Technical Note · Test Setup
Fixture Design for Vibration Testing
A fixture's job is to deliver the controller's motion to the test article unchanged. This note explains what that demands of materials, mass, stiffness and joints, how close to its first resonance a fixture can be used, and how to check it before the test.
Key takeaways
- The ideal fixture is rigid and massless. Real fixtures trade the two: stiffness raises the first resonance, mass uses up shaker force.
- Magnesium, aluminum and steel have almost the same stiffness-to-weight ratio. Magnesium wins because it is light: for the same mass you can make it thicker, and it resonates higher.
- For a lightly damped fixture, the first resonance should be at least 1.85× the highest test frequency to keep mounting points within ±3 dB, and about 3× for ±1 dB.
- Joints are where most fixtures fall short: bolts need enough preload to stay clamped at peak acceleration, and the load path should run straight to the shaker.
- Always verify the fixture on the shaker before the test. Calculations and finite element models predict; a multi-channel sweep confirms. ShakerScope turns that sweep into a report that classifies each resonance and points to its likely cause.
What a fixture has to do
The vibration controller controls the motion at the control accelerometer. The specification, however, defines the motion the test article should receive at its mounting points. The fixture connects the two, and the test is only as valid as the fixture's ability to make them the same. A good fixture:
- Moves as a rigid body across the test band, so every mounting point sees the controlled motion.
- Adds as little mass as possible, leaving shaker force for the test article.
- Restrains cross-axis motion, so the article is tested in the intended axis only.
- Represents the real mounting at the interface: the same attachment points, fasteners and, where specified, mounting stiffness.
- Survives the test repeatedly, without loosening, cracking or wearing its threads.
Materials: why magnesium?
A fixture's first natural frequency depends on how stiff it is relative to how heavy it is. For common fixture metals that ratio, the specific stiffness E/ρ, is almost identical.
| Material | Young's modulus E | Density ρ | Specific stiffness E/ρ | Mass for the same geometry | Notes |
|---|---|---|---|---|---|
| Magnesium (AZ91 / ZM5) | 45 GPa | 1810 kg/m³ | 24.9 MJ/kg | 0.67× aluminum | Lightest; good inherent damping; castable; needs steel inserts for threads; fire precautions when machining |
| Aluminum (6061-T6) | 69 GPa | 2700 kg/m³ | 25.6 MJ/kg | 1.0× | Easy to machine and weld; widely available; low damping |
| Steel | 200 GPa | 7850 kg/m³ | 25.5 MJ/kg | 2.9× aluminum | Very heavy for its stiffness; used for small adapters, inserts and low-frequency or high-load fixtures |
Because E/ρ is nearly the same, a fixture of a given shape resonates at about the same frequency in any of these metals. The difference is mass. And because mass is the real constraint, the useful comparison is frequency for a given mass: a lighter material lets you make thicker, stiffer sections within the same weight budget.
For bending-dominated parts such as plates, cubes and expanders, the advantage grows with the ratio of densities: at equal mass a magnesium plate resonates about 1.5 times higher than aluminum and over 4 times higher than steel. Cast magnesium alloys also have higher internal damping than aluminum, which lowers the peaks of the resonances that do fall in band. That is why Spectral Dynamics head expanders are cast in ZM5 magnesium.
Mass: the price of every kilogram
Shaker force is shared by everything that moves: the armature, the fixture and the test article. The maximum acceleration is F divided by their total mass, so every kilogram of fixture reduces the level available for the test.
Figures 1 and 2 are the two sides of the central design trade: more material raises the first resonance but takes force away from the test. The best fixtures put material only where it carries load and keep the rest out, through ribbed or pocketed castings, box sections instead of solid blocks, and short, direct load paths.
How close to its first resonance can a fixture be used?
A fixture does not stop working suddenly at its first resonance. Its mounting points start to move more than the control point well below it, as the transmissibility rises toward the peak (see our Resonance note). The question is how much error is acceptable.
| Allowed error at mounting points | Fixture first mode ≥ | For a 2000 Hz test | For a 500 Hz test |
|---|---|---|---|
| ±3 dB | 1.85 × highest test frequency | ≥ 3700 Hz | ≥ 925 Hz |
| ±1 dB | 3.0 × highest test frequency | ≥ 6100 Hz | ≥ 1520 Hz |
For tests to 2000 Hz these targets are often impossible for anything larger than a small block, which is why large fixtures and slip tables are usually operated with modes in band. In that case the strategy changes from avoiding fixture resonances to managing them: adding damping, controlling from several accelerometers at once, and limiting responses where justified. These targets also assume the test article itself is light compared with the fixture. A heavy article lowers the fixture's resonances, so the survey should be done with the article or a mass simulator installed.
