# Loose Mounting and Soft Foot: the Cheapest Find on the Site

> You've shown up with an instrument and a plan to balance an induced-draft fan. Pick up a wrench, a feeler gauge, and a dial indicator first. Mechanical looseness and soft foot show up more often than imbalance, cost less to fix than any other repair, and meanwhile counterfeit other machines' symptoms: they raise 1x (vibration at the rotor's running speed), scatter harmonics across the spectrum, and make the phase unstable. Let's go through how to tell one from the other and how to find both in a single walk-down.

**In short:** Mechanical looseness is a loss of stiffness at a joint: a bolt isn't holding, a fit has worked loose, a shim has crushed, a weld has cracked. Soft foot works differently: the bolts are tight, but the foot's base and the mounting pad don't lie in the same plane, and tightening deforms the machine casing. You find looseness with a wrench, a feeler gauge, and tapping; you find soft foot with a dial indicator while loosening the bolts one at a time, with a common working criterion of 0.05 mm. Until the joints are restored, balancing is pointless: the influence coefficients — the machine's response to a trial weight — drift from run to run, and the result won't hold.

Source: https://axiline.pt/en/articles/loose-mounting-soft-foot/  
Publisher: AXILINE · Vila Nova de Gaia, Portugal · +351 931 831 229 · axilinegeral@gmail.com

## Why Looseness Gets Checked First

Weigh up the cost of each fix. Balancing requires an instrument, three runs, and weights. Shaft alignment requires time to strip the coupling and work with shims. Bearing replacement requires a spare part and downtime. A loose foundation bolt requires a wrench and five minutes.

Meanwhile, looseness counterfeits nearly any other diagnosis. It raises the running-speed component, adds 2x (vibration at twice running speed), scatters the spectrum into whole-number harmonics, lifts the noise floor, and makes the phase unstable. A technician sees a high 1x in the spectrum — the breakdown of vibration by frequency — and orders a balancing job. A weight gets installed, vibration drops for a week, then comes back.

Looseness itself is rarely the root cause. It works as an amplifier. A joint that's lost its stiffness lets through more motion for the same exciting force. An imbalance that would produce 2 mm/s on a rigid base produces eight on a loose frame. You end up chasing the rotor while the frame is the culprit.

There's also a feedback loop: vibration itself destroys the mounting. The bolt loses preload, the fit works loose, the grout delaminates, and a month later the level is higher than it was. The process runs in a circle and accelerates.

> The order of work on site is always the same: mounting and stiffness first, then shaft alignment, then balancing. Doing it in reverse order almost guarantees a repeat visit.

## Six Types of Looseness You'll Find in the Field

### Machine-to-Base Mounting

Loose foundation bolts, an anchor that's lost preload, a nut with no lock-nut, a washer that's crushed into the foot. The most common and cheapest case. Checked with a torque wrench in a single walk-down.

### Fit Looseness

An impeller or fan wheel on the shaft, a coupling half, a pulley on a taper bushing, a bearing in its housing or on the shaft. Here the clearance rotates together with the rotor, so the picture shows up as 1x with a drifting phase, and it grows as the machine warms up.

### Cracks in the Frame, Bedplate, and Welds

A crack in the weld of a support post or in a corner of the frame produces directional compliance. The amplitude in one direction is several times higher than in the perpendicular one, and the spectrum scatters into harmonics. Found by inspection after degreasing and by tapping.

### Foundation and Grout Settling

The concrete under the foot has crumbled, the epoxy grout has delaminated at the edges, a gap has opened up under the frame's base. Tightening the bolts doesn't help here: there's nothing left to pull against. Fixed by repairing the base.

### Crushed and Excess Shims

A stack of seven thin shims, painted or rusted plates, a shim smaller than the foot's base. Tightening compresses this stack, and then the machine settles and the joint loosens again.

### Fatigued Vibration Isolators

The rubber has hardened or cracked, the springs have settled unevenly, the isolator's mounting has loosened. Uneven settling at the corners tilts the machine and creates soft foot on its own.

## Soft Foot: Not Looseness, but Its Opposite

Soft foot means a mismatch in geometry: the foot's base and the mounting pad don't lie in the same plane. You tighten the bolts to normal torque, the foot pulls down flush, but the machine casing bends in the process. On paper everything is tight. In reality the machine is standing on three feet out of four, and you've forced the fourth one down.

