You need more specifications for the MEMS accelerometers; such as frequency range (min - max.), mounted natural frequency, and output range (min. - max.). The sound sensor (microphone) should be specified in dB-SPL (20 micro-Pascal pressure reference) and and not a voltage reference. The rest of the information from Party A and B make little sense! Will you be analyzing the output (analog voltage or digital) with your software or just looking at spectral data from the sensor OEM? You need the bearing fault frequencies andany other periodic forcing frequencies related to shaft speed.

Some resources:

MEMS Accelerometer

explain volts/LSB for mems accelerometer - Bing

Accelerometer Specifications - Quick Definitions

From <https://www.analog.com/en/prod...ons-definitions.html>

How to Build a MEMS-Based Solution for Vibration Detection in Condition Monitoring

From <https://www.analog.com/en/tech...-based-solution.html>

ADXL357 Datasheet and Product Info | Analog Devices

Walt

Thank-you Walt for your timely reply.

Pointwise response;

[Walt] - You need more specifications for the MEMS accelerometers; such as frequency range (min - max.), mounted natural frequency, and output range (min. - max.).

[Tony] - All i have from the existing Wireless Sensor Spec is attached. No reference to mounted natural frequency. Sensors are typically epoxied mounted but for this application the intention will be to stud mount them considering the low speed.

[Walt] - The sound sensor (microphone) should be specified in dB-SPL (20 micro-Pascal pressure reference) and and not a voltage reference.

[Tony] - All i have for Acoustic and Ultrasonic is attached. Sensitivity appears to be 94 dB SPL @ 1 kHz.

[Walt] - The rest of the information from Party A and B make little sense! Will you be analyzing the output (analog voltage or digital) with your software or just looking at spectral data from the sensor OEM?

[Tony] - Both i believe. Digital will be analysed and the software is capable.

[Walt] - You need the bearing fault frequencies and any other periodic forcing frequencies related to shaft speed.

[Tony] - Yes bearing numbers are known and fault frequencies will be determined.

@Walt Strong and others, the queries are mainly;

1. Is the existing MEM sensor with specs provided a good/adequate sensor and suitable for this Low Speed (~6 RPM) application?
2. Can the combination of Vibration and Acoustic monitoring with a Wireless sensor operating 24/7 provide the Client the ability to detect a fault?
3. What guidance on good practice can be provided in using this existing MEMS based Wireless sensor for this low speed application.

I am no expert and hence why i joined Maintenance.org to seek the support of experts in this field to support a Client.

Thank-you for your Guidance.

Tony

Please can anyone provide a learned response?

Thank-you in anticipation.

T

Before diving into spectral data configuration and FFT frequency resolution which I probably don't understand anyway, let me key in on a very important statement in your request, "even with a shaft fault existing such as misalignment or unbalance, the machine can live for years". This leads me to believe a simple PM to replace the bearing every 2 years may be preferred to trying to define a questionable low speed vibration monitoring system? Also, what is the COF (Consequence of Failure)? If this is a mixer or agitator that can be easily shut down for repair or especially if it's redundant, it may be more cost effective to run to failure or install one of the old "earthquake switch" devices on it to indicate a problem. Bottom line what's the ROI? Always invest your money and time in the best return.

What is the OEM and model of the digital accelerometer that you have?

You can compare some of the critical specifications of the digital accelerometer to piezoelectric accelerometer such as this one:

Typical Piezoelectric Accelerometer with ICP power

CTC AC203 Series Low & High Frequency Accelerometer

AC203 Datasheet (ctconline.com)

Sensitivity: 100 mv/g

Frequency Response: (+/- 3 dB) 6-600,000 cpm or 0.1-10,000 Hz

Dynamic Range: +/- 80 g

Spectra Noise in micro g per square root Hz: 1.3 @ 10 Hz, 0.2 @ 100 Hz, 0.1 @1000 Hz

Temperature Range: -58 to 250 F (50 to 121 C)

Resonant Frequency: 18,000 Hz

Mounting: 1/4-28 stud for tapped hole or with adhesive mounting disk

The digital accelerometer has one third the sensitivity, and it has over 800 times more noise spectral density than the piezoelectric accelerometer. The digital accelerometer has a lower temperature limit. The piezoelectric accelerometer is superior in many specifications.

