Ultrasonic Bearing Diagnosis: Ultraprobe, dBμV, Lubrication

Ultrasonic bearing diagnosis is a condition monitoring method using sound waves at frequencies 20–100 kHz to detect friction, impact, and leakage in rolling bearings before vibration symptoms emerge.

This technique captures high-frequency signals beyond human hearing, converts them to the audible range, and displays readings in dBμV. When a bearing runs normally, the signal remains stable. As degradation begins, amplitude spikes abruptly — typically 8–12 dB ahead of what vibration analysis can detect.

Definition and Application Scope

Ultrasonic bearing monitoring operates on the principle of acoustic radiation. Every mechanical system generates sound waves — rolling bearings produce characteristic signals that depend on speed, load, and lubrication condition. When the lubricant film breaks down and metal contacts metal, stress waves propagate through the bearing housing and can be measured from the external surface.

Application range spans electric motors, industrial fans, centrifugal pumps, gearboxes, conveyor systems, and any equipment using rolling bearings. It excels especially with low-speed bearings (below 100 rpm), the zone where vibration analysis often lacks sensitivity.

ISO 29821:2011 specifies measurement methods and signal analysis procedures for ultrasonic monitoring of rolling bearings, establishing baseline protocols and alarm thresholds used throughout industry.

Fault Detection at 40 kHz

Most ultrasonic bearing monitors operate at a center frequency of 40 kHz. This choice is deliberate: at 40 kHz, signals from damaged bearings stand out sharply against typical machinery noise floor, while attenuation is sufficient to avoid crosstalk from distant sources.

Two signal types must be distinguished:

Friction signals: Arise from inadequate lubrication. The sound resembles continuous "hissing" with dBμV rising steadily as lubricant depletion increases. This is the earliest symptom, often appearing before bearing temperature rises measurably.

Impact signals: Occur when rolling surfaces sustain damage — spalling, chipping, or cracking. The sound becomes rhythmic and periodic, matching BPFO (Ball Pass Frequency Outer race) or BPFI (Ball Pass Frequency Inner race) theory. Envelope analysis on the ultrasonic signal isolates the characteristic fault frequency.

The BPFO formula for deep groove ball bearings with n balls, contact angle α, pitch diameter Dm, and ball diameter Db:

BPFO = (n/2) × rpm/60 × (1 − Db×cosα/Dm)

Example: 6308 C3 bearing (8 balls, α=0°, Dm=65 mm, Db=15.87 mm) at 1,450 rpm:

BPFO = (8/2) × 1450/60 × (1 − 15.87/65) = 4 × 24.17 × 0.756 = 73.1 Hz

If ultrasonic spectrum analysis shows high amplitude at 73 Hz and its harmonics, the outer raceway is degrading. [Harris, T.A. Rolling Bearing Analysis, 5th Ed.] provides complete formulas for other bearing types.

Common Monitoring Equipment

Three platforms dominate the industrial market:

UE Systems Ultraprobe 15000: Dual-mode sensor — contact and airborne transducers. Frequency range 20–100 kHz, displays dBμV and real-time FFT spectrum. Includes recording for offline analysis. Best suited for route-based inspection programs.

SDT International SDT270: Specialized for low-speed bearing diagnosis. Multi-channel modules integrate with SDT Ultranalysis Suite software. Particularly strong for bearings under 100 rpm — popular in cement mills and metallurgical plants.

SKF TMSU 10: Handheld unit with magnetic sensor head. Simpler than the two above, no FFT display, shows only total dBμV. Suitable for maintenance teams wanting fast deployment without advanced training.

Beyond handheld devices, fixed systems like Emerson AMS 2140 or Fluke ii900 enable 24/7 continuous monitoring. Higher cost, but immediate fault detection independent of inspection schedules. SKF Maintenance Products Catalogue lists full technical specifications for each platform.

Sensor selection depends on bearing access. Contact transducers (magnetic or probe-tip) measure structure-borne signals through the housing — preferred for route-based programs where technicians touch a fixed spot marked on the housing. Airborne transducers capture leakage signals around steam traps and pressure valves and are not used for bearing diagnostics. For permanently installed sensors, piezoelectric acoustic emission sensors bonded near the bearing outer ring provide the cleanest signal for bearings above 500 rpm; for slower machines the resonant frequency of the sensor must fall in the 30–45 kHz range to capture the low-energy impact signals from infrequent ball passes.

Common bearing codes encountered in Vietnamese ultrasonic inspection programs:

dBμV Measurement and Baseline Trending

dBμV is a logarithmic unit expressing ultrasonic amplitude relative to 1 microvolt:

dBμV = 20 × log₁₀(V / 1μV)

Absolute values matter less than trending over time. A 6316 C3 bearing (d=80, D=170, B=39 mm, C=72 kN) on a 750 rpm fan might read 28 dBμV when freshly installed — normal for that machine, at that speed, with that grease. The same reading on a 6205 at 2,900 rpm would be abnormal.

Baseline establishment procedure:

1. Measure within 24–72 hours after fresh lubrication (signal stabilizes as grease distributes evenly)
2. Record dBμV at a fixed measurement point (mark with paint or tape for repeatability)
3. Log speed, load, and ambient temperature at the time of measurement
4. Repeat at least 3 times over 1–2 weeks to confirm baseline stability

Alert thresholds from ISO 29821 and field experience:

Trending (the rate of change) matters more than any single reading. A bearing rising 1–2 dB per week behaves completely differently from one jumping 6 dB overnight. Software like SDT Ultranalysis or SKF @ptitude plots trends automatically and predicts failure date based on escalation rate.