The consequences reach inside the machine. The bearing housing bores go out of alignment with each other, clearances change, an electric motor's air gap tilts and a line appears at twice mains frequency (100 Hz on a 50 Hz supply). On a pump, soft foot drags the shaft alignment out of true: you carefully aligned the unit, and after tightening the misalignment came back. For balancing, what matters more is that the support's dynamic stiffness changes, and the influence coefficient changes along with it.

Soft foot comes in four types, and each is fixed differently.

- Parallel. The base is parallel to the pad, but there's a gap across the entire surface between them. Missing shims. The simplest case: add a shim of the right thickness and the issue is closed.
- Angular. The base sits at a wedge, gap on one side. Caused by an uneven pad, a deformed foot, or a twisted frame. Flat shims are pointless here: you'll get a point contact and deformation all over again. Needs a stepped shim set or machining of the pad.
- Squishy. Under the foot: a stack of thin shims, dirt, paint, rust, burrs. Tightening compresses the stack, the machine runs, the stack settles, the preload is lost. Fixed by cleaning up and replacing the stack with a single solid shim.
- Induced (strained). The geometry is fine, but an external force is pulling on the machine: a rigid pipe, a duct, a pinched cable gland, an over-tightened coupling. The indicator shows movement when the bolt is loosened, even though the shims have nothing to do with it. Checked by disconnecting the flange.

> A separate case is bolt-bound: the hole in the foot doesn't line up with the anchor's axis, the bolt jams against the side of the hole, and it won't let you shift the machine during alignment. This isn't soft foot, but it gets fixed in the same pass, either by opening up the hole or by switching to a smaller-diameter bolt.

## How to Check for Soft Foot with a Dial Indicator

1. **Prepare the Pads** — Tighten all the bolts to the specified torque. Clean the feet's bases and the base pads of paint, rust, oil, and remnants of old gaskets. Remove burrs. As long as there's a layer of paint under the foot, every reading is a lie, and you'll be chasing a millimetre that isn't really there.
2. **Go Over It with a Feeler Gauge** — Before reaching for the indicator, go around the machine with a 0.05 mm feeler gauge along the full perimeter of each foot with the bolts tightened. If the gauge slides in, you've already found a gap, and how far it goes in tells you its shape: across the whole base, or wedge-shaped on one side. This takes two minutes and gets you half the diagnosis.
3. **Set Up the Dial Indicator** — Put the magnetic base on the foundation or the frame, not on the machine itself. Rest the indicator's plunger on the top face of the foot, as close to the bolt as possible, strictly vertical. Zero the dial.
4. **Loosen One Bolt at a Time** — Loosen one bolt fully, leave the others tight. Watch the indicator. If the foot lifts, you've found soft foot and you know its value in millimetres. Tighten the bolt back down, zero the dial, move to the next foot. Don't loosen two bolts at once — the result becomes meaningless.
5. **Compare Against the Criterion** — Working guideline: a lift of up to 0.05 mm is considered acceptable for most industrial machines. For high-speed units and machines with a small air gap, use 0.02–0.03 mm. The exact value still comes from the machine's documentation, and the general rule only serves as a starting point.
6. **Fit Shims and Repeat the Round** — Add shims matched to the measured value. For an angular foot, use a stepped shim set or machine the pad. Tighten and recheck all the feet around the circle: fixing one changes the picture at the others. Repeat the cycle until every foot is within tolerance.

> Keep the stack no more than four shims thick, and use stainless plates with a U-shaped cutout so you can swap them without removing the bolt entirely. One 1.0 mm plate is always better than ten 0.1 mm ones.

## What Shows Up in the Measurements

Looseness rarely draws one clean line. It messes up the entire spectrum, and that mess is itself the sign. The cause is nonlinearity: a joint with a gap transmits force in one direction and not in the other, the sine wave gets clipped, and a clipped sine wave decomposes into a set of harmonics.

- [x] A comb of whole-number harmonics. 1x, 2x, 3x, 4x, and up, sometimes as far as 8x–10x. The classic signature of lost stiffness.
- [x] Fractional harmonics at 0.5x, 0.33x, 1.5x. Appear when the gap is large enough to produce a real impact. This is no longer "a bit loose" — it's knocking.
- [x] A raised noise floor. The spectrum stops being a set of lines on a clean background; the whole bottom lifts up. Often it's the floor that catches your eye first when comparing against a previous measurement.
- [x] Unstable 1x phase. Phase is the angle that ties vibration to the rotor's revolution. Take it over three runs in a row. A spread of ±5° is normal. Jumps of 20–40° mean the system's stiffness is changing while it runs.
- [x] Sharp directionality. 9 mm/s horizontal, 1.5 mm/s vertical with the same sensor mounting: the compliance is directional — look for a loose foot or a crack in that direction. Imbalance gives a more even picture across the radial directions.
- [x] An amplitude drop across a joint. Measure on the bearing housing, on the foot, on the frame, and on the concrete nearby. A drop of more than one and a half to two times at a single joint points directly to where stiffness is being lost.
- [x] Impacts in the time waveform. Don't limit yourself to the spectrum. Looseness produces an asymmetric signal with clipped peaks and discrete impulses that the spectrum smears into the floor.
- [x] A change in the picture after tightening. The most convincing test of all: measure, tighten one joint, measure again at the same point with the same sensor. A change of more than 20–30% settles the question without argument.

