Low Level Sine Sweeps

Intepreting Results

• Sine sweeps are meant to test the satellite at a range of sinusoidal frequencies, which lets us generate a response spectrum.

  • This response spectrum helps to identify resonant frequencies.
  • It also helps us identify any damage to the structure during testing, as cracks and material yielding will change the location and amplitude of the peaks in the spectrum response.

• They are called low-level because they are not meant to be performed at the same magnitude as the high-level tests (for HERON, these were random and sine burst).

• Some reasons for differences between pre- and post-test responses include:

  1. Loss of preloading
  2. Fatigue cracks and ruptures

  This is meant to document the data analysis and lessons learned for HERON Mk II Vibration Testing in July 2021. This can be helpful for FINCH and beyond. The full structural verification report can be found here, in the restricted drive. A lot of this information comes from a series of papers that can be found here.

• Loss of Preloading: Bolt preload is when there is tension produced between a nut and bolt. When preloading is optimal, the load placed on the bolt will be distributed through the interface where the bolt is installed, meaning the bolt handles less load.

  • Vibration testing can cause relative rotation of the bolt or screw within the threads of the hole or nut.
  • It can also produce stress levels above the material's elastic limit, which can also reduce preload.
  • A good way to test this (if possible), is to refasten available bolts and rerun the sine sweep. If the response returns the original, preloading loss was the issue.
    • This was not done for HERON (as bolts were also difficult to access)
  • Another good way to identify this is torque striping: draw a stripe with aerospace marker on the bolt head and any neighbouring interface (see Figure below). If the line breaks, we can identify relative rotation.
    • This should be considered for future vibration testing.

• Fatigue Cracks: Random high stress tests can cause the material to yield before it ruptures, but cyclic high stress tests can cause a crack to form and then grow at high stress areas.

  • As cracks grow, natural frequencies can decrease, which is not good.
  • If retorquing the fasteners doesn't increase the natural frequencies back to approximately the original frequencies, you should inspect for cracks.

• These criteria are generally considered a test failure. As the test is being performed, if the frequency or response peak drops by a percentage greater than the ones listed in the table, stop the test and investigate

• Frequency OrderNatural FrequencyResponse PeakFundamental Mode10%30%Higher Order Modes10%50%

Lessons Learned

• Torque striping visible bolts in the satellite structure is a good way to identify preload loss, and understand any possible changes in frequency or response peaks.
• A 0.25g sine sweep is good for ensuring the satellite is not overtested during the sine sweep phase (remember that the sine sweep does still apply some stress to the satellite).
  • However, it may be too sensitive for some shaker tables to appropriately handle, and this should be clarified before testing.
• Sine sweep tests are not enough to determine that your satellite made it through vibe successfully! You need to do full diagnostic testing, including software, electrical, and payload (if possible or applicable)

Figure: Example of a sine sweep spectrum response

Figure: Torque Striping [[source](https://theholdline.wordpress.com/2018/02/02/torque-striping/#:~:text=Torque Striping February 2%2C 2018 cessna04n Losing a,the bolt and adjoining structure after it’s tightened.)]

Random Vibration

Background

The point of random vibration testing is to simulate ground transportation and launch conditions (from acoustic waves). The vibe table will shake the satellite at multiple frequencies simultaneously, and randomly vary the acceleration at each frequency.

The random environment is not defined by a time history (because random vibration is not predictable), so instead the acceleration spectral density is used, which is the mean square acceleration divided by the bandwidth [source]

The random environments are defined as follows [source]