Guided Lubrication Protocol

Overgreasing damages bearings as much as undergreasing. Excess grease generates heat through shearing, creates overpressure inside the housing, and pushes seals outward. Ultrasound solves this by giving technicians real-time feedback directly from the bearing as grease is applied.

Listen → Grease → Stop procedure:

Step 1 — Listen: Connect the sensor to the lubrication nipple or nearest measurement point. Record baseline dBμV before pumping. If the signal is +8 dB or higher than baseline, confirm underlubrication before proceeding.

Step 2 — Pump: Add small increments of 0.5–1 gram at a time. After each pump, pause 15–30 seconds and monitor dBμV. The signal typically drops immediately as fresh grease reaches rolling surfaces.

Step 3 — Stop: Cease pumping when dBμV reaches baseline or drops below it. Further pumping adds no benefit and creates unnecessary internal pressure.

This method works best with lithium complex NLGI 2 or NLGI 3 grease for high-speed bearings. For low-speed heavy-load applications like 22220 EK/C3 (d=100, D=180, B=46, C=365 kN) on a mill, calcium sulfonate complex NLGI 2 offers superior load capacity and water resistance.

[NTN Industrial Bearing Technical Reference CAT. No. 3017/E, 2021] recommends relubrication quantity: G = 0.005 × D × B (grams), where D is outer diameter (mm) and B is width (mm). For 22220 EK/C3: G = 0.005 × 180 × 46 = 41.4 grams — pump 5 grams at a time, listening after each increment.

Early Detection vs. Vibration Analysis

Ultrasound and vibration analysis do not compete — they complement each other across different fault stages:

Stage 1 (very early): Ultrasound detects only. Lubricant film thins, high-frequency hissing appears. Overall vibration spectrum remains normal. Action: relubricate.

Stage 2 (early): Both methods show signs. Ultrasound reveals rhythmic impacts; vibration shows slight amplitude rise at BPFO/BPFI frequencies. Action: plan replacement within 30–90 days.

Stage 3 (moderate): Vibration analysis becomes clearer. Sidebands form around fault frequencies. Ultrasonic signal strengthens. Action: replace at next scheduled stop.

Stage 4 (advanced): Full broadband vibration rise. Ultrasonic signal saturates. Infrared thermography shows heat. Action: immediate shutdown.

Ultrasound value lies entirely in Stages 1–2. This is why advanced predictive maintenance programs use ultrasound as the primary filter — only sending bearings to detailed vibration analysis after ultrasound confirms a problem. This approach saves analysts substantial time without sacrificing coverage.

Real-World Case

At a cement manufacturing plant in Hai Duong Province, the maintenance team conducts ultrasonic route inspections every two weeks across 47 electric motors ranging from 75 kW to 315 kW.

During a March inspection run, an Ultraprobe 15000 detected a +14 dB spike compared to baseline on a 185 kW motor driving a kiln cooling fan. The bearing on the drive end was 6316 C3 (d=80, D=170, B=39 mm, C=72 kN), installed with lithium complex NLGI 2 grease and an 800-hour relubrication interval established at commissioning. Simultaneous vibration measurement showed 3.2 mm/s overall — within ISO 10816 Zone B (3.0–7.5 mm/s), below the standard warning threshold.

Based on the ultrasonic signal, the team performed a guided relubrication using the Listen → Grease → Stop protocol. Recommended quantity per formula G = 0.005 × D × B: G = 0.005 × 170 × 39 = 33 grams. They pumped 22 grams total of lithium complex NLGI 2 in 5-gram increments, stopping when dBμV dropped to baseline +3 dB — confirming fresh grease had reached the rolling surfaces. The following week's check showed stable readings.

Two weeks later, the next inspection revealed the signal climbing again — this time +16 dB from original baseline and not responding to additional grease. Ultrasonic FFT spectrum analysis showed rhythmic impacts at 73 Hz, matching the BPFO prediction for this 6316 at 1,450 rpm. BPFO confirmed: (8/2) × (1,450/60) × (1 − 15.87/170 × cos 0°) = 4 × 24.17 × 0.907 = 87.7 Hz — slightly higher than the 6308 example because the 6316 has a larger pitch diameter and shallower ball-to-pitch ratio. The team cross-checked the FFT peak against this calculation, confirming outer raceway fault. Bearing replacement was scheduled for the next planned shutdown window two weeks away.

Disassembly revealed outer raceway spalling approximately 8 mm² in area — Stage 2 fault, entirely amenable to planned replacement. The replacement bearing was ZVL 6316 C3, matched dimensionally and in clearance class to the original. Without ultrasonic monitoring, vibration analysis would have remained silent until Stage 3–4, risking sudden failure on an operating kiln.

According to [Timken Engineering Manual, 2022], detecting Stage 1–2 faults converts bearing replacement from emergency to scheduled maintenance — saving 60–80% of unplanned downtime costs in heavy industrial applications. The plant estimated four hours of kiln downtime avoided, valued at approximately 85 million dong in lost throughput.