GB2175406A - Testing digital audio systems - Google Patents

Abstract

A low frequency test signal, e.g. 30 Hz, from a generator (10) is applied to a digital audio system (11) under test comprising an analogue to digital converter (12) and a digital to analogue converter (13). The analogue output is applied to an amplifier (14) and a high pass filter (15) which removes the test signal frequency and passes just the quantizing error and distortion components. These are passed through an amplifier (16) to a low resolution analogue to digital converter (17) providing error value samples to a computer (19). Each such sample is accompanied by a value from another low resolution analogue to digital converter (18) responsive to the unfiltered signal. These second values indicate the segment of the transfer characteristic of the system (11) under test. The computer (19) processes the inputs to ascertain the peak to peak amplitude of the quantizing error plus distortion signal for each segment of the transfer characteristic. In another embodiment (Fig. 3, not shown) a single high resolution analogue to digital converter is used and the computer performs digital filtering to form the error value samples and truncation to form the segment values. <IMAGE>

Description

SPECIFICATION Method and apparatus for testing digital audio systems The present invention relates to a method and apparatus for testing a digital audio system, more specifically for measuring discontinuities in the transfer characteristics of a digital system. It is important to be able to detect and measure such discontinuities when measuring the distortion of audio signals by digital equipment. The normal analogue noise and "total distortion" measurment methods are insensitive to the distortion products which arise from such discontinuities. Moreover, the effect of discontinuities upon audio quality depends upon their position in the transfer characteristic [Gilchrist, N.H.C. 1982 Digital audio impairments and measurements. Audio engineering Society Conference Publication "Digital Audio".Broadly speaking, discontinuities at central coding levels produce noticeably worse effects than discontinuities at the extreme coding levels.

The object of the present invention is to provide a method and apparatus which will provide the required kind of information for proper assessment of digital audio systems.

The invention is defined in the claims below.

The invention will be described in more detail, by way of example, and with reference to the accompanying drawings, in which: Figure 1 is a block diagram of apparatus embodying the invention.

Figure 2 is a flow chart for the operations carried out by a computer in the apparatus.

Figure 3 is a block diagram of a second apparatus embodying the invention.

Figure 4 is a flow chart for the operations carried out by the computer in the apparatus of Fig. 3.

Fig. 1 shows a low frequency audio signal generator 10 acting as the source of a test signal. The test signal is a pure sine wave of constant amplitude and its frequency may be 30 Hz. The test signal is applied to the digital audio system 11 under test which includes an input analogue to digital converter 12 and an output digital to analogue converter 13. In between these converters there will typically be some form of transmission channel, signal processing or record/playback channel.

The audio output from the system under test will be the test signal contaminated by quantizing error and the distortion components arising from any discontinuities in the system 11 under test. The output signal is applied to amplifier 14 and one output from this amplifier is fed through a high pass filter 15 and another amplifier 16 to an analogue to digital converter 17. The high pass filter 15 removes the test signal itself, and the output from the filter may comprise signal components in the frequency band 300 Hz to 15 KHz. The output of the filter comprises the quantizing error and distortion components within the pass band and these components are converted into digital form by the converter 17.

A further analogue to digital converter 18 operates on the unfiltered signal from the amplifier 14. The digital signals from the two analogue to digital converters 17 and 18 are applied to a computer 19 which analyses the digital values in a manner described below.

The converters 12 and 13 within the digital audio system 11 may be assumed to be of reasonably high resolution, at least 13 or 14 bits. The digital audio system under test may retain the digital audio samples in linear form throughout or, alternatively, employ companding. The converters 17 and 18 in the measurement apparatus do not have to be of high resolution; typically they may be 4 to 12 bit converters. The converter 17 is handling only the low level distortion components and the converter 18 is required to indicate only fairly coarsely the region of measurement within the transfer characteristic of the system 11. At a resolution of 8 bits the transfer characteristic is quantized into 256 levels. In theory the second input to the computer could be taken from the most significant bits at the input to the digital to analogue converter 13 within the system 11.In practice it is less trouble and more flexible to provide a cheap additional analogue to digital converter 18 on the output of the amplifier 14.

