Design of a DSP-based 24 bit Digital Audio Equalizer for Automotive

0 downloads 0 Views 309KB Size Report
+C37. 1u. VBATT/2. VFILT1. AVDD/2. AIM2. AIP2. 10. 9. 8. U4:3. TL074. 12. 13. 14. 11 ... 102. D4. 103. D5. 104. D6. 113. D7. 114. D8. 115. D9. 116. D10. 117. D11. 118. D12 .... A serial port consistent with the IEEE 1149.1 (JTAG) standard is ...
Design of a DSP-based 24 bit Digital Audio Equalizer for Automotive Applications Alberto Bellini*, Eraldo Carpanoni*, Giacomo Frassi*, Monica Tesauri**, Emanuele Ugolotti** *

Dipartimento di Ing. dell’Informazione, University of Parma Parco Area delle Scienze, 181/A, I-43100, ITALY [email protected] ** ASK Industries Via F.lli Cervi, 79, 42100 Reggio Emilia, ITALY

ABSTRACT The project deals with the design and realization of a dedicated board DSP-based for the processing of audio signals for automotive applications (DIGIcar). The I/O interface of the board is composed of two stereo inputs and two stereo outputs. The prototype was realized in a four layer PCB with SMD components. The layers were designed in order to reduce high frequency effects of digital signals on analog audio signals. The DIGIcar board was tested stand-alone with an Audio precision System 2022. Experiments were performed inside a car too, where the DIGIcar board was interfaced with a four channel power amplifier and connected between audio source and car loudspeakers. A development tool in MATLAB was exploited to synthesize the suitable equalizing filters given standard acoustics car measurements. Then the filters are stored in the board Flash EEPROM. A few options are available to tailor the equalizer for different cars. Listening tests and acoustic measurements show the effectiveness and the functionality of the designed board.

1. INTRODUCTION

Automotive engineering is an attractive field from any point of view, especially when dealing with accessories aimed at increasing comfort. As for hi-fi car audio systems the traditional low quality radio system is being moving to a sound system, even if the specific car compartment spatial composition makes sound reproduction a very hard task. Reverberations, reflections, echo, noise and vibrations are some of the issues to account for. A first step in the direction of increasing sound comfort is that of equalizing the acoustic pressure response in the frequency domain. To accomplish this task the inversion of the amplitude of the measured Sound Pressure Level (SPL) should be performed [1,2]. The main target of the equalization procedure is to increase sound comfort at the driver position, option which does not directly results in an

homogeneous quality in the whole car cockpit. However also spatial processing of audio source, like phase filters, delays between channels, and between rear and front loudspeaker has been found of large importance in order to increase sound quality reproduction [3,4]. Sound preprocessing results in a pleasant increase of sound quality, but often requires large computational power and high quality A/D and D/A converters, in order to prevent the introduction of quantization errors and additional conversion noise. Every car features a peculiar conformation, which moreover is influenced by the number of occupants, by their position and also by environmental conditions, like running motor noise. It is clear that common radio systems are insufficient to produce a pleasant sound reproduction. Audio Digital Processing Systems are becoming more and more wide spread in the consumer market, and they have entered car audio system too. They require peculiar hardware characteristics at a very low price, because of the target cost of a car acoustic equalizing system. From this rationale, the idea of designing a dedicated DSP board for the equalization of car cockpits was originated (DIGIcar).

2. HARDWARE DESIGN Peculiar car characteristics, power supply levels, noise floor, and typical applications drew the technical specifications of DIGIcar. The target number of channels was four audio-in and four audio-out, in order to handle the left/right front/rear loudspeakers separately. Therefore the technical specifications of the design were as follows: Bandwidth 20-20000 Hz, Sampling rate 48kHz, A/D conversion resolution 24 bit, 100 dB Signalto-Noise Ratio, Low distortion, THD< 0.01%. The DSP is a fixed point processor, since the target application is Digital Audio Processor for automotive sound system, where cost is a major concern.

