We don't just say it works. We show you.
Spectral Dynamics builds a simulated shaker, fixture and test article around the Panther controller, with the channels, geometry and dynamics we choose. We use it to develop the software, to regression-test every application every night against a known baseline, and to show the result of a test before it runs on real hardware.
| Application | Scenarios | Pass | vs baseline |
|---|---|---|---|
| Random | 62 | 62 | 0 |
| Sine | 48 | 47 | 1 |
| Shock | 36 | 36 | 0 |
| Replication | 40 | 40 | 0 |
| MIMO Random | 31 | 31 | 0 |
| RMA | 24 | 24 | 0 |
| Total | 241 | 240 | 1 |
Baseline vs regression run = results · 1 FLAGGED · Sine: control error above baseline. Sent to engineering with the plots.
Development, regression and validation
A shaker lab is a slow place to find a software problem. Hardware time is scarce, a test article can be damaged, and some conditions are hard to create on purpose. The simulator puts those conditions on an engineer's desk.
Build against the hard cases
New control features are developed against rigs with sharp resonances, deep anti-resonances and strong cross-coupling, not just the well-behaved setups that are easy to find in a lab.
Every application, every night
Each night every Panther application runs a full set of scenarios, typically 200 to 300, so a change that breaks something is found the next morning, not by a customer.
Show the result, don't just claim it
When we say Panther can run a specific test to a specific result, we can run it on a simulated rig and show the control spectra, the errors and the verdict.
Your channels, your geometry, your dynamics
| Model | What we set |
|---|---|
| Control and measurement channels | Single-exciter (MISO) and multi-exciter (MIMO) control. The simulator has no fixed channel limit, so it can model systems larger than any rig we have in the building. Nightly MIMO regression covers 31 configurations from 2 × 2 to 12 × 6. |
| Shaker geometry | Number and placement of exciters and sensors, and the transforms between them, up to six degrees of freedom. |
| Injected dynamics | Resonances, anti-resonances and Q factors, with variance applied so no two runs see exactly the same rig. |
| Validation of the model | The plant simulator is validated to within 0.1 dB against measured data from a customer installation. |
Numbers from the simulator, not from a brochure
These results come from Spectral Dynamics' own simulated test system and the regression suites that run on it.
Plant simulator against measured data from a customer installation.
MIL-STD-810H 519.8 gunfire tests passed all nine checks on a resonant fixture with modes inside the gunfire band.
MIMO configurations, 2 × 2 to 12 × 6, regression-tested every night.
Unit tests that must pass before any MIMO build is accepted.
Baseline vs regression run = results
Every scenario has a baseline: a known-good result. Each night the regression run is compared against that baseline, and the difference is the result. Doing that by hand for 200 to 300 scenarios would take several engineers several weeks. AI-driven analysis does it overnight and hands our engineers a report that points straight at anything that moved.
Baseline
Each scenario has a stored, known-good baseline result to measure against.
Regression run
The day's software is built and every application runs its scenarios on the simulated test system, typically 200 to 300 in all.
Compare
AI-driven analysis compares every regression run to its baseline and to its pass criteria, looking for failures and anomalies.
Results
Engineers start the day with a report of what passed, what moved from baseline and what needs a closer look.
To run and review a few hundred scenarios across every application.
Every application, every night, with a report ready for engineering review in the morning.
One wrong sign, found before a shaker moved
In a multi-exciter test, the transform between sensors and degrees of freedom has to be exactly right. On the simulator, a single wrong sign in one row of that transform made the test abort during ramp-up. With the sign corrected, the same test passed. One variable changed, one clear answer.
On a real rig, that kind of error costs lab time and can put a test article at risk. On the simulator it costs a few minutes.
Your setup, simulated
The same simulation that tests our software can model yours. With the SD Support Package, Spectral Dynamics simulates your shaker, fixture, channels and dynamics, so you can see how a test will run before it reaches your lab.
- Your channel count and shaker geometry, single-exciter or multi-exciter.
- Your fixture and test-article dynamics: resonances, anti-resonances and Q factors.
- Your test profile, run on the simulated rig, with the control results and verdict to review.
Spectral Dynamics' simulation-based process for software development and customer support, including the SD Support Package, is patent pending. Talk to your sales representative for details.
Frequently asked
What can the simulated test system model?
Single-exciter and multi-exciter configurations, shaker and sensor geometry up to six degrees of freedom, and injected dynamics: resonances, anti-resonances and Q factors, with variance from run to run.
How do you know the simulation is realistic?
The plant simulator has been validated to within 0.1 dB against measured data from a customer installation.
Can you simulate my setup and my test?
Yes. The SD Support Package simulates customer setups, including channels, shaker geometry and dynamics, so you can see how a test will run before it reaches your lab. Talk to your sales representative for details.
Ask us to show you, not just tell you.
Simulation-backed development, nightly regression testing for Panther, and customer setups simulated with the SD Support Package. Talk to your sales representative for details.
