# Trial (calibration) weight: why it's needed and how to choose it

> A fan starts humming after the impeller is cleaned, you arrive with the instrument, fit sensors to the bearing housings, and take the baseline vibration reading. The software shows 1x — the vibration at rotation frequency — its phase, and the spectrum (vibration broken down by frequency), but it doesn't tell you how many grams to fit or where. It's missing one run — the one with the trial weight. Below: why the calculation is impossible without it, how to choose the mass, how to tell whether the trial run worked, and where people go wrong most often.

**In short:** A trial (calibration) weight is a temporary weight whose mass, mounting radius, and angular position are precisely known. You fit it to a correction plane, make a second run, and by doing so give the machine a known change in imbalance. From the difference between the before-and-after vibration vectors, the software calculates the influence coefficient — the specific machine's sensitivity to the added weight — and only then calculates the mass and angle of the correction weight. A trial run counts as successful if the 1x amplitude changed by at least 20-30% or the phase shifted by at least 20-30°.

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

## What a trial weight is

A trial weight (also called a calibration weight) is a temporary mass you fit to a correction plane for just one run. It doesn't fix the vibration. Its job is to make the machine show how it responds to an added imbalance.

The word “known” is the key here. The weight only does its job when you know three things about it precisely.

- Mass. Weighed, in grams, not “about twenty.” Scales are included with the Balanset-1A.
- Mounting radius. The distance from the rotation axis to the weight, in millimeters. The instrument uses it to convert vibration into imbalance (g·mm) and to check tolerance.
- Angular position. The point the software will later measure the correction weight's angle from. Mark it with chalk or a marker right away.

> If you plan to balance this same rotor later using saved coefficients, fit the trial weight at the same angular position as the tachometer's reflective mark. That way the influence coefficients stay valid.

## Why the correction can't be calculated without a trial run

After the first run, the instrument knows one thing: the vibration vector at rotation frequency — the 1x amplitude and phase. That's not enough. The software doesn't know the rotor's mass, the stiffness of the bearing housings, how compliant the frame is, the effect of a belt drive, or how much the foundation gives under load. The same 50 g·mm imbalance on a rigid steel base will produce one vibration reading, and on a welded frame with soft vibration mounts, a completely different one.

The trial weight closes that gap. You introduce an imbalance whose magnitude you know exactly (mass times radius) and watch how far and in what direction the vibration vector moves. The ratio of one to the other is the influence coefficient — the specific rotor-supports-foundation system's sensitivity, capturing both the magnitude and the direction of the response.

Formally, it looks like this: α = (V1 − V0) / Ut, where V0 is the 1x vector before the weight, V1 is the vector after, and Ut is the trial weight's imbalance. The software then solves the inverse problem and outputs the correction imbalance Uk = −V0 / α — that is, the mass and angle. The subtraction here is vector subtraction, not arithmetic: 5.4 mm/s after the weight instead of 7.2 mm/s before it doesn't mean things got “25% better.” The phase could have swung by 135°, in which case the change in the vector is larger than either measurement on its own.

A worked arithmetic example, not a site report. Starting point V0 = 7.2 mm/s ∠ 75°. Fit 30 g at a 350 mm radius, get V1 = 5.4 mm/s ∠ 210°. The vector difference ΔV ≈ 11.7 mm/s ∠ 236°, influence coefficient α ≈ 0.39 mm/s per gram, correction mass around 18.5 g at the same radius. The balancing software does this calculation itself and draws a polar diagram, but it's worth understanding the logic — you'll immediately notice when a result looks suspicious.

> The three-run method for two planes, and the two-run method for one, only works on a linear system. If the speed, temperature, belt tension, or sensor position changes between runs, the influence coefficient is now describing a different machine.

## The 30/30 rule: the criterion for a valid trial run

The trial weight has done its job if the machine responded noticeably. The practical threshold is this: the 1x amplitude changed by at least 20-30% (in either direction, up or down), or the phase shifted by at least 20-30°. Balanset's documentation calls this the 30/30 rule.

The logic is simple. Any vibration measurement has spread: the speed drifts, the flow in a fan varies, the bearing temperature rises. If the weight changes the reading by 5%, you can't tell its effect apart from that noise, and the influence coefficient comes out as essentially a random number. The software will dutifully calculate a correction, you'll fit 40 g instead of the 12 you actually needed, and the vibration will go up.