C30

C45

680p J2:5

680p

R10

R12

R14

R20

10K

10K

10K

470

C34

+

5

J3:5

R28

R30

R32

10K

10K

10K

2 3

C42

RIGHT

470p

470p

U4:2

6

VBATT/2

C38

C40

10n

100n

7

AVDD/2

1n

10K

R25 56K

C53 10n

U5:2

5

R43 56K

7

5

6 TL074

TL074

R16

R21

10K

470

C35

+

R34

R39

10K

470

C50

+

AIP1

1u

FRONT

AIP3

1u

REAR C31

C46

680p R11

R13

10K

10K

680p

R17

R22

10K

470

C36

+

3

C28

LEFT

AIM2

R29

R31

10K

10K

R26 56K

8

4

C41

10n

100n

C48 AVDD/2

1n

10K

R27 56K

C54 10n

U5:4

13

R45 56K

14 12

VFILT3 TL074

TL074

R19

R23

10K

470

C37

+

AIP2

C44 100n

11

11

AIM4

TL074

VBATT/2

C39

12 C29 100n

+

1u R44 56K

R36

14

C51

8

470p

AVDD/2

1n

U4:4

13

470

10

C33

R18 10K C26 100n

R40

10K U5:3

C43

LEFT

TL074

470p

VBATT/2

R35

3

9

10

VFILT1

J3:3

1u

U4:3

9

4

J2:3

C47

R33 AVDD/2

1n

10K

100n

AIM3

R42 56K

TL074

C32

R15 C25

+

1u

1 3

TL074

VBATT/2

C49

470

U5:1

2 R24 56K

1

R38

5

U4:1

C27

RIGHT

AIM1

1u

1u

ASK Titolo : Numero :

R37

R41

10K

470

C52

+

AIP4

1u

S.p.A.

DSP PER AMPLIFICATORE 4 CANALI - SMT S.204.051.1

Codice c.s. :

S.204.051.1

Rev.

Figure 1 DIGIcar schematics analog audio input interface. Data : 18-06-01 File : S2040511B.SCH

File coll. : S2040511C.SCH

Foglio 2 di 4 Disegnato da :

The CODECs are connected to the DSP exploiting the two DSP serial ports, fig. 2. The power supply is obtained from the car battery operated by a circuitry based on LM317, which produces 3.3 Volt (for digital circuitry), 1.8 (core DSP supply), 12 Volt (opAmps power supply), with proper analog and digital ground management.