Geometry: keep the load path short and straight
The shape of a fixture matters as much as its material. Cantilevered features, such as an L-bracket that holds a test article out to the side, turn the fixture into a spring.
| Guideline | Why |
|---|---|
| Use box, cube and ribbed shapes; avoid thin plates and cantilevers | Walls loaded in plane are far stiffer than plates loaded in bending. |
| Keep the test article's center of gravity low and centered over the armature | An offset or tall CG creates overturning moments, rocking modes and cross-axis motion. |
| Put stiffness under the mounting points | Material away from the load path adds mass without adding useful stiffness. |
| Make the fixture symmetric where possible | Symmetry keeps rocking modes from being excited by the axial input. |
| Bolt the fixture to the armature or head expander at many points, close to the load path | The interface is part of the fixture; a few widely spaced bolts make it a spring. |
| Avoid gaps, shims and stacked adapter plates | Each interface adds compliance, possible rattle and damping that varies with level. |
Bolted joints and overturning moments
Most fixture problems that appear during a test, such as resonances that drift, rattling, nonlinear response, or a coherence drop, come from joints rather than from the fixture body.
Bolt preload
A bolted joint stays stiff only while it stays clamped. At peak acceleration, the test article pulls on its fasteners with a force of mass × acceleration, shared between the bolts. If that force exceeds the preload, the joint opens on every cycle.
| Quantity | Value | Note |
|---|---|---|
| Test article mass | 10 kg | |
| Peak acceleration | 30 g | e.g. 3σ peak of a 10 Grms random test |
| Number of bolts | 4 | |
| Separating force per bolt | 736 N | 10 × 9.81 × 30 / 4 |
| Minimum preload with a 2× margin | ≈ 1.5 kN per bolt | Before allowing for moments and prying |
| Overturning moment if the CG is 150 mm above the interface (horizontal axis) | 441 N·m | 10 × 9.81 × 30 × 0.15; adds to the load on the outer bolts |
Ordinary fasteners tightened to their rated torque usually give far more preload than this example requires, so the more common failures are bolts that were never torqued, threads stripped in soft fixture material, and preload lost to vibration over a long test. Practical measures: use steel threaded inserts in magnesium and aluminum, torque every bolt to a documented value, re-check torque between axes or after long durations, and use enough bolts close to the load path rather than a few large ones far apart.
Overturning moments on slip tables and expanders
In horizontal testing the article's inertia acts above the slip plate, producing a moment that tries to tip the plate. The slip table's bearings must restrain it, which is why bearing type matters for tall or heavy articles. Spectral Dynamics hydrostatic journal-bearing (BTJ) and T-bearing (BTT) slip tables are designed for high overturning moments and resistance to bending. Keeping the article's CG low reduces the moment, the rocking it excites, and the cross-axis motion it produces.
Damping
When fixture resonances must fall in band, damping keeps their peaks manageable. Figure 3 shows that damping has little effect well below resonance, but near and at resonance it sets the peak.
- Material choice: cast magnesium has more internal damping than wrought aluminum or steel.
- Constrained-layer damping: a viscoelastic layer bonded between the fixture surface and a stiff constraining plate dissipates energy as the fixture bends.
- Bolted or bonded built-up construction: friction and viscoelastic adhesives at interfaces add damping, but bolted interfaces must stay clamped or they add rattle instead.
- Avoid relying on damping to excuse a weak design. Stiffness moves resonances out of band; damping only lowers their peaks.
Verify the fixture on the shaker
Hand calculations and finite element models are essential for design, but the test uses the fixture as built, bolted to that shaker or slip table with those fasteners and carrying that test article. Only a measurement on the rig shows how it actually behaves. The classic method is a fixture survey: a low-level sweep with accelerometers at the control point and the mounting points, followed by transmissibility, coherence and phase for each point, as described in our Measuring Transmissibility note.
Characterizing the fixture with ShakerScope
ShakerScope, an application for Panther and Lynx controllers, packages that survey into one procedure and goes further: it characterizes the whole rig, from shaker and amplifier through fixture and slip table to the test article, and produces a report a technician can hand to the customer. A first run takes about ten minutes, with the sweep itself lasting two to five.
- Describe the rig. Enter the shaker's rated specifications and the fixture, and ask the built-in AI where to mount each accelerometer for this setup.
- Set the channels and the sweep. Up to ten channels, each with its role, position, axis and sensitivity. Install the test article or a mass simulator, and sweep beyond the test profile so the fixture's roll-off and first modes are measured rather than assumed.
- Press Start. Pre-test checks, the low-level sine sweep and a controlled shutdown run automatically through Panther or Lynx.
