A frequently asked question is “How can I optimize the parameters of my stepped sine sweep (“StweepTM”)?”. If you are a regular SoundCheck user, you are aware of the many parameter control options in the Stweep stimulus editor but you may not be using them all to your advantage.
- Stimulus Type
- Stimulus Level
- Frequency resolution
- Start frequency
- Stop frequency
- Step size
- Minimum cycles
- Minimum duration
In this brief note, we will cover the basic essentials of creating a Stweep to satisfy your measurement requirements.
It is well known that the greater the measurement averaging time, the greater the measurement confidence. In some cases, there may be few, if any test time constraints (R&D) and in others, test time may be critical (production test) which requires a delicate balancing act of speed vs. accuracy.
In the SoundCheck stimulus editor, the primary tools for optimizing your Stweep are those that control the Step Size and Frequency Resolution.
Step size includes two parameters:
Min Cycles: minimum amount of sine periods that occur per step in the sweep
Min Duration: minimum time in seconds spent at each step in the sweep
Depending on the values of these two parameters, there will be a transition frequency above which the step size will be satisfied by the Min Duration value and below which, the step size will be satisfied by the Min Cycles value.
Step size also affects the width of the FFT tracking filters applied during HarmonicTrak or Heterodyne analysis and therefore the out of band noise present in the measurement.
Frequency resolution options include standard 1/n octave spacing as well as options for user defined linear spacing (e.g. 20 Hz) and user defined log spacing (e.g. 1/10 octave).
Optimization Example - Production Line Tests
The main goal is to gather accurate data in the shortest period of time. One of the trade-offs for this measurement is determining if you want a higher throughput of product or more confidence in the results of your measurement.
One of the first places to start optimizing your stimulus is the resolution of your Stweep. In some cases, people are using resolutions that are unnecessarily high for their testing purposes and this means they are wasting time and losing throughput!
When determining the maximum practical resolution of a Stweep, it is recommended to start at R10 (1/3 Octave) and incrementally increase the resolution until the measurement improvement becomes marginal. Why test using 1/24 octave stimulus when you can get the same results from 1/12 octave in half the sweep time? From here, it is the user’s choice to determine which resolution gives them the best value or this may be determined for them by an existing test specification or industry standard.
SoundCheck® also gives you the option of creating a compound Stweep where for example, the low frequency range can be configured for ⅓ octave resolution and the high frequency range configured for 1/12 octave. Sweeping through the low frequencies at a lower resolution can save a significant amount of sweep time.
A separate issue that comes up is that production facilities can be noisy, whether it be HVAC noise or machinery, and this noise can contaminate measurements and reduce their accuracy, particularly at lower frequencies.
Rejecting noise in this region can be done by increasing the Min Cycle number for the lowest frequency range of interest. This is especially applicable if you are sweeping below the resonant frequency of a speaker where the response begins to significantly roll off and approach the background noise floor. Spending more time at each step can increase the signal to noise ratio of your measurement and improve the measurement confidence. For more on measurement confidence, see the SoundCheck example sequence Confidence and Noise which can be found in your SoundCheckXX/Sequences/How To examples folder.
A useful way of applying these techniques without adding too much testing time can be done by creating a compound stimulus as shown in the table below. With this method we are able to add more noise rejection to the lower frequency band, and save time in the upper frequencies where there is less competing noise.
Lastly, when optimizing your stimulus, consider the frequency range of your sweep. If your DUT has no usable frequency response below 100 Hz there is no sense in having your sweep range extend down to 20 Hz.