While you're at it, check the stability of the readings themselves. The 1x amplitude and phase shouldn't change by more than 10-15% over the course of a reading. If they drift more than that, the machine is running close to resonance, and balancing it in that state is pointless.

| What the trial run showed | What it means | What to do |
| --- | --- | --- |
| 1x amplitude changed by 30% or more, or phase by 30° or more | The system responded, the influence coefficient is reliable | Calculate the correction, fit the weight |
| Change of 5-15% | The weight is too small, the response is lost in the measurement spread | Increase the mass 2-3 times, weigh it, enter the actual value, repeat the run |
| Readings didn't change at all | The weight came off, it's not where you think it is, the speed is different, or imbalance isn't the main cause of the vibration at all | Check the weight's mounting, the tachometer mark, the speed; compare 1x with the overall vibration |
| Amplitude and phase drift more than 15% from reading to reading | Speed is unstable, or the machine is close to resonance | Change the speed or the foundation-mounting conditions; don't balance in resonance |
| Amplitude increased several times over | The weight landed close to the heavy spot and added to the existing imbalance | This is a normal working situation, the calculation is still correct. If the vibration is unsafe, reduce the weight or move it by roughly 180° |

> The Balanset-1A tells you itself whether the trial weight was adequate. That prompt doesn't replace your own read of the numbers: the instrument evaluates the change in the vector, not whether the weight is actually held on securely.

## How to choose the mass and radius

1. **Assess where you're starting from** — Look at the baseline 1x and the speed. The heavier the rotor and the lower the speed, the larger the mass you'll need. The higher the speed, the more careful you need to be: centrifugal force grows with the square of the rotation frequency, and a weight that's harmless at 500 rpm creates 36 times the load at 3000 rpm.
2. **Start small** — Take a mass that clearly won't drive the vibration up to a dangerous level. One wasted run is cheaper than a weight flying off. You can always build the mass up later.
3. **Weigh it and enter the actual figures** — Enter the weighed mass in grams and the measured radius in millimeters into the software. Percent mode is convenient for a quick pass, but it costs you the residual-imbalance calculation in g·mm and the ability to save the influence coefficients.
4. **Fasten it at a strong, marked point** — Choose a spot where the weight holds securely and where you can actually fit the correction mass afterward. Mark the point: the software will measure the angle from it.
5. **Make the run at the same speed** — The trial run's speed has to match the baseline run's speed. Wait for a steady operating condition before starting the measurement.
6. **Check the change against the 30/30 rule** — Change is enough — move on to calculating the correction. Not enough — remove the weight, fit a heavier one, enter the new mass, and repeat. Don't try to squeeze an influence coefficient out of readings that barely moved.

> Fit the correction mass at the same radius as the trial weight. If that's not possible, recalculate it inversely proportional to the radius, or use the software's function for recalculating weights onto other planes and radii. The correction weight's angle is measured from the trial-weight location, in the direction of the rotor's rotation.

## Mounting and safety

A trial weight is only temporary in purpose. In terms of mounting requirements, it's no different from a permanent one: it spins at the same speed and the same radius. A 50 g plate that comes loose on a fan impeller will punch straight through the guard.

- [x] Fasten the trial weight as securely as a permanent one: a bolt, a clamp, a tack weld; a magnet only at low speed and on a clean surface.
- [x] Check clearances: the weight must not touch the guard, the diffuser, guide vanes, or any other stationary part.
- [x] Don't fit the weight to thin sheet metal, a blade of doubtful strength, or a location with a crack.
- [x] Account for the load on the bearing housings: the trial weight adds a dynamic force, and on worn bearings this is noticeable.
- [x] Stay clear of the plane of rotation during the run, and don't leave anyone standing near the guard.
- [x] Stop the machine and wait for it to come to a complete standstill before removing or repositioning the weight. Never reposition it during coast-down.
- [x] Decide in advance at what vibration level you'll abort the run, and don't chase a nice-looking number if the machine is clearly overloaded.

> A weight that's too heavy is more dangerous than an extra run. If you're unsure about the mass, fit less and do one more run instead.

## Common mistakes

### The mass was entered “by eye”

You wrote down 25 g, but 31 g is actually hanging on the rotor. The influence coefficient is off by 20%, and the correction mass comes out wrong. The instrument won't spot the discrepancy, and you'll end up with extra iterations.

### The radius was measured roughly

This barely affects the correction mass, as long as the correction weight goes at the same radius. But the residual imbalance in g·mm, and the tolerance check against grade G (the standard's residual-imbalance limits), become fiction.

### The trial weight was left on without saying so

It couldn't be removed, and that wasn't noted in the software. The correction is calculated as if the weight weren't there, and you end up with an extra imbalance sitting at a known location.

### The sensor was moved between runs

A different point on the bearing housing, or a different measurement direction, is a different system. The phase shifts, and the influence coefficient becomes garbage. Mount the sensors once and leave them alone until the job is done.

### The speed changed

Runs at 1450 and 1480 rpm are already incomparable, and if you moved closer to resonance the difference will be dramatic. Work at one steady operating point.

### The angle was measured in the wrong direction

The weight ends up mirrored, and vibration goes up instead of down. The simpler safeguard is fixed-position mode: the software gives you a blade or hole number, and no protractor is needed.

### The weight was fitted somewhere awkward

The trial weight ends up on the inside of the wheel, somewhere you can't later reach with the correction mass. Choose a point that works for both operations.

## When a trial weight doesn't help

The influence-coefficient method assumes the system is linear and that imbalance is what's causing the vibration. Where that assumption doesn't hold, the trial weight will show you honestly: there's a response, but balancing isn't improving the picture. Building up the mass any further from there is pointless.