Bearing fault detection for a shaft at 6 rpm can be a real challenge for the following reasons:

1. It may only be possible to detect a significant bearing fault rather than an early warning of a developing fault.
2. Low frequency and amplitude acceleration values for bearing fault frequencies.
3. Bearing vibrations may not be detectable with high vibrations from accelerometer noise and from other machine components or other machines.
4. A long measurement time of 5 to 10 revolutions is needed which would be 50 to 100 seconds of data per measurement. The vibration analyzer must have long time record measurement capability.
5. Vibration signal analysis capability should include overall peak acceleration level, time waveform, FFT spectrum, and demodulated (envelope or PeakVue) spectrum.
6. An accurate shaft speed measurement that may require multiple pulses per revolution.

Here is some additional information:

Low Speed Bearing Fault Detection

Search: bearing vibrations on low speed shafts

bearing vibrations on low speed shafts - Bing

Low Speed Bearing Analysis

ABSTRACT: (vibration.org)

Low Speed Bearing Analysis – Vibration Institute (vi-institute.org) purchase of same article?

Low Speed Trunnion Bearing Problems -- good setup information

Microsoft PowerPoint - P2 Low Speed Trunnion Bearing Problems (rms-reliability.com)

Condition monitoring of low-speed bearings - A review

From <https://www.researchgate.net/p..._bearings_-_A_review>

An Overview of Bearing Vibration Analysis

Vibration brochure (final art) (schaeffler.com)

Slow speed bearing defect detected though vibration analysis

From <https://theseasonedanalyst.gur...-vibration-analysis/>

Low-speed bearing monitoring Challenges and Solutions

PowerPoint Presentation (bearing-news.com) by SKF

Walt

Thank-you Walt.

I appreciate the detailed response.

I believe the spec of the underlying accelerometer within the sensor is this: Tri-Axis, ±40g, 3.3V, Analog, 3x3x0.9mm LGA - KX220-1072 - global (kionix.com).

Thoughts Walt on this unit?

The differences in performance between a MEMS vs Piezoelectric Accelerometer are well documented. In comparison to their weaknesses, MEMS sensors do provide other benefits.

It will be a challenging use case undoubtedly. Success is unclear as interpreted within your response.

You mention a measurement time of 5-10 revolutions. Other papers mention 15-20. This would be a significant amount of data!

Are you aware, or can you outline, the best process to follow in collecting results worthy of analysis?

Thank-you again Walt and Vic for your responses.

If there is any other information for consideration considering the limitations of the sensor and complexity of the use case, i would really appreciate other advice for consideration.

Tony

The Kionix company does not appear to be heavily involved with machine condition monitoring applications. All of the triaxial accelerometers on their website appear to be labeled "not for new designs". The accelerometer you have appears to be an old design. There are better performing MEMS accelerometers on the market, such as available from Analog Devices that have considerably lower noise.
There is no "best practice" for low-speed bearing vibration measurements that I'm aware of. There are plenty of vibration monitoring pitfalls that could cause false alarms or no fault detection at all until complete failure occurs; based on reasons previously mentioned. The accelerometer choice is important, but the real challenge can be finding a suitable signal analysis method including overall levels, waveform, and spectrum. I cannot confirm if the product you have will provide an adequate warning prior to the shaft stops rotating or visible smoke appears!

"or install one of the old "earthquake switch" devices on it to indicate a problem" This total piece of junk for this application!!

Walt

What's the COF? What's the cost of downtime? What's the ROI? I don't need FFT & modal analysis techniques to replace the bearing on a wheel barrow. I'll just replace it when it squeaks. Again, the author emphasized the asset can operate for years out of balance or misaligned. So, my level of concern is minimal. There needs to be a reason or return to invest more money. By the way, I reviewed a bunch old WWII oil flooded recip's once in China and the old earth quake (ball & pivot) mercury switches provided sufficient protection >99% reliability for 50+ years. No financial return to invest in anything else.     

"WWII oil flooded recip's once in China and the old earth quake (ball & pivot) mercury switches provided sufficient protection >99% reliability for 50+ years." Does this indicate the vibration switches detected every machine fault/failure, or there were no machine failures, or the switches never activated (including no earthquakes) in 50+ years?

I am proactive with my wheelbarrow (garden carts) by keeping the bearings lubricated. The mining machine in the OP probably has a greater COF, cost of downtime, and ROI compared to a wheelbarrow!

Walt

So, that's what this is. Well, if the downtime cost justifies the investment, then so be it.

Thank-you Walt and Vic. Your learned and timely responses are very much appreciated.

Tony