> The unpleasant part: soft foot can raise a clean 1x with no harmonics at all, because there are no impacts involved. A spectrum without a harmonic comb doesn't excuse skipping the walk-down with a wrench and a dial indicator.

## Table: What You See, Where to Look, How to Check

| What You See in the Measurements | Where to Look for Looseness | How to Check by Hand |
| --- | --- | --- |
| Comb of 1x, 2x, 3x… and a raised noise floor | Machine-to-frame mounting, frame-to-foundation mounting, a cracked weld | Torque check on all bolts, 0.05 mm feeler gauge around the feet, tapping the frame and welds, flashlight inspection after degreasing |
| Fractional harmonics at 0.5x and 0.33x, impacts in the time waveform | A real gap: a worn bearing fit in the housing, a loose bolt, delaminated grout | Dial indicator on the outer ring while the shaft is jacked up, fretting marks (reddish powder) on the fit, tapping the grout with a hammer |
| High 1x, phase drifting by tens of degrees | A loosened fit of the impeller, fan wheel, coupling half, or pulley on the shaft | Dial indicator on the hub while rocking it with a lever, marks of spinning and peening on the key, a repeat measurement on the warmed-up machine |
| Horizontal four times higher than vertical | Directional compliance: one foot, one support post, a crack in the frame | Dial indicator on each foot in turn, a bump test on the frame, comparing amplitude at the foot and on the concrete nearby |
| Level changes noticeably after retightening a bolt | Exactly the joint you retightened | Measure before and after at the same point. Mark the nuts with paint so a month later you can see which ones worked loose again |
| 2x has risen, axial vibration has risen | Misalignment caused by soft foot or base settling | Soft foot with a dial indicator first, only then shaft alignment. Otherwise you'll align the machine while it's in a distorted state |
| A line at twice mains frequency (100 Hz on a 50 Hz supply) | Motor casing distortion from soft foot, loose stator mounting | Check the feet with a dial indicator, run a power-off test: the electromagnetic line disappears instantly, a mechanical one decays with the coast-down |
| Vibration rises over hours of running and drops after shutdown | Thermal distortion of the frame, a pinched pipe, a strained foot | Measure on the cold and the warmed-up machine, disconnect the pipe flange with a dial indicator on the foot |
| Level jumps from run to run for no obvious reason | Fatigued vibration isolators, uneven settling, a rigid contact bypassing the isolator | Measure the free height of each isolator and compare them, find and remove any metal bridges |

> The table points you toward where to look, not to a conclusion. Looseness, misalignment, and bearing defects can look alike, and on one machine they often coexist. Assess the overall level separately, in mm/s RMS (vibration velocity root-mean-square) over the 10–1000 Hz band, per the applicable part of ISO 20816. A subtlety: a loosened foundation effectively turns a rigid base into a flexible one, but that's not a reason to apply the more lenient flexible-foundation limits to the machine.

Sources: [ISO 20816-1:2016](https://www.iso.org/standard/63180.html) · [ISO 13373-3:2015](https://www.iso.org/standard/40840.html)

## Shims, Anchor Bolts, Grout, and Vibration Isolators

### Shims

Use factory-made stainless shims with a U-shaped cutout. Keep the stack no more than four pieces thick: the more layers, the more springiness, and the faster the machine will settle. The shim should cover the entire base of the foot, not just the spot under the bolt. Throw out rusty and painted plates — they're exactly what causes a squishy foot.

### Anchor Bolts and Fasteners

A bolt holds the machine down through tensile force, not by its head bearing down. A short stud in worn threads loses preload with any temperature cycle. Torque to spec, mark the nuts with paint, and a month later check which marks have shifted. An anchor that spins freely or pulls out together with a cone of concrete needs to be reset, not tightened harder.