16 15

LRCLK MCLK

R51 33,2

TLC320AD77C C67 100n

S2

R65 C73 1n R66

S3

33

33

13 62 39 46 17 127 129 30 92 18 64 65 66 67 63

C74 1n R55 R56 R57 R58 R59

U7

C57 100n

+ C62 22u

3 + C65 1u 4 5 C66 100n 6

L6

7

220uH

AINRM AINRP

VREFP

AOUTR VCOM AOUTL

26 AOP3 25 VCOM2 24 AOP4

VREFM VRFILT

23

AVDD

H4 8 H3 9 H210

AVSS

11 DVDD

12 13 14

L4 220uH C55 C58 100n 10n

MOD2 MOD1 MOD0

DVDD

TLC320AD77C

14

7

8

AVDD

LRCLK MCLK

21H1 20H0 19 18 17

10K 10K 10K 10K 10K

32 98 H0 H1 H2 H3 H4

DVDD R48 10K

C68

H6 H7 R50 33,2

58 69 81 95 120 124 135 6

H6 H7

R49 10K

16 15

100n

OSC1 1

DVDD R60 R61 R62 R63 R64

22

SPDMODE PDN_RSTB DEM1 DEM0

DVSS SDIN SDOUT SCLK

10K 10K 10K 10K 10K

C70 100n

AVSSREF TEST

REAR

12,2880Mhz

AINLM AINLP

28 AIM3 27 AIP3

H6 H7

R54 100

HAS HBIL HCNTL0 HCNTL1 HCS HDS1 HDS2 HOLD HPIENA HR/W INT0 INT1 INT2 INT3 NMI

HD0 HD1 HD2 HD3 HD4 HD5 HD6 HD7

H7

H6

8 HINT/TOUT1 TOUT0 CLKOUT CLKMD1 CLKMD2 CLKMD3

U9

R86 22K

BC846B

10 19 21 22 20 24 28 55 61 29 25 26 23 27

A0 A1 A2 A3 A4 A5 A6 A7 A8 A9 A10 A11 A12 A13 A14 A15 A16 A17 A18 A19

R87 22K

U10:2 4 6 DVDD

5 74LV32

U10:1 1 3 74LV32 2 DVDD C76 R77 100n 10K R72 R81 10K 51 10K 82 94 R82 10K 77 78 10K R78 R79 10K 79 U12

TMS320VC5402

T2

T1

9 READY DS IS PS MSTRB HOLDA HRDY IACK IAQ IOSTRB MSC R/W XF

DL2

BC846B

14 7

R75 22K

R76 22K

MP/MC RS

96 97

AIM4 AIP4 C61 100n

1 2

BCLKX0 BCLKX1 BCLKR0 BCLKR1 BDR0 BDR1 BDX0 BDX1 BFSR0 BFSR1 BFSX0 BFSX1 BIO

R89 470

DL1

13112 13211 13310 134 9 136 8 137 7 138 6 139 5 14027 14126 5 23 7 25 8 4 9 28 10 29 11 3 105 2 10730 108 1 109

A0 A1 A2 A3 A4 A5 A6 A7 A8 A9 A10 A11 A12 A13 A14 A15 A16 A17 A18

M29W040B E G W DQ0 DQ1 DQ2 DQ3 DQ4 DQ5 DQ6 DQ7

22 24 31 13 14 15 17 18 19 20 21

0 1 2 3 4 5 6 7

DVDD VCC

32C77 100n

VSS

16 DVDD

R88 E 10K

R47 10K

DVDD

R84 470

11

U11:3 74LV08

48 49 41 42 45 47 59 60 43 44 53 54 31

R46 10K

17

DVDD SDIN SDOUT SCLK

CVDD

R69 10K

DVDD

21H1 20H0 19 18

SPDMODE PDN_RSTB DEM1 DEM0

DVSS

RESET RESET NC PFO PFI GND

TPS3707-33

U11:4 74LV08

13

14

22

MR

DVDD

6 5 12

7

MOD2 MOD1 MOD0

C72 1n

U11:2 74LV08

4

3 74LV08

F R85 10K

12 13 14

3

33

C75 100n

U11:1

1 11 2

C R80 10K D R83 10K

11

R52

U10:4

74LV32

A R73 10K B R74 10K

H4 8 H3 9 H210

VDD

R53 10K

C71 100n S1

TEST

FRONT

AVDD C69 100n

AVSS

12 8 13

74LV32

8 R70 10K 9 R71 10K

23

AVDD

AVSSREF

U10:3

10 1 2 8 7 6 5

142 130 128 125 112 111 106 93 91 76 75 70 68 57 56 52 50 40 34 33 16 14 4 3

VREFM VRFILT

9 R68 10K U8 4

D0 D1 D2 D3 D4 D5 D6 D7 D8 D9 D10 D11 D12 D13 D14 D15

220uH

26 AOP1 25 VCOM1 24 AOP2

CVDD DVDD DVSS CVDD DVDD DVSS DVSS DVSS CVDD DVSS DVDD DVSS CVDD DVSS DVDD CVDD DVSS DVSS DVSS DVDD CVDD DVSS DVDD DVSS

7

28 AIM1 27 AIP1

AOUTR VCOM AOUTL

TDI TRST TCK TMS TDO EMU0 EMU1/OFF

C64 100n 6

AINRM AINRP

VREFP

X1 X2/CLKIN

+ C60 22u

L5

DVDD

DVDD

AINLM AINLP

86 87 88 89 85 83 84

C56 100n

1 2

3 + C63 1u 4 5

099 1 100 101 2 102 3 103 4 104 5 113 6 114 7 8 115 9 116 A 117 B 118 C 119 121 D 122 E 123 F

U6 AIM2 AIP2 C59 100n

R67 270 DVDD

ASK Titolo :

