Setting Production Line Audio Test Limits: 3 Methods and 4 Useful Tips

Audio test limits are perhaps the hardest part of production line testing, but it is critical that it is done correctly. If the limits are set too wide, bad units may get shipped to unhappy customers. If the limits are too narrow, good units may fail and production yields may suffer. There are several different ways of setting production line limits.

3 Methods for Audio Test Limit Setting

Compare to a reference standard, or golden unit.

With this method, an ideal speaker, or ‘golden unit’ is selected, and all measurements are compared to how this speaker measures on any given day. This means that the test environment, temperature, humidity, measurement chamber and test jigs have no influence on the measurements since they are all relative to the reference standard. A Golden sample must be carefully chosen – often it is the best fit to average from a good batch. The method of choosing, as well as how to handle any changes in the sample should be well documented to ensure consistency. Samples must be carefully stored, and re-measured periodically, for example at the beginning of every batch or shift. The disadvantages to this method are that the reference loudspeaker can deteriorate and drift over time, and after a while it needs to be replaced with a new reference loudspeaker. 

Compare to a statistical average

If you choose to compare your speaker to a statistical average, a number of approved samples are measured, and those measurements are used to set absolute limits. Once limits have been set, the samples are not re-measured. This is simpler than using a Golden Unit as there is no need to continuously re-measure, but the environment must be well-controlled and test tools and measurement equipment well maintained since any change in these will affect the absolute measurements. Setting absolute limits is not straightforward, and a significant number of speakers, all conforming to specification, is required. For each performance characteristic (except rub & buzz and loose particles), the average, standard deviation, and the Ppk for the batch of speakers must be calculated. A good starting point for all curve limits, apart from rub & buzz, is the mean curve ± 3 sigma or standard deviations. Distortion limits are a little different. For THD, a +5 sigma limit is a good starting point. Rub & Buzz limits should be  set manually to levels such that they always reject audible distortion or some maximum acceptable level of audible distortion depending on the application and price point of the product. For example a lot more distortion is acceptable in a micro speaker than a high end loudspeaker.

Absolute flat-line limit for Perceptual Methods

It’s simple to set audio test limits with perceptual distortion methods. Limits are based on audibility, which is constant regardless of frequency and even product. For perceptual Rub & Buzz measurements, the limit is a flat line at the acceptable audible distortion level. In the case of transient distortion, the threshold is  simply a count of the number artifacts above the audibility threshold, which is also constant across frequencies and products. This simplicity makes these measurements extremely fast and simple to configure.

4 Tips for Production Line Audio Test Limit Setting

1. Whatever method you use to set audio test limits, they may need to float in both level and frequency to overcome reflections from fixturing and from the test box itself. 
2. It’s important to measure devices multiple times for Gage R&R, to understand any variation in the product and microphone measurement positioning to ensure measurement repeatability. There is often a certain amount of trial and error involved in setting limits, tweaking them until there is acceptable yield  and good correlation with human listeners.
3. Correctly setting limits is an extremely complex process, and it’s well worth being on-site to assist with limit-setting when you are putting a device into production so you can truly see and hear what is going on. 
4. At the end of the day, your limits need to correlate well with what you hear, so make sure units that pass sound good, and units that fail sound bad. If you can find units that sound marginally or borderline good, these are great to use as a sanity check – they should barely pass your audio test limits!

Setting Audio Production Line Test Limits in SoundCheck

Check out chapter 19 of the SoundCheck manual for detailed information on the various methods for setting production line audio test limits that are available in SoundCheck.