- Read the report. Every channel is analyzed after the sweep, and the results are delivered as a five-plot PDF on site, with an AI-written narrative when online.
| Fixture problem | What ShakerScope measures | What to do |
|---|---|---|
| Fixture or head-expander mode in the test band | FRF peak on the mounting-point channels; listed in the resonance inventory with frequency, Q and class (fixture, suspension or armature) | Compare the mode frequency with Table 2; stiffen or damp, re-torque, or plan multi-point control |
| Loose bolt, damaged cable or poorly mounted accelerometer | Coherence below 0.9, shaded on the coherence plot | Re-torque, re-mount and re-sweep |
| Rattling joint or impacting | Rising total harmonic distortion (THD) on the affected channel | Find the loose interface; check preload (Table 4) |
| Rocking or slip-table tilt from an offset CG | Cross-axis response between paired channels; the Orbit Inspector plots the motion between two channels at any frequency | Lower or center the CG; check slip-table bearings and alignment |
| Usable range of the complete setup | Optimal (±12 dB) and usable (±20 dB) bands of the control response, each requiring coherence above 0.9 | Lay the test profile over the bands before testing |
| Changes after a re-torque, a repair or a year of use | Compare Systems overlays a baseline and a new run and reports new, lost and shifted resonances | Save each accepted fixture setup as a baseline |
Two features matter particularly for fixture work. First, a closed control loop can produce a peak-and-notch pair that is not in the structure at all; ShakerScope flags these controller artifacts separately, so nobody stiffens a fixture that doesn't need it. Second, results are compared with the rated specifications, so a fixture that is eating into the shaker's performance is shown as exactly that. Mounting-point transmissibility and the ±3 dB acceptance band of Table 2 still apply; ShakerScope's FRF and resonance results give the data to check them, together with the likely physical cause behind anything that fails.
For day-to-day confidence, its companion ShakerCheck runs a five-minute low-level check before every test, confirming that the channels, fixture and shaker are right on the day.
Common misconceptions
| Misconception | Reality |
|---|---|
| "Steel is stiffer, so a steel fixture is better." | Steel's stiffness-to-weight is the same as aluminum and magnesium. A steel fixture of the same shape resonates at the same frequency and weighs nearly three times as much as aluminum. |
| "Thicker is always better." | Thickness raises resonances but adds mass, reducing available acceleration. Put material in the load path, not everywhere. |
| "The first mode only needs to be above the test band." | Mounting points deviate well before resonance: +3 dB at 0.54·fₙ for a lightly damped fixture. |
| "A bare-fixture survey is enough." | The test article's mass lowers fixture resonances. Survey with the article or a mass simulator installed. |
| "If the control spectrum is in tolerance, the article got the right input." | Only at the control point. Without a survey, mounting points may see very different levels. |
| "Every peak in the sweep is a fixture problem." | Some are suspension or armature modes, some belong to the test article, and a closed loop can create a peak-and-notch pair that is not in the structure. Classify before redesigning. |
Quick reference
| Quantity | Formula or rule | Example |
|---|---|---|
| Available acceleration | a = F / (marmature + mfixture + marticle) | 8.9 kN / 22 kg → 41 g |
| Fixture first mode for ±3 dB | ≥ 1.85 × highest test frequency | 2000 Hz test → ≥ 3700 Hz |
| Fixture first mode for ±1 dB | ≥ 3 × highest test frequency | 500 Hz test → ≥ 1520 Hz |
| Plate frequency scaling | f ∝ thickness × √(E/ρ) / span² | Halving the span quadruples the frequency |
| Equal-mass frequency gain | ∝ 1/ρ for the same footprint | Magnesium ≈ 1.5× aluminum |
| Bolt separating force | m · apeak / nbolts, plus moment effects | 10 kg, 30 g, 4 bolts → 736 N each |
| Overturning moment | m · a · hCG | 10 kg, 30 g, 150 mm → 441 N·m |
How Spectral Dynamics helps
- Magnesium head expanders: cast in ZM5 (AZ91-equivalent) magnesium and designed with finite element and modal analysis for high usable frequency at low mass, with stainless steel inserts, pressed-in steel washers to protect the magnesium, and top and bottom surfaces flat and parallel to better than 0.15 mm. Aluminum versions are available where usable frequency matters more than mass.
- Slip tables for every load: mechanical linear-bearing (MT), oil-film T-guide on granite (GT), and hydrostatic journal (BTJ) and T-bearing (BTT) designs for high overturning moments.
- ShakerScope: one multi-channel sweep on Panther or Lynx characterizes the shaker, fixture and test article together, with an inventory of resonances classified as fixture, suspension or armature, coherence, THD and cross-axis checks, the usable bandwidth, and prioritized setup recommendations. Compare Systems tracks the same fixture over time.
- ShakerCheck: a five-minute, low-level check before every test.
- Fixture surveys with Panther: transmissibility, phase and coherence from the control point to every mounting point, with 24-bit, phase-matched channels.
- Control strategies for real fixtures: Panther's MISO control lets several control accelerometers contribute to one drive, so the controlled level represents the whole fixture interface when resonances fall in band.
References and further reading
- C. M. Harris and A. G. Piersol (eds.), Harris' Shock and Vibration Handbook, 5th ed., McGraw-Hill (chapter on test fixtures).
- D. S. Steinberg, Vibration Analysis for Electronic Equipment, 3rd ed., Wiley.
- A. W. Leissa, Vibration of Plates, NASA SP-160, 1969.
- MIL-STD-810H, Method 514.8, Vibration.
- Spectral Dynamics: Slip Tables & Head Expanders; Vibration Shaker Systems Buyer's Guide.
- Spectral Dynamics Technical Library: Resonance; Measuring Transmissibility.
