmax = 2000 Hz, number of lines = 200, and resolution = 10 Hz. In this case, we only measure data at frequencies which are multiples of 10: 10, 20, 30, 40, 50, 60, and so on. And, we have missed the 1x operating frequency of the motor and fan and the blade pass frequency.
Picture the vibration spectrum from a two-cylinder reciprocating refrigeration compressor with frequency resolutions of 1 Hz, 5 Hz and 10 Hz. With the 5 Hz and 10 Hz resolutions, there is a significant loss of information regarding the health of the compressor, both amplitude and frequency information. In this case, the data loss would be significant since the vibration report would show that the compressors are running smoother than actual.
Take-home Point: Set the vibration analyzer up correctly for the measurement which needs to be made because when you don’t, useless data is obtained, leading to making measurements again, increased cost, and lowered profit.
Error #3: Instrumentation Compliance and Calibration
NEBB has instrumentation requirements we use to do our jobs and a list of instruments for each instrument type, which comply with NEBB requirements. Every year, we must send these instruments in for a National Institute of Standards and Technology, or NIST, traceable calibration. Periodically, one of our instruments does not comply with NEBB requirements, and we must replace the instrument or the sensor attached to the instrument or send the instrument back to the factory for internal adjustments.
Side note: The NEBB requirements for sound level meters and analyzers are a subset of the American National Standards Institute, or ANSI, Type 1 (Class 1) and Type 2 (Class 2) specifications. Manufacturers indicate their instrument meets ANSI Type 1 or Type 2, which then implies that the instrument meets NEBB requirements.
This discussion applies to models and manufacturers of sound level meters and analyzers which meet NEBB requirements and are on the NEBB list of pre-approved instruments list, and do not pass an annual calibration. Why would an approved instrument not comply with NEBB requirements?
As a sound and vibration instrumentation calibration laboratory, most of the instruments we calibrate yearly comply; however, there are some which are not in compliance. The most common reason an instrument is not in compliance is degradation of the microphone, and many times in the lower-end sound level meters, they most likely did not come out of the factory in compliance.
Why would an instrument come from the factory uncalibrated or not in compliance with the ANSI standards the instrument is advertised to meet? The issue might be cost. Calibrating every instrument takes time, and if a large enough fraction of the customers do not require the instrument to be in compliance, the company can save money. (Note: This is only my opinion.)
As a calibration laboratory, we see this issue only with the lower-priced sound level meters that are purchased online or from a wholesale distributor. As a standard practice, these manufacturers do not factory calibrate the meters or analyzers prior to shipment unless a customer specifically requested and paid for this factory calibration. Also, this requested calibration is an additional cost.
As a matter of practice, whenever we purchase any instrument for NEBB-related work, we cannot necessarily rely on the vendor or distributor to know enough to ensure we get a calibrated instrument. We need to do our own due diligence and explicitly ask if the instrument comes with a NIST traceable calibration, and if not, explicitly ask for one.
I hope this article has helped clarify some nuances of NEBB sound and vibration instrumentation and measurement practices. Remember, we humans are about 60% water, and the first resonance of a water molecule is about 22.235 GHz. So happy vibrating.
Looking to get NEBB Certified? Request your application today.