You get the first sign even before the trial run. Compare the overall vibration (the total level across all frequencies) with the 1x running-speed component. If the overall level is several times higher than 1x, something else is producing most of the vibration, and balancing won't remove it.

- Resonance. Amplitude jumps around in a narrow speed range, the 1x phase swings by almost 180°, readings drift from measurement to measurement. Change the speed or the foundation-mounting conditions, then balance.
- Loose fasteners and “soft foot.” The system is nonlinear, the influence coefficient changes from run to run. Torque the fasteners and check the supports first.
- Shaft misalignment. 2x is strong in the spectrum, axial vibration is often high. It's fixed with shaft alignment, not weights.
- Bearing defects. Peaks at high frequencies not related to running speed, a raised noise floor. Balancing won't help here.
- Cracks, rubbing, build-up, and uneven wear. Imbalance comes and goes on its own, the result is unstable. Look for a mechanical cause.
- The rotor is flexible at operating speed. A single pair of correction planes doesn't describe its behavior; different methods and a different part of the standard are needed.

> A trial run is useful even in these cases. It shows the system's response, and that response often makes it clear that the vibration isn't coming from imbalance. That saves you half a day of trying to chase a number down with weights.

Sources: [ISO 20816-1:2016](https://www.iso.org/standard/63180.html) · [ISO 13373-3:2015](https://www.iso.org/standard/40840.html) · [ISO 13373-5:2020](https://www.iso.org/standard/62202.html) · [ISO 21940-12:2016](https://www.iso.org/standard/50429.html)

## Who does this, and with what

Everything described here is done by one person with a two-channel instrument. The Balanset-1A gives you two accelerometers for the bearing housings, a laser phase sensor with a tachometer, a USB module, and Windows software. The instrument measures speed, the amplitude and phase of vibration velocity, overall vibration and 1x, and builds the FFT spectrum. The software tells you whether the trial weight was adequate, calculates and stores influence coefficients, splits the weight across fixed positions, proposes drilling, checks tolerance against grade G, and stores the results in an archive for the report. Balancing is done in one or two planes, directly in the machine's own bearings.

Treat the grade-G numbers and the vibration zones in mm/s as a working guide. Before contractual acceptance, check against the applicable part and current edition of the standard, and record the measurement points, the frequency band, the operating mode, and the support type.

If there's no time to work through all this on your own fan or crusher, AXILINE's engineers come to the site and balance it themselves. We design and manufacture the Balanset instruments and use them ourselves on site visits, so the question “what trial weight should I use on this wheel” is routine for us. We start with vibration diagnostics: we check 1x against the overall level and the spectrum, so we're not balancing something balancing can't fix. The instrument can also be bought and learned independently — consulting support is there for that case.

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

## Frequently asked questions

**What mass should the trial weight be?**

There's no universal figure: the response depends on the rotor's mass, the stiffness of the supports, and the speed. Here's how it works: take a small mass you have on hand, make a run, look at the change in 1x and phase. Too small — increase it 2-3 times, reweigh it, enter the new value, and repeat the run. The Balanset software tells you whether the trial weight was adequate, so you're not guessing by eye.

**Does the trial weight have to be removed before fitting the correction weight?**

Usually yes — by default, the trial weight is removed after the trial run. If it can't be removed (welded on, sitting somewhere awkward), tick the “leave trial weight in plane 1” box in the software — then its mass gets factored into the calculation. You can't just leave the weight there without telling the software: the correction will miss.

**Why enter the actual mass in grams instead of working in “percent” mode?**

In percent mode, the software gives you the correction mass as a fraction of the trial weight, and you'd have to measure it off a weight you never actually weighed. On top of that, without grams and a radius you can't calculate the residual imbalance in g·mm or check tolerance against grade G, and you won't be able to save the influence coefficients for next time either.

**Can you skip the trial run entirely?**

Yes, if you've already balanced this same rotor with this instrument and saved the influence coefficients along with the trial weight's mass. Then one run is enough. The condition is strict: the same machine, or an identical type, the same sensor points and directions, the same speed. For a new rotor, the trial run is needed.

**You fitted the trial weight, but the readings barely changed. What now?**

First check the obvious: the weight is in place, the speed is the same, the sensors haven't moved, the reflective mark is being read cleanly. Then increase the mass 2-3 times and repeat the run. If there's still no response, look at 1x against the overall vibration and at the spectrum: most likely something other than imbalance is producing most of the vibration, and there's nothing here to balance.

**Which is more dangerous: a trial weight that's too small, or one that's too large?**

A small one ruins the calculation, a large one breaks the machine. A small weight produces a vibration change on the order of the measurement spread, and the influence coefficient comes out unreliable — but nobody gets hurt. A heavy weight can drive vibration up to unacceptable levels, overload the bearing housings, fly off at speed, and wreck the mounting. That's why the mass is built up in steps, not thrown in “with margin” from the start.