a HUSH Stereo Noise Reduction System with Adaptive Threshold SSM2000 PIN CONFIGURATION 24-Lead Plastic DIP 24-Lead SOIC L IN 1 R IN 2 L VCF C1 3 L VCF C2 4 V+ 5 ACOM 6 24 L OUT 23 R OUT 22 R VCF C1 21 R VCF C2 FEATURES Up to 25 dB of Noise Reduction from Virtually Any Audio Source without Sonic Artifacts "Single-Ended" Operation Eliminates Need for EncodeDecode Process Adaptive Threshold Dynamically Adjusts to Changing Nominal Signal Levels Effectively Decodes Dolby B(R) Encoded Sources Direct VCA Control Port Access for Additional Level Control Functionality Logic-Controllable Bypass and Muting 100 dB Dynamic Range (Noise Reduction OFF) 0.02% Typical THD+N (@ 1 kHz, Noise Reduction OFF) +7 V to +18 V Operation No Royalty Requirements APPLICATIONS Auto Radio Sound Processing Multimedia PC Sound Cards Television Sound Processing Cassette Tape Players AM/FM Receivers Telephone & Wireless Links Professional Audio GENERAL DESCRIPTION SSM2000 20 V- TOP VIEW 19 DIGITAL GND VCA PORT 7 (Not to Scale) 18 NC VCF DET IN SUM OUT 8 9 17 MUTE 16 DEFEAT 15 AUTO THRESHOLD CAP 14 DEFAULT THRESHOLD 13 NC VCA DET IN 10 VCF DET CAP 11 VCA DET CAP 12 NC = NO CONNECT 6.8F L IN 6.8F 1 2 3 1nF 24 23 22 1nF L OUT R OUT R VCF CAP 1 R VCF CAP 2 V- DIGITAL GND DO NOT CONNECT1 MUTE DEFEAT 0.22F R IN L VCF CAP 1 L VCF CAP 2 V+ ANALOG COMMON (V+/2) VCA CONTROL PORT 20nF 2nF The SSM2000 is an advanced audio noise reduction system based on proprietary HUSH(R) circuitry. HUSH combines a dynamic filter and downward expander to provide a high level of effectiveness without the sonic artifacts normally associated with noise reduction systems. In addition, an Adaptive Threshold circuit detects nominal signal levels and dynamically adjusts both thresholds, thereby providing optimal results regardless of program source. Since it is a single-ended system, HUSH can be used on virtually any audio source, including audio and video tapes, radio and television broadcasts, or any other source with objectionable noise. The SSM2000 can be used with Dolby B encoded sources with excellent results. A key feature is direct access to the Voltage Controlled Amplifier port enabling additional functions such as dc volume control, automatic leveling, compression, etc. with minimal external circuitry. Dolby B is a registered trademark of Dolby Laboratories, Inc. HUSH is a registered trademark of Rocktron Corporation. 4 5 21 20 SSM2000 6 7 8 500 1k 19 18 17 16 15 14 13 VCF DET IN 20nF 9 SUM OUT VCA DET IN 1.0F 3F 10 VCF DET TC 3.3F AUTO THRESHOLD CAP DEFAULT THRESHOLD2 DO NOT CONNECT1 11 12 VCA DET TC NOTES: 1MAKE NO CONNECTION TO PINS 13 AND 18 2DEFAULT THRESHOLD. NORMALLY CONNECTED TO ANALOG COMMON Figure 1. Typical Basic Application REV. 0 Information furnished by Analog Devices is believed to be accurate and reliable. However, no responsibility is assumed by Analog Devices for its use, nor for any infringements of patents or other rights of third parties which may result from its use. No license is granted by implication or otherwise under any patent or patent rights of Analog Devices. (See page 16.) (c) Analog Devices, Inc., 1996 One Technology Way, P.O. Box 9106, Norwood, MA 02062-9106, U.S.A. Tel: 