Vibration Wont Go Down: Finding the Cause

← Balanset-1A Knowledge Base — all chapters

The key fact from years of support: in roughly 90% of “it won’t balance” cases the culprit is a mechanical fault of the machine or resonance — not the device and not unbalance. The device measures honestly, but balancing cannot cure play, a crack or a resonance.

The diagnostic tree

Work through the steps top to bottom.

Step 1. Are the readings stable at all?

Take 2–3 measurements in a row at the same RPM, changing nothing. Amplitude must repeat within ~10–20%, phase within ~10–20°.

Readings “float” → it is too early to balance. Causes of instability:

1. Bearing play — rock the rotor with a pry bar, belts off.
2. A loosened structure — cracks, untightened bolts, wandering mounts, freely hanging flaps/frames.
3. Резонанција — change the RPM by 50–100: if vibration changes severalfold, you are on or near a resonance.
4. Caked/shedding material on the rotor, wandering free knives.
5. Tachometer trouble — double triggers, glare (see chapter 08): phase jumps because of false marks.

Step 2. Look at the spectrum (FFT) — who is shaking?

Step 3. Resonance signs

• After installing the calculated weight, vibration grew;
• the device “asks for ever more mass” each iteration;
• a small RPM change → a big vibration change;
• phase swings sharply when RPM changes.

Решење:

1. Record a RunDown (coast-down) chart: spin up → cut the drive → the device logs amplitude and phase versus RPM. Peaks = resonance zones; in a resonance zone the phase makes a sharp “bend”.
   

   

   

   Without RunDown, hunt resonances with the vibrometer: starting from low RPM, raise it in 50–100 rpm steps and watch the vibration — a sharp rise at some step reveals the resonance zone. Real example: from 650 to 750 rpm vibration grew fivefold → resonance near 755. Another: 70 mm/s at 1300 rpm but 50 at 1400 → resonance near 1300; balance at 800.

   

2. Pick balancing RPM ван the resonance zones — usually clearly below (600–900 rpm for mulchers). Example: resonances at ~420 and ~644 rpm → balancing at ~844 rpm; another case: no luck at 1600, fine at 1000 rpm.
3. Once balanced at “clean” RPM, climb to working RPM and trim.
4. The radical route: change the structure’s stiffness (reinforce the frame, prop it, change the supports). A machine standing “bare metal on concrete” often behaves badly — vibration mounts/springs help.

Step 4. Nonlinearity — “I mount the weight but the response is off”

Signs: the response to a weight is disproportionate; on removing the weight the vibration does not return to the initial value (something in the machine shifts/hangs).

• Инсталирај a half or a third of the calculated mass per iteration.
• Check whether the trial weight is too big: the change it causes should be 20–30% but no more than ~100%. An overweight trial breaks the linear model and leads to overcorrection.
• If the initial vibration is very low (<1 mm/s), even 100–200 g can cause a sharp spike — reduce the trial mass.
• If vibration doesn’t return when weights are removed — look for a mechanical defect (something is shifting: a fit, a key, a crack).

Step 5. Check your own mistakes

Run through chapter 09. The most frequent:

• the angle counted против rotation (symptom: vibration roughly doubled instead of dropping);
• percent instead of grams (radius not entered);
• the “remove mass” checkbox while welding weights on;
• the tacho mark moved between runs;
• sensors mounted axially or X1/X2 swapped;
• RPM changed mid-series;
• the trial weight not removed when the program assumed it was;
• balancing without the knives/hammers that were fitted afterwards.

Step 6. Defects found in real cases

Real “culprits” from practice (when balancing “wouldn’t work”):

• a destroyed clutch (twin disk) — broken pieces inside, invisible from outside; unstable vibration;
• a bent shaft — 4–20 mm of run-out; balancing compensates only partially, the result is worse than usual (11 mm/s instead of 3);
• play in the motor’s rear bearing — a ~85 Hz peak; waited for a new motor;
• a crack in a pulley;
• broken mulcher brackets/mounts that cracked again after repair;
• a worn bearing seat — cured by build-up welding and re-machining the journal;
• the tachometer counting 2 pulley holes as 2 marks — RPM “doubled”, the maths meaningless;
• aerodynamics — duct suction with insufficient stiffness: nothing to balance; clean the filters and stiffen the structure;
• the variable-frequency drive — a 300 Hz peak (6×50) vanishing when the VFD was switched off.

Step 7. When to stop

Some machines cannot be balanced in the field until they are repaired. Even seasoned support has had a handful of refusals in 10+ years (a combine chopper, a 2.5 m brush cutter, an excavator-mounted mulcher). If the mechanics are defective — tell the customer honestly: repair first (bearings, shaft, frame), then balancing. That is a professional conclusion, not a defeat.

Quick cheat sheet: “why did vibration grow after mounting the weight?”

1. Wrong angle-counting direction → check you counted with the rotation.
2. The “remove mass” checkbox while welding on → inverts the result.
3. Resonance → RunDown, different RPM.
4. Too big a step (full mass on a nonlinear machine) → half the mass.
5. Mechanics (play/crack) → overhaul.
6. False tacho marks → phase is wrong, see chapter 08.