### Epoxy Grout

Grout transfers the load from the frame into the concrete, and it fails predictably: it delaminates at the edges, cracks around the perimeter, and leaves a gap under the frame's base. Tap it with a hammer — a dull, hollow thud instead of a solid ring means a void. Slide a 0.05 mm feeler gauge under the edge of the frame. Repairing the grout on a machine like this removes more vibration than any balancing job would.

### Vibration Isolators

Rubber ages and hardens, springs settle unevenly, and uneven settling at the corners creates soft foot on its own. Measure the free height of each isolator and compare them. An isolator only works when the mounting's natural frequency is well below running speed: hardened rubber raises the stiffness, the frequency creeps up toward running speed, and you end up with a resonance out of nowhere.

> Separately, check that the machine isn't being short-circuited anywhere by a rigid contact bypassing the isolators: a pinched cable, a guard pressing against something, a drain cast into concrete, a rigidly welded pipe. One such bridge nullifies the entire vibration-isolated mounting, and no amount of shim-fitting will fix it.

## Why Balancing a Machine with Looseness Won't Hold

Balancing rests on two assumptions: the system is linear and constant over time. You install a trial weight of known mass at a known radius, the instrument watches how the 1x vector — amplitude and phase — changed, and derives an influence coefficient. It then solves the inverse problem and gives you the correction mass and angle.

Looseness breaks both assumptions at once. The joint's stiffness depends on amplitude: at small motion the foot sits on the pad, at large motion it lifts off. An influence coefficient taken at one amplitude doesn't work at another. On top of that, it changes from run to run along with tightening and temperature.

In practice you'll recognize it from the symptoms. The trial run gives a strange response, the calculated correction doesn't match reality. After installing the weight, vibration drops by half of what was promised, and a trim weight pulls the vector off in a third direction. Four fine-tuning attempts instead of one almost always means the mechanical checks were skipped.

Worse still, balancing here can actually make things worse. A weight fitted to a machine distorted by soft foot turns into extra imbalance the moment you add a shim and the casing springs back into shape. And if the impeller's fit has loosened, the weight doesn't even stay at a fixed angular position: the hub has spun a few degrees, and the correction is no longer where it was calculated to be.

And there's a safety question. Running up a machine with a cracked frame or a spinning anchor and adding trial-weight mass to it is not something you want to do. Mounting first, weights second.

> Balancing with known looseness present is honest only as a temporary measure until a scheduled repair, agreed on in advance and recorded in the report. In that case you agree upfront that the result will last only until the base is repaired, not for six months. More on why the level comes back is in the article on vibration returning after balancing.

## A Single Walk-Down and Where We Can Help

The sequence is simple and almost always the same. First you restore the joints and the geometry, then you check the shaft alignment, and only then do you balance. That way balancing fits into a single visit, the influence coefficients come out stable, and the result holds.

AXILINE's engineers design and manufacture Balanset instruments and use them to do field balancing themselves. When we arrive on site, we don't start with weights. First, two accelerometers on the bearing housings, overall vibration against 1x, spectrum, time waveform, and phase across three runs. Then the wrench, the feeler gauge, and the dial indicator. If the vibration is coming from looseness or soft foot, you'll hear that before you pay for a balancing job that wouldn't have held anyway.

You can get the instrument yourself and do the walk-down on your own. Balanset-1A is a portable two-channel vibration analyzer and balancer: two accelerometers, a laser phase sensor using a reflective tape mark, a two-channel USB module with preamplifiers, integrators, and an ADC, and Windows software. It shows overall vibration and 1x side by side, phase, speed, the FFT spectrum, and the time waveform, and its archive lets you put before- and after-tightening measurements into a single report. The next step, once the mounting is sound, is single- and two-plane balancing, fixed positions, drilled-hole calculation, saved influence coefficients, trim balancing (fine correction with a small weight), and tolerance calculation by G accuracy class per ISO 21940-11. There's a Balanset-1A OEM version without the case for building into machine tools and test stands. Consulting support is included.

Let's be direct about the limits. The instrument computes velocity RMS (the same as the mm/s RMS figure) over the 5–200 Hz band: looseness harmonics and fractional harmonics show up well there, and that's enough to settle the "mounting or rotor" question. Early bearing defects sit higher, in the low-kilohertz range, and need an analyzer with a high-frequency path. And no instrument at all will find soft foot. That's found by a dial indicator in the hands of someone who didn't mind loosening a bolt.