The computer 19 receives pairs of values each pair consisting of a sample of the combined quantizing error and distortion measurements and a sample of the instantaneous level of the test signal, i.e. the position within the digital audio system transfer characteristic.

The first input may be called the error sample and the second may be called the segment number since the samples from the converter 18 identify 256 segments of the transfer characteristic, assuming 8 bits resolution. The computer 19 can readily be programmed to process the input information and provide the engineer with useful information in the form of a print out or a VDU display, e.g. a bar chart display. In the present embodiment the basic processing operations determine the peak amplitude of the quantizing error plus distortion signal in each of the segments of the transfer characteristic. The effective resolution of the digital system and the magnitudes and approximate positions on the transfer characteristic of discontinuities may be deduced from the information. It is not necessary to determine the exact points on the transfer characteristic at which discontinuities occur.

Fig. 2 is a flow chart showing a possible basis of a simple computer program. The functions of the various blocks are explained more fully below.

Block 20. Initialisation comprises normal operations such as clearing the computer stack and clearing variables including an array or table which stores three quantities for each segment of the transfer characteristic. These quantities are the maximum (most positive) error sample value denoted MAX, the minimum (most negative) error sample value denoted MIN and the peak-to-peak error value.

Block 21. The computer waits until the segment number (read in block 22) changes to indicate that a new segment is just being entered. The segment number is input from the analogue to digital converter 18, and the error sample value is input from the converter 17 (block 23). The sample rate istthe same for both of these converters, and is locked to a master clock e.g. within the computer. The main processing loop of the progrmme is also locked to this master clock.

Block 24. The main processing loop commences each time by acquiring the new pair of values constituted by the segment number and the associated error value. Because of the low resolution of the converter 18, the segment number will remain unchanged for a number of pairs of values.

Block 25. A test is made to see whether the pair of values just stored pertain to the same segment as the preceding pair of values or to a new segment.

Block 26. If in the same segment, the error value is compared with MAX and with MIN.

These comparisons are dealt with further below.

Block 27. If a new segment has just been entered the first operation is to update the peak-to-peak error value stored for the old segment. This is done by reading MAX and MIN from the table, forming the difference therebetween and overwriting the peak-topeak value in the table with the difference thus formed.

Block 28. The segment number is now updated to correspond to the new segment.

Block 29. The values MAX, MIN and peakto-peak error in the table are accessed by some form of indirect or indexed addressing and the table address is updated in accordance with the new segment number. Block 26 is again reached for the error value to be compared with MAX and MIN.

Block 30. This tests if the error value is greater than MAX. If it is MAX is overwritten by the error value (block 31).

Block 32. If the error value is less than MIN, MlN is overwritten by the error value.

After the comparisons and possible updating of MAX or MIN the program returns to block 24 to process the next pair of values.

The test signal may be at a level close to the maximum which the digital system 11 will convey without clipping. Because the access to the digital system is via analogue connections, remote digital equipment may be tested via high quality analogue telephone circuits.

In the embodiment described with reference to Figs. 1 and 2, two low resolution analogue to digital converters 17 and 18 are employed in producing the first and second digital samples. Figs. 3 and 4 relate to a second embodiment in which the output of the system 11 under test is applied to a single analogue to digital converter 35 providing digital samples to a computer 36 which carries out all the necessary operations in generating the first and second digital samples. It is a disadvantage of the embodiment of Fig. 3 that the analogue to digital converter 35 must be a high resolution, audio converter of high quality but there are compensating advantages as will be mentioned below.

Fig. 4 shows the flow chart for the computer 36. The only differences from Fig. 2 concern the way in which the first and second digital samples are generated for input to the blocks 21 and 24.

The input from the high resolution analogue to digital converter 35 (block 37) is digitally high pass filtered in block 38. An advantage of this digital filtering is that it can be performed with zero group delay. Block 39 sets the level of the error signal, i.e. performs any necessary gain adjustment corresponding to analogue amplification in Fig. 1, and provides the first digital samples.

