Component Manufacturer Uses Calibration to Reduce Test Risk

Component Manufacturer Uses Calibration to Detect Changes in Test System Measurement Uncertainty

Reduce Risk, Cost and Measurement Uncertainty

Company

The latest wireless devices, such as front-end modules, filters, and RF switches, require more integration, better performance, and greater coverage of frequency bands. In this case study, a wireless component manufacturer produces filters for use in the latest generation of wireless devices. As the margins in devices become tighter and the stack up of margins increase, the margins available for device filter design decrease. This required the manufacturer to have even greater asurance that its test system remained within specification. It counted on the test limits to determine whether the devices pass or fail. Even a small drift or fault in the test system could lead to increased measurement uncertainty; reducing margins and dramatically increasing the risk of missing ramp-to-volume requirements.

Company:

• Industry-leading manufacturer of components for wireless devices 

Key Issues:

• How to use correction factors for S-parameter measurements to detect changes in system level performance
• Drift or faults in the system leads to greater measurement uncertainty
• Ramp to volume targets while reducing costs

Solutions:

• Keysight’s risk calculator demonstrated reduction in risk and costs Risk Calculator

Results:

• Reduced risk by 41% due to decrease in false passes from 1.56% to 0.92%
• Reduced costs by an estimated 51% annually 

Key Issues

To meet its new design specifications and cost goals, the wireless component manufacturer wanted to ensure that shipping filters met performance requirements for insertion loss, ripple deviation (flatness), Voltage Standing Wave Ratio (VSWR), and attenuation. One of the key specifications for the customer’s filter test systems was ripple deviation, which ensures that signals are not distorted in the filter passband. The test engineer uses a network analyzer to measure the reflection and transmission characteristics of the filter, also known as its S-parameters. Regular calibration of the network analyzer reduces many of the systematic errors in the test, ensuring the results are as accurate as possible. A filter with a higher ripple deviation than allowed by requirements introduces distortion to the signal passing through the filter. The resulting distortion increases the risk of a wireless device incorrectly sending or receiving data.

To meet volume requirements, the customer tested the filters according to the maximum allowed ripple deviation. The results indicated four possible production test outcomes: true pass, true fail, false fail (good products fail), and false pass (defective products pass) as shown in Figure 1. The highest risk of missing production volume targets comes from false passes, and the most significant costs are due to good filters mistakenly failed during testing. A mathematical equation defines these test outcomes for calculating risk as established by the Joint Committee for Guides in Metrology (JCGM), a world-recognized organization formed by seven international standards organizations in JCGM 106:2012 per section 9.51 .

Solutions

Real-world Data Demonstrates Reduction in Risk and Costs

Keysight presented an example of how calibrating a system can detect instrument or cable drift and system faults. The model incorporated real-world data into Keysight’s risk calculator to demonstrate the potential savings and reduction in risk when calibration is performed using Keysight customized solutions. 

In the real-world example presented to the customer, the filter specification for maximum ripple deviation was 1.5 dB. The test limit used was ±0.75 dB. Due to a fault in a connector, the measurement uncertainty of the ripple deviation measurement had increased from ±0.2 dB to ±0.5 dB. Using the proposed calibration solution with a Keysight Electronic Calibration Module, Keysight identified the fault in the connector by analyzing the correction terms. Once it was replaced with a new connector, the measurement uncertainty returned to ±0.2 dB.