Sie sind auf Seite 1von 4
Application and Technical Support for Audio Precision APWIN Users Computing Group Delay from Phase Data Using

Application and Technical Support for Audio Precision APWIN Users

Computing Group Delay from Phase Data Using APWIN

TN-29
TN-29

Introduction

Group delay is a property of a device or a system: a plot of the change in phase of the response as a function of frequency; it is the negative derivative (slope) of the phase-vs-frequency characteristic of a device.

Group delay is of interest to audio engineers, particularly in the design and test of low-pass filters used in digital audio and in loudspeaker design areas. For instance, an anti-aliasing filter will typically have a phase response curve which slopes sharply down at high frequencies. This means that the high-frequency components will be delayed longer in their passage through the filter, resulting in a loss of precision in musical transients and a more diffuse stereo image. It is possible to correct the group delay distortion of such filters by using an all-pass network, but this is seldom done in practice.

A display of the Analog Analyzer Phase.AT2 sample file is shown in Figure 1. The plot
A display of the Analog Analyzer Phase.AT2 sample
file is shown in Figure 1. The plot shows the frequency
and phase response of a sweep through a 20-kHz
filter. Figure 2 shows the same frequency response
along with the group delay response.
Figure 1. Frequency and
phase response of a sweep
through a 20-kHz filter.
Constant group delay across the
frequency band means that all
portions of a wideband signal arrive
simultaneously. A pure time delay,
equal at all frequencies, gives a level
straight-line plot of phase versus
frequency. In an audio component,
this plot may vary with frequency,
and the component is said to
produce group delay distortion.

Figure 2. Resultant test showing frequency response and group delay

How Do I Do This

Extract the self-extracting file Group Delay.EXE. This will produce 10 files. The at1, at2, and at2c file are example tests for the System One, System Two, and System Two Cascade. The two .txt files and the .bmp are for information displayed when the Macro (.abp) is run. The Group Delay. apb file is the Macro/Procedure file that makes it all happen.

Start the APWIN software and load one of the test files (.at?) as appropriate for your system. There are two test files for each system.

The Anlr Phase.at? test files contain test settings to use the analog analyzer to measure the Phase and Frequency response on your device. The MLS Phase.at? files are vary similar but use the digital generator in your system to produce the MLS sequence to be used as a test signal.

Don't forget to adjust the generator output amplitude and verify that the input and output configurations are correct for your device.

Making Phase Measurements Using the Analog Generator and Analyzer

Run the test and collect real data on your device.

Now that you have loaded a test and run it, it is time to display the group delay calculations. To do this load the Group Delay.apb macro file by selecting the Open/Procedure from the APWIN File menu. Now that the Macro/Procedure is loaded run it by selecting Run from the APWIN Procedure menu. See the “How It Works” section below.

Making Phase Measurements Using (MLS) Maximum Length Sequence

There are a few things to be aware of when using MLS as a test signal to measure the phase characteristics of a test device.

Each of the Audio Precision systems have a different internal delay. Because of this we have produced three tests, one for each system, that have the appropriate time delay entered into the Digital Analyzer Quasi - Anechoic Acoustical Tester Time Delay field for a sample rate of 48 kHz. The System One delay is approximately 21 s (1 sample), System Two 63 s (3 samples), and System Two Cascade is 291 s (14 samples). Use the test for your system as a starting point for your Phase data collection. If you use a different sample rate compute the delay by using the following formula:

Number of samples / Sample rate.

When making phase measurements using MLS remember that this is a two-step process. First you set the test up to produce the Impulse response, then you configure the test setup to produce the Phase and Frequency response. Our examples have this setup for you. After loading the test and adjusting input and out settings press GO on the Sweep panel or F9 to run the test and produce the Impulse response. Then press the Spectrum/Waveform tool bar button or Sweep panel button to compute the Frequency and Phase responses.

How It Works

The procedure scans through memory to find suitable phase data. This data may have been loaded from a file, or created by a just-run test. If suitable data is not found, a prompt is displayed informing the user as shown in Figure 3.

How Do I Do This Extract the self-extracting file Group Delay.EXE. This will produce 10 files.

Figure 3. Data Not Found window

If suitable data is found, a prompt (Figure 4) informs the user that the current test is being saved.

How Do I Do This Extract the self-extracting file Group Delay.EXE. This will produce 10 files.

Figure 4 Saving Current Test window

The dialog box of Figure 5 may appear if the analyzer encounters phase wrapping as shown in Figure 9. See the following section on Data Anomalies for more information regarding the affects of phase wrapping. Click “Yes” to remove phase wrapping. However, some devices may have large phase shifts that are normal, and therefore you may not want to unwrap the display.

The dialog box of Figure 5 may appear if the analyzer encounters phase wrapping as shown

Figure 5. Unwrap phase data dialog box

The dialog box of Figure 5 may appear if the analyzer encounters phase wrapping as shown

Figure 7. Log spacing where the spacing is too close at low frequencies

The Graph displaying the Group Delay response is then updated. If frequency response data was available in the source data, it is also displayed along with the group delay. A dialog box (Figure 6) allows you to elect to restore the original data or retain the Group Delay response before the procedure terminates.

The dialog box of Figure 5 may appear if the analyzer encounters phase wrapping as shown

Figure 6. Save Group Delay data or Restore original test dialog box

Click NO to keep the current data, then click File > Save to save it.

Data Anomalies

In the graph shown in Figure 6 note the low frequency artifact that looks like ringing in the group delay response. This artifact is caused because the phase measurements are closely spaced together at low frequencies. To reduce this artifact use fewer steps in the sweep or use a linear sweep with wider frequency spacing.

This artifact will also show up across the frequency band in a linear sweep if the frequency spacing is too close as shown in Figure 8.

The dialog box of Figure 5 may appear if the analyzer encounters phase wrapping as shown

Figure 8.

Linear spacing too close across the entire frequency band

This noise artifact is produced because of phase meter quantization. As the difference in phase and frequency decreases the quantization becomes more evident.

Frequency spacing is a compromise. If the spacing is to great then you run the risk of missing complete 360 degree shifts in the phase response. This type of data anomaly is shown in Figure 9.

Figure 9. Example of Phase wrapping In this example the phase mode was set to 180

Figure 9. Example of Phase wrapping

In this example the phase mode was set to 180 to -180 deg and the measurement wrapped. When this type of phase data is used to calculate the group delay a spike will appear in the group delay response as shown in Figure 10.

Figure 9. Example of Phase wrapping In this example the phase mode was set to 180

Figure 10. Group Delay curve showing phase wrapping characteristic

Figure 11 shows a zoomed in view of the group delay response. Notice that the group delay curve is very similar to Figure 2 and 8 except for the spike and quantization artifacts.

Figure 9. Example of Phase wrapping In this example the phase mode was set to 180

Figure 11. Group Delay with negative spike caused by phase wrapping

TECHNOTES are hints and information to assist APWIN users to perform specialized tasks or unusual tests.

Figure 9. Example of Phase wrapping In this example the phase mode was set to 180

Audio Precision

5450 SW Arctic Drive

Beaverton Oregon 97005

Tel: 503-627-0832 Fax: 503-641-8906

US Toll Free: 1-800-231-7350 email: info@audioprecision.com web: audioprecision.com

Copyright © 2001 Audio Precision, Inc. TN29-0152-01-T-0401-5C-BB