The input from the high resolution analogue to digital converter is also truncated (block 40) to form the second digital samples which are the segment numbers associated with the first digital samples. The second digital samples are subjected to digital delay in block 41 to compensate for the absolute delay in the high pass filtering block 38. It is an advantage of this embodiment that the delay compensation can be exact.

The computer program represented by the flow chart of Fig. 4 can be used to check the analogue to digital converter 35 for any discontinuities. Moreover, the program can be used to check any other analogue to digital converter, with a resolution up to the maximum word length of the computer, providing that the sampling of the converter can be locked to the program repetition rate.

Claims (15)

1. A method of testing a digital audio system wherein a low-frequency analogue test signal is applied to the input of the system and the analogue output of the system is processed to derive first digital samples which represent the analogue output of the system filtered to remove the test signal and leave an error signal representing quantizing error and the distortion components arising from discontinuities in the transfer characteristic of the digital audio system, each first digital sample is associated with a second digital sample corresponding to the instantaneous level of the analogue output of the system, and the first and second digital samples are analysed in a computer to determine the error associated with each of a plurality of segments of the transfer characteristic of the digital audio system under test.

2. A method according to claim 1, wherein the analogue output of the system is high pass filtered and then applied to an analogue to digital converter which provides the first digital samples.

3. A method according to claim 2, wherein the second digital samples are provided from the analogue output of the system by another analogue to digital converter.

4. A method according to claim 1, wherein the analogue output of the system is applied to an analogue to digital converter whose output is digitally high pass filtered to provide the first samples.

5. A method according to claim 4, wherein the output of the analogue to digital converter is truncated to provide the second samples.

6. A method according to any of claims 1 to 5 wherein the maximum and minimum values of the first digital samples over an interval of time are stored in respect of each different second digital sample value.

7. A method according to claim 6, wherein peak-to-peak values formed as the difference between the maximum and minimum values are also stored.

8. Apparatus for testing a digital audio system having a low frequency test signal applied thereto and providing an analogue output signal representing the test signal contaminated by quantizing error and the distortion components arising from discontinuities in the transfer characteristic of the digital audio system, comprising a filter for removing the test signal from the output signal, to leave an error signal containing the quantizing error and distortion components, a first analogue to digital converter operative on the error signal to provide first digital samples, a second analogue to digital converter operative on the output signal to provide second digital samples, and a digital computer programmed to analyse the first and second samples to determine the error associated with each of a plurality of segments of the transfer characteristic of the digital audio system.

9. Apparatus according to claim 8, wherein the computer stores the maximum and minimum values of the first digital samples over an interval of time in respect of each different second digital sample value.

10. Apparatus according to claim 9, wherein the computer also forms and stores peak-to-peak values representing the difference between the maximum and minimum values.

11. Apparatus for testing a digital audio system having a low frequency test signal applied thereto and providing an analogue output signal representing the test signal contaminated by quantizing error and the distortion components arising from discontinuities in the transfer characteristic of the digital audio system, comprising an analogue to digital converter operative on the analogue output signal to provide high resolution digital samples, digital filtering means for removing the test signal from the high resolution digital samples so as to provide first digital samples representing an error signal containing the quantizing error and distortion components, truncating means operative on the high resolution digital samples to provide second digital samples, and a digital computer programmed to analyse the first and second samples to determine the error associated with each of a plurality of segments of the transfer characteristic of the digital audio system.

12. Apparatus according to claim 11, wherein the digital computer implements the digital filtering means and the truncating means.

13. Apparatus according to claim 11 or 12, wherein the computer stores the maximum and minimum values of the first digital samples over an interval of time in respect of each different digital sample value.

14. Apparatus according to claim 13, wherein the computer also forms and stores peak-to-peak values representing the difference between the maximum and minimum values.

15. Use of the apparatus according to claim 11 to test the digital output of an analogue to digital converter or other digital audio device.