Vibration Monitoring for Rotating Machinery: When to Walk Around, When to Go Online

Every rotating machine eventually fails. The question is whether you detect the failure mode early enough to schedule a shutdown on your terms—or discover it when the rotor seizes at 3 a.m. The right vibration monitoring approach is not about buying the most expensive system. It is about matching the consequence of failure to the speed of detection. Three tiers of protection exist, and the table below lays out what separates them.

Approach	Cost per Point	Detection Speed	Best For
Handheld walk-around	Lowest	Weeks to months between checks	Non-critical pumps, fans, small motors
Loop-powered transmitter	Moderate	Continuous 4-20mA to DCS/PLC	Medium-criticality machines with existing control infrastructure
Rack-based online system	Highest	Real-time, 1 ms sampling	Turbo-machinery, API 670 compliance, safety-critical rotors

Tier 1: The Walk-Around Meter

A handheld vibration meter costs a few hundred dollars and lives in a technician's tool bag. You walk the plant on a route—weekly, monthly, quarterly—and record a number at each bearing housing. The Extech VB400 pen-style vibration meter gives you acceleration and velocity at IP65, small enough to carry alongside your multimeter. For machines where a failure means "we swap the spare pump and rebuild this one next week," this is enough.

The limitation is obvious: a bearing can go from normal to destroyed between two monthly readings. If you walk the pump every 30 days, you have a 30-day window where a failure can go undetected. For a cooling water circulation pump, that may be acceptable. For a 50 MW steam turbine, it is not.

Technician using a handheld vibration meter on a motor bearing housing — A handheld vibration meter on a routine walk-around route. Data is only as good as the route interval—a bearing that fails 3 days after the monthly check will run to destruction before the next reading.

Tier 2: The Continuous Transmitter

A 4-20mA vibration transmitter bridges the gap between handheld and rack-based. It mounts permanently on the machine, sends overall vibration level to your existing DCS or PLC, and alarms when the trend crosses a threshold. You get continuous coverage without buying a dedicated monitoring rack.

Two transmitters in our catalog illustrate the range:

Wilcoxon PC420A: A loop-powered 4-20mA vibration transmitter outputting peak or RMS velocity. Wire it into any analog input on your existing control system. For pumps, cooling tower gearboxes, and motor-fan sets where you want continuous trending without a separate vibration rack.
Wilcoxon iT301: Dual 4-20mA outputs with Modbus/RS485. It transmits both overall vibration and temperature from a single accelerometer, giving you a second parameter for the price of one transmitter. Useful when you suspect temperature rise precedes bearing failure in your machine.

The trade-off: a transmitter gives you one number—overall level. It cannot tell you whether the vibration is coming from imbalance, misalignment, a bad bearing, or a loose bolt. For that, you need the waveform, and for that you need Tier 3.

Tier 3: The Rack-Based Online System

A rack-based monitoring system like the Bently Nevada 3500 series samples vibration signals thousands of times per second and analyzes the full waveform—not just the overall level. This matters because different failure modes produce different frequency signatures. Imbalance lives at 1× running speed. Misalignment shows up at 2× and 3×. A bearing defect generates a high-frequency ring that a 4-20mA transmitter's RMS averaging buries completely. You cannot diagnose what you cannot see.

The 3500 is a modular rack: you populate it with the monitor modules your machine needs, not the ones Bently Nevada ships in a default bundle:

3500/42M Proximity Monitor — 4 channels, eddy-current probes for radial vibration and thrust position on fluid-film bearings. This is the API 670 workhorse for turbo-machinery.
3500/50M Tachometer Module — 2 channels with Keyphasor output. Provides the once-per-turn reference pulse that makes orbit plots and phase analysis possible. Without a Keyphasor signal, you cannot balance a rotor in place.
3500/61 Temperature Monitor — 6 channels, RTD or thermocouple. Bearing metal temperature is the second parameter every API 670 machine requires, and the 3500/61 brings it into the same rack as your vibration data.
3500/77M Cylinder Pressure Monitor — 4 channels for reciprocating compressors. Rod drop and cylinder pressure tell you about ring wear and valve condition—mechanical issues that vibration alone can miss on a recip machine.
3500/91M EGD Gateway — bridges the 3500 rack to your plant's Ethernet network via Modbus TCP. If the protection rack cannot talk to your DCS, the operators cannot see what the machine is doing.

Bently Nevada 3500 rack with multiple monitor modules installed — A Bently Nevada 3500 rack with proximity monitor, tachometer, and temperature modules. Each slot accepts one monitor module. A typical steam turbine train uses 4–8 modules depending on the number of bearing planes and the API 670 protection requirements.

How do I know if my machine needs API 670 protection?

API 670 applies to turbo-machinery in petroleum, chemical, and gas industry services—steam turbines, gas turbines, centrifugal compressors, and large pumps above roughly 500 hp. If your machine has fluid-film bearings and an unplanned shutdown costs more than the protection system, you are in API 670 territory. Machines with rolling-element bearings below 500 hp rarely justify a rack system unless they are safety-critical. Start with a transmitter, and only move to a rack when the failure consequence—production loss, safety risk, equipment damage—exceeds the installed cost of the 3500.

Can I mix Bently Nevada 3500 modules from different generations?

Yes, with a constraint. The 3500 rack backplane supports 3500/42M (current generation), 3500/40M (previous), and 3500/42 (original) proximity monitors in the same rack. However, the 3500/91M EGD gateway requires rack firmware revision 5.0 or later. If you are expanding an older rack that has been running on firmware 3.x, plan the gateway upgrade alongside the module adds—otherwise the new modules will report data the gateway cannot see.

What is the difference between a proximity probe and an accelerometer in a turbine application?

Proximity probes (eddy-current) measure shaft displacement relative to the bearing—non-contact, directly observing the rotor. They are the API 670 standard for fluid-film bearing machines. Accelerometers measure casing vibration—contact-mounted on the bearing housing. For a sleeve bearing machine, a proximity probe sees the rotor orbit before the vibration reaches the casing. For a rolling-element bearing machine, an accelerometer is often the better choice because the bearing transfers vibration efficiently from the rolling elements to the housing. The 3500/42M works with proximity probes. For accelerometer-based monitoring, pair a Wilcoxon iT301 transmitter with your existing PLC.

Industrial control room with monitoring systems and instrumentation panels displaying real-time machine data — An industrial monitoring panel with instrumentation displays. A rack-based online system feeds vibration, temperature, and position data to a central interface like this—giving operators continuous visibility of every monitored machine without walking the plant floor.