617/329-4700 Fax: 617/326-8703 (VS = +8.5 V, ACOM = VS/2, f = 1 kHz, RL = 100 k , 0 dBu = 0.775 V rms. TA = +25 C, Noise Reduction and Adaptive Threshold enabled (Pin 14 at VS/2), unless otherwise noted.) Parameter AUDIO SIGNAL PATH Signal-to-Noise Ratio Headroom Dynamic Range Total Harmonic Distortion Effective Noise Reduction Symbol SNR HR THD+N Conditions Min SSM2000 Typ Max 86 4.5 91 0.02 15 10 8 7 300 60 -85 1 37 3 37 3.8 -70 20 22 1 +1 26 10 18 11 Units dB dBu dB % dB dB k pF dB dB dB kHz kHz kHz k dB mV/dB mV V mA dB k SSM2000-SPECIFICATIONS Input Impedance ZIN Output Impedance, Dynamic ZOUT Capacitive Load Channel Separation Mute Output Gain Matching, L & R Channels Gain Bandwidth GBW DYNAMIC FILTER Minimum Bandwidth Maximum Bandwidth VCA CONTROL PORT Input Impedance VCA Voltage Gain Range Gain Constant Control Feedthrough POWER SUPPLY Voltage Range Supply Current Power Supply Rejection VCA, VCF DETECTOR Input Impedance BWMIN BWMAX VIN = 0 V, 20 Hz to 20 kHz (Flat) 80 Clip Point, THD = 1% Clipping to Noise Floor VIN = 300 mV rms, 2nd & 3rd Harmonics1 20 kHz Bandwidth (Flat) Downward Expander Section Dynamic Filter Section Pins 1 and 2 6 Pins 23 and 24 No Oscillation f = 1 kHz, VIN = 300 mV rms VIN = 300 mV rms VCA at AV = 0 dB NR Disabled VCF C = 0.001 F 0.04 AV Pin 7 VIN = 300 mV rms (Pin 7 = 2.0 V & 0 V) Pin 7 VS ISY PSRR+ RIN Pins 8 and 10 +7.0 7.5 70 4.0 5.4 7.0 NOTES 1 NR in defeat mode. Specifications subject to change without notice. PASSBAND FREQUENCY RESPONSE 0.5 dB 0 -10 -20 -30 -40 INPUT LEVEL -50 -60 -70 -80 -90 -100 0 -10 -20 -30 -40 -50 -60 -70 -80 -90 -100 MASKED NOISE FILTERED NOISE fC = 3 kHz TO 37 kHz OUTPUT LEVEL AUDIO SIGNAL 6 dB/OCTAVE SLOPE Figure 2. Downward Expander Attenuation Characteristic (In dBt; 0 dBt = Threshold) Figure 3. Dynamic Filter Characteristic (For CF = 0.001 F Defeat Mode) -2- REV. 0 SSM2000 PIN DESCRIPTION Pin # 1 2 3 4 5 6 Name L IN R IN L VCF C1 L VCF C2 V+ ACOM Function and Explanation Left Audio Input. Right Audio Input. Left VCF Filter Cap Port. Left VCF Filter Cap Port. Positive Supply. Analog Common Voltage. Internal Circuit Ground Point. Must be held to a voltage approximately halfway between V+ and V-. Should be a clean low impedance voltage source capable of at least 4 mA, such as from a buffer driven off a supply-splitting voltage divider. Input to VCA Control Port. Allows external adjustment of attenuation with a +22 mV/dB slope. Zero volts relative to ACOM gives zero additional attenuation. Should be connected to ACOM if function not required. Input to VCF Detector. Combined Left Plus Right Output. Input to VCA Detector. Voltage Controlled Filter. Time Constant Capacitor Port. Voltage Controlled Amplifier. Time Constant Capacitor Port. Pin # 13 14 Name NC DEFAULT THRESHOLD Function and Explanation Make No Connection to this Pin. Default Threshold Adjustment Port. Allows reduction of noise reduction action if signal source is relatively clean such as CD sources. Normally connected to Analog Common. -1.2 V is recommended for CDs; -2 V completely defeats noise reduction. 15 7 VCA PORT 16 17 18 19 20 21 22 23 24 8 9 10 11 12 VCF DET IN SUM OUT VCA DET IN VCF DET CAP VCA DET CAP AUTO THRESHOLD CