Is It Resonance, or Is It the Force? How to Tell in an Afternoon

Everyone has watched this experiment. During spin-up, a washing machine hits one particular speed and shudders so hard it walks across the floor — then, at a higher speed, it goes quiet and smooth. Nothing was repaired in between. The machine simply passed through a natural frequency: a speed where the shaking force from the spinning drum happened to line up with the frequency at which the machine's own structure wants to vibrate.

That one observation contains the whole diagnosis. There are two frequencies in play. The forcing frequency comes from rotation — once per revolution for unbalance — so it moves up and down with speed. The natural frequency belongs to the structure itself, set by its mass and stiffness, and it does not care what speed the machine runs at. Trouble arrives only where the two meet.

Balancing reduces the force. Resonance multiplies whatever force is left. If a machine runs at or near a natural frequency, even an excellent balance leaves a small residual force that the structure amplifies several times over — so the vibration comes back, the balancing contractor comes back, and the cycle repeats. The problem was never the amount of force; it was the structure's eagerness to respond to it at that particular frequency.

The response curve below is the signature of the situation. Below the natural frequency the structure barely responds; near it, response per unit of force climbs steeply; above it, things calm down again — which is exactly why the washing machine is smooth at full spin. Machines living near the peak also show tell-tale symptoms: vibration that is extremely sensitive to small speed changes, and trial balance weights that never behave quite the way the vectors said they should.

With the machine switched off, strike it — a soft-faced hammer, or a block of timber on a large structure — and record what rings. A structure struck once vibrates at its own natural frequencies and nothing else, exactly like a struck bell. The amplitude dies away, but the frequency of the ring does not change: that frequency is a property of the structure, and it is the number you came for. Hitting harder only makes the ring louder — it never changes the note. What your choice of striker does set is bandwidth: a hard tip means a short contact that spreads energy far up the spectrum, while a soft face or timber block lengthens the contact and concentrates the energy low down — exactly what a big, low-frequency structure needs.

The kit is minimal: one accelerometer on the bearing housing or pedestal, a small DAQ, and an FFT. A two-channel PhonoVibe D is enough — 24-bit capture, IEPE sensor power, USB into a laptop — with the ring analysed in TVIB's TSAP201 spectrum analyzer. If a ring frequency sits at or near your running speed's 1× frequency, you have found your suspect. In a lab setting the same idea is done properly with an instrumented impact hammer and TVIB's FRF tools, but for an afternoon field answer, the simple bump is remarkably effective.

Now use the machine itself as the exciter. Cut power and record vibration continuously as it coasts to rest. The residual unbalance force slides down through every frequency between running speed and zero, and each time it crosses a natural frequency the amplitude swells, peaks, and falls away — a resonance sweep you get for free, using energy already stored in the rotor.

The reading is direct: the speed at which amplitude peaks marks a natural frequency (strictly, the frequency at which the forced response peaks, which sits a shade below the structure's true natural frequency by an amount set by damping — and a machine that coasts down fast can smear the peak slightly later still, so let it take its time through the zone), in the units that matter most — RPM on your own machine. A run-up gives the same curve in reverse. If the peak sits at or close to normal running speed, the case is closed: no amount of re-balancing will hold, because the machine operates on top of its own resonance. If instead the amplitude simply shrinks smoothly with speed and shows no peak, the problem really is the force — and balancing, alignment, or looseness work is the right response.

A representative case, not a specific customer: a plant fan is balanced three times in a year. Each visit the vibration drops, and within weeks it creeps back. Operators also mention it runs noticeably rougher on hot days, when the drive settles at a slightly different speed. An afternoon of testing ends the argument: a bump test on the stopped fan rings at a frequency within a few percent of running speed, and a coast-down shows the amplitude peak almost exactly at operating RPM. The fan is not badly balanced — it is parked on a natural frequency of its support structure.

The fix in a case like this is structural, not rotational: stiffening the pedestal raises the natural frequency safely above running speed, and the next coast-down shows the peak moved well away from the operating point. The balance report finally stays valid, because the amplifier — not the force — was the disease.

Once resonance is confirmed there are exactly four levers. Stiffen the structure and the natural frequency moves up. Add mass and it moves down. Add damping and the peak shrinks and broadens while staying essentially where it is. Or move the forcing instead — change operating speed, or remove the excitation source — so the machine no longer sits on the peak. Whichever lever you pull, verify it the same way you found it: repeat the bump test and the coast-down, and watch the peak move away from the running speed. On critical structures, teams take this further under controlled excitation: a shaker on a vertical or lateral stand, driven with sine sweeps or noise from the TWGM 206 waveform generator, with FRFs measured through PhonoVibe and TVIB — and a machinery fault simulator like TMFSS makes a safe classroom demonstration of the whole resonance story, speed sweep and all.

If you want the theory under this explainer, TIERA's free primers at 101.tieraonline.in cover it — start with Modal & Resonance 101. They are free introductions, not accredited certifications. For formal, career-grade training, TIERA's TCAT programme (see /services) runs structured courses with proctored certification exams at exams.tieraonline.in.

Both field tests in this post run on a two-channel kit. A PhonoVibe D — 24-bit capture, IEPE sensor power, USB into a laptop — records the bump-test ring with the machine off and the full coast-down after power is cut, and the bundled TVIB TSAP201 analyzer puts a cursor on the ring frequency and the amplitude peak. If the peak lands at running speed, you have the answer in data, not in opinion — and the same trace proves the fix worked afterwards.

When the structure is critical enough to deserve controlled excitation, the same chain scales up: TVIB's TFRT and TIST 205 modules measure frequency response functions with a shaker positioned on TIERA's vertical or lateral stands, driven by sine sweeps, bursts, or band-limited noise from the TWGM 206 waveform generator. And if you would rather someone came and settled it, TIERA runs on-site vibration and modal testing — coast-down surveys through root-cause reporting.