J6

1 2 3 4 5 6 7 8 9 10 11 12 13 14

The computational power of the adopted processor is 100 MIPS, i.e. around 500 FIR taps/sample/channel at 48kHz. Therefore the following components were chosen: a TMS320VC5402 DSP [5], two stereo TMS320AD77C CODECs (integrated A/D and D/A up to 96 kHz sampling rate) [6], a watchdog, input and output interface circuitry based on discrete components and on TL074 opAmps, power supply circuits designed to be interfaced to a standard 12 V car battery. Moreover a Flash EEPROM is provided for stand-alone bootload. Input and output circuitry is specifically designed for anti-aliasing filters and to perform single-ended to double-ended signal conversion, as required by the CODECs. This conversion is needed since the companion amplifier uses single-ended signals. The opAmps are not biased in dual voltage supply, because of the peculiar application. A proper virtual ground at Vbatt/2 (half of the battery voltage) is generated by the power supply stage and is adopted for all discrete circuitry. The four channels are processed by an analog circuitry, in order to convert single ended signal provided by the audio amplifier in double ended signals, consistent with the specification of the adopted CODEC, which is a TMS320AD77C. The CODEC features a 24 bit resolution and is operated at 48kHz. The output interface is again designed with opAmps to provide necessary filtering and conditioning, fig. 1.

Numero :

S.p.A.

DSP PER AMPLIFICATORE 4 CANALI - SMT S.204.051.1

Codice c.s. :

Figure 2 DIGIcar schematics, digital interface. CONN-7X2P-P2,54

Data : 18-06-01 File : S2040511C.SCH

File coll. : S2040511D.SCH

S.204.051.1

Rev.

Foglio 3 di 4 Disegnato da :

A serial port consistent with the IEEE 1149.1 (JTAG) standard is available too, which allows for both testing and programming of the DSP for debugging purposes through a host PC. Some switches and leds are added in the board for diagnostic purposes, and to provide user defined interrupts, which may be used for instance to change filter coefficients during acoustic system design. Specific care was placed in the layout design of the input/output interface in order to obtain accurate symmetry for the four channels paths. Specifically the high frequency paths of the crystal are separated by audio signals, so that the glitches of the digital clock don’t affect the quality of the signals. A separate analog and digital ground is provided, and the board is shielded with grounded planes. The prototype board is reported in fig. 3. It was tested at first in a lab, programming the DSP through the JTAG TI emulator and Code Composer Studio. At first a simple talk-through program was used, which reproduces input signals to the output channels. Audio Precision System 2022 was used to characterize the performances of DIGIcar, and the overall THD-N of the DIGIcar board, programmed as a talk-through system is around –85dB (medium quality opAmps are probably the main responsible for this level, which is fully consistent with car-audio high quality systems).

Figure 3 DIGIcar prototype board (top) as a power amplifier daughter board (bottom).

3. CAR CHARACTERIZATION AND EXPERIMENTAL SET-UP The DIGIcar board is capable of processing the four channels separately. The performances when four different FIR are processed are up to 500 taps for channel at a sampling rate of 48kHz. As an example four filters were programmed and tested in the lab. In this case the code is hand written in assembly code, and four FIR filters are implemented with the MAC instruction in four Interrupt Service Routines, synchronized by the Frame Sync signal of the CODEC and the inner TIMER. The synthesis of the suitable filter coefficients for the digital equalizers requires at first car acoustic characterization, and then an automatic procedure to compute and store filter coefficients (fig. 4).

Figure 4 Software development system for equalizing filter synthesis.

Two software architectures are available. The former implements four FIR filters (one for each channel), while the latter four IIR filters. The former option requires the computation of the filters taps, given car acoustic measurements. The latter option is usually employed, implementing 5-bands or 7-bands parametric equalizer, mimicking common acoustic equalizer.

FIR filter taps computation is automatically performed starting on car acoustic measures, performed with a Bruel&Kjaer 4192L microphone separately for the four channels (front-left, front-right, rear-left, rear-right), digital acquisition is made with a Layla sound card by ECHO. A dummy head placed in the driver position is used to this aim. Several characteristics can be obtained in this way. The frequency spectrum of the car Sound Pressure Level can be obtained with MLS or sine-sweep measurements [7]. Then the four channel SPL can be inverted and synthesized with four minimum phase FIRs. Specifically the frequency response amplitude is processed so that the obtained spectrum convolved with the car acoustic measurements result in a target “comfort” curve. The phase of the measured frequency response is unchanged, in the sense that only linear-phase FIR are used. The “comfort” curve is not necessarily a flat curve, but it depends on the specific car physical conformation or target customers. Another option is the measurements of the car AQT [8], which provides a frequency spectrum referring to the signal dynamics and not to steady state measurements. This allow to characterize more accurately how the car inside reacts acoustically to music. Then as before the AQT frequency response is inverted and implemented with minimum phase FIR filters.