- [x] Tools: a torque wrench, a feeler gauge set starting at 0.03 mm, a dial indicator with a magnetic base, a hammer, a flashlight, marking paint, a set of stainless shims.
- [x] Before-measurement: overall vibration (the total level across all frequencies) and 1x at every bearing housing, horizontal, vertical, and axial, plus phase. This is your reference point — without it there's nothing to compare against.
- [x] Wrench walk-down: bolts from the machine to the frame, the frame to the foundation, bearing housing covers, the guard, the coupling half. Mark with paint as you tighten.
- [x] Feeler gauge around the perimeter of each foot with the bolts tight, on all four sides. Record where it goes in and how far.
- [x] Dial indicator: loosen the bolts one at a time, record the lift value at each foot, with a 0.05 mm criterion or tighter per the documentation.
- [x] Tapping: the frame, welds, grout, support posts. Listen for a difference in sound — a dull spot means a void or a crack.
- [x] Inspection: cracks along welds and in the corners of the frame, fretting and spin marks on the fits, the condition of the vibration isolators, pinched pipes and cables.
- [x] After-measurement: the same point, the same sensor, the same direction, the same operating condition. Compare against the before-measurement and decide whether balancing is even needed.
- [x] Report: what was retightened, where a gap was found, how much shim was added, how vibration changed. Otherwise, a month later no one will remember what's already been checked.

Sources: [ISO 20816-1:2016](https://www.iso.org/standard/63180.html) · [ISO 21940-11:2016](https://www.iso.org/standard/54074.html) · [ISO 13373-3:2015](https://www.iso.org/standard/40840.html) · [Balanset-1A operation manual](https://vibromera.eu/balanset-1a-operation-manual/) · [Balanset-1A manufacturer specification](https://vibromera.eu/product/balanset-1/)

## Frequently asked questions

**How is soft foot different from loose mounting?**

Looseness is the absence of a joint: a bolt is free, the foot lifts, the gap acts like a hinge. Soft foot is the opposite in meaning: there is a joint, and it's under strain. The bolts are tight, but the geometry doesn't match, so the machine sits twisted, like a table on an uneven floor. Residual stresses live inside the casing, the bearing housing bores go out of alignment with each other, the support's dynamic stiffness changes. You hear looseness with a hammer; you see soft foot only with a dial indicator.

**What tolerance for soft foot counts as normal?**

The industry working guideline: a foot lift of up to 0.05 mm when the bolt is loosened is considered acceptable for most industrial machines. For high-speed units, precision drives, and machines with a small air gap, use a tighter figure, 0.02–0.03 mm. This is a guideline, not a standard for your specific machine. The exact value comes from the manufacturer's documentation, and for contractual acceptance, record that value — not the general rule.

**Can soft foot be checked without a dial indicator, using only a feeler gauge?**

A feeler gauge finds a gap but doesn't show deformation. It works well for parallel and angular soft foot: run a 0.05 mm gauge around the base's perimeter with the bolts tight, and if it goes in, there's a gap, and how deep it goes tells you the shape. A squishy foot from a stack of shims and a strained foot from a pinched pipe won't be caught by the gauge: there the base sits flush, but the casing is still being pulled. Start with the feeler gauge, finish with the dial indicator.

**We retightened all the bolts and vibration didn't change. Does that mean there's no looseness?**

It doesn't. A retightening check only rules out one type of looseness — lost preload in a bolted joint. A worn bearing fit in its housing, a spun impeller hub, a cracked weld, delaminated epoxy grout, and settled vibration isolators aren't fixed by tightening and don't respond to it. Go through the rest of the walk-down: dial indicator on the feet, tapping the frame and grout, inspecting the fits for fretting and spin marks.

**Looseness has raised 1x. How do you tell it apart from real imbalance?**

Look at three things. First, 1x phase repeatability: imbalance holds phase within ±5° from run to run, looseness swings it by tens of degrees. Second, directionality: imbalance produces comparable amplitudes across the radial directions, looseness sharply favours one direction. Third, the response to tightening: a change of more than 20–30% after retightening one joint means you've found the source, and it isn't the rotor.

**Can a machine be balanced if looseness has been found but the foundation repair is postponed?**

Sometimes yes, but only as a deliberate temporary measure. Agree in advance that the result will only last until the repair, and record that in the report along with the defects found. Technically, expect the worst: the trial run will give an unstable response, the correction won't converge on the first attempt, and once stiffness is restored the weights will have to be recalculated. Work on a machine with a cracked frame or a spinning anchor doesn't start at all until the mounting has been restored.