PHASEDRAFTVIBRATION ANALYSIS

A bearing fault, before and after.

Two readings of one measurement point on a 1800 rpm machine with a 6206 bearing. Below is what the pipeline computed — not a screenshot, not a summary written afterwards. Every figure is recomputed from those two files on each test run and checked against this page, so a number that drifted would fail the suite rather than quietly stand here.

No sign-in, no invite code, nothing uploaded. The two files are linked at the bottom of this page if you want to run them yourself.

The two spectra

Before Before: velocity spectrum with a bearing outer-race faultVelocity spectrum from 0 to 399.5 hertz. The tallest peak is the bearing outer-race tone at 107 hertz, amplitude 1.4 millimetres per second, with a second harmonic at 214 hertz at 0.6. The shaft line at 30 hertz stands at 0.9.1x 30 HzBPFO 107 Hz2xBPFO 214 Hz0100200300400 Hz
Outer-race tone at 107 Hz, amplitude 1.4 mm/s, with its second harmonic at 214 Hz. The 1× shaft line sits at 30 Hz.
After After: the same spectrum with the outer-race tone reducedThe same velocity spectrum after the work. The outer-race tone at 107 hertz has fallen to 0.05 millimetres per second and the second harmonic at 214 hertz is no longer among the peaks. The shaft line at 30 hertz is unchanged at 0.9. Drawn to the same amplitude scale as the Before chart.1x 30 HzBPFO 107 Hz0100200300400 Hz
The same tone at 0.05 mm/s — the second harmonic is gone. The 1× shaft line is unchanged, which is the point: this was a bearing job, not a balance job.

Both charts are drawn to the same amplitude scale (0–1.5 mm/s) and bin for bin — all 800 lines, no smoothing and no resampling. Auto-scaling the second chart to its own maximum would draw a 0.05 mm/s peak as tall as a 1.4 mm/s one, and the picture would contradict the table.

The verdict

Overall vibration moved from 1.77 to 0.90 mm/s (-49.0%); ISO zone B to A — improved. The Before reading's committed fault is not verified as resolved.

A change is called significant at a ratio of 1.25 (about +1.9 dB) or its reciprocal 0.80; smaller movements are reported with their figures and called no significant change.

Overall level, velocity spectrum, axis y.
BeforeAfterChange
Overall vibration 1.77 mm/s 0.90 mm/s -0.87 mm/s (-49.0%)
ISO 20816-3 zone B A down
Peaks the two readings share, and the one the After reading no longer has. Shaft 30 Hz; compared over 0–399.5 Hz at 0.5 Hz per line.
FrequencyOrder BeforeAfter ChangeVerdict
30.0 Hz 1.00× 0.9 0.9 0.0% no significant change
107.0 Hz 3.57× 1.4 0.05 -96.4% improved
214.0 Hz 7.13× 0.6 gone — not among the After reading's peaks
Band levels. These are root-sum-square bin amplitudes: their absolute value depends on the export's scaling, their ratio does not — so read the percentage.
BandBefore AfterChangeVerdict
Sub-synchronous — 0.0–24.0 Hz 0.006928 0.006928 0.0% no significant change
Running speed — 24.0–36.0 Hz 0.9 0.9 0.0% no significant change
Shaft harmonics — 36.0–300.0 Hz 1.523 0.05502 -96.4% improved
High frequency — 300.0–399.5 Hz 0.01414 0.01414 0.0% no significant change

And here is the part worth reading twice

The outer-race tone fell 96.4% and its second harmonic is gone. The pipeline still did not call this repaired.

Bearing outer race fault (BPFO) at 107.0 Hz — not verified. The After analysis committed this same fault on its own evidence.

The After reading commits the same fault on its own evidence, so the honest answer is that the tone got quieter, not that the fault is gone. A tool that graded itself on how confident it sounded would have written “repaired” here. There is no spelling of “repaired” in this code.

That is what the whole product is for. A committed diagnosis where the evidence supports one, a named uncertainty where it does not, and never the third thing — a confident sentence with nothing behind it.

Run it yourself

These are the two files, unedited. They differ at exactly 2 rows out of 800 (107.0 Hz: 1.4 → 0.05, 214.0 Hz: 0.6 → 0.001); everything else is identical, which is why the charts are readable.

before.csv after.csv Sample report (PDF)

Both files are above. Analysed as two readings of the same machine — 1800 rpm, ISO group 2, rigid support, bearing 6206 — they build the trend a machine page shows. The side-by-side verdict on this page is what the pipeline computes when it is handed both at once.

Running an analysis needs an invite code today — the form will refuse without one. Ask at jk@khaingsinternational.com. Reading this page needs nothing.