4. EXPERIMENTAL RESULTS Then as a typical example application the DIGIcar board was validated with a car cockpit acoustic equalization. The filter obtained in one of the three methods described in section 3 (IIR parametric filter, inverse-SPL FIR, inverse-AQT FIR) are convolved with the stereo audio source and sent to a 40 W 4-channel amplifier that drives the car loudspeakers. An Opel Corsa was used as the test car. It featured a standard loudspeaker system, with four channels. Four different 385 FIR filters were implemented on each channel of DIGIcar. Each filter synthesizes the inverse SPL of each channel, or the inverse AQT of each channel. Another option is the use of 5-band IIR filter, where the tuning is made on the basis of listening tests. For listening tests listeners (seated in the driver position) can blindly change between the three different equalizations and the no-equalization condition. In the following experiments are reported. Fig. 5 shows the SPL of the Front-Left channel, measured with MLS de-convolution. Fig. 6 show the electric measurements of DIGIcar Front-Left channel, an Audio Precision System 2020 was used to this aim. Fig. 7 shows a comparative spectrum analysis, addressing the effect of equalization. The two dashed lines refer to SPL-inverse and AQT-inverse equalization.

Standard SPL measurements are reported. Fig. 8 shows the same comparative analysis reporting AQT measurements. Listening tests show how the equalized car is mostly preferred, because of its increased sound image.

in lab tests (with Audio precision 2022 system) and in commercial cars. Experiments are consistent with typical requirements of car audio systems.

Figure 8 AQT measurements of Opel Corsa Front-Left, Figure 5 Opel Corsa cockpit acoustic measurements (Front-Left channel SPL).

with DIGIcar equalization (dashed lines) and without (solid line).

Three different equalizing structures are available: a 5band IIR, SPL-inverse FIR, and AQT-inverse FIR. They can be chosen at testing time depending on the specific car acoustics conformation. Both subjective and objective tests show the effectiveness and the functionality of the designed board.

REFERENCES

Figure 6 DIGIcar FL channel programmed to implement a 385 FIR filter, synthesizing the AQT inverse response.

Figure 7 SPL measurements of Opel Corsa Front-Left channel, with DIGIcar equalization (dashed lines) and without (solid line).

CONCLUSIONS A DSP-based four analog-in four analog-out board, tailored for car-audio acoustic equalization system was designed, realized in 4 layers and experimentally validated

[1] G. Cibelli, A. Bellini, E. Ugolotti, A. Farina, C. Morandi “Experimental validation of loudspeaker equalization inside car cockpits”, preprint #4898 AES 106th convention, Munich, May 1999. [2] S.T. Neely, J.B. Allen, “Invertibility of a room impulse response”, Journal of the Audio Engineering Society, May 1979, vol. 66, pp. 165-169. [3] A. Bellini, A. Cannizzaro, G. Cibelli, E. Ugolotti, A. Farina, C. Morandi, “Experimental validation of stereo dipole systems inside car cockpits”, in Proc. of ICSPAT'99, Orlando, FL, November 1999. [4] Ole Kirkeby and Philip A. Nelson, “Local sound field reproduction using two closely spaced loudspeakers”, Journal Acoustic Society of America, vol. 104, no. 4, October 1998, pp. 1973—1981. [5] Texas Instruments, TMS320VC5402 Fixed-Point Digital Signal Processor. 1998 [6] Texas Instruments, TLC320AD77C 24-Bit 96 KHz Stereo Audio Codec. 1999. [7] A. Farina, E. Ugolotti, Automatic Measurement System For Car Audio Application. Pre-print of the 104 AES Convention, 15-20 May 1998 [8] A. Farina, G. Cibelli, A. Bellini, “AQT – A New Objective Measurement of The Acoustical Quality of Sound Reproduction In Small Compartments”, preprints of AES 110, Amsterdam, Olanda, May 2001 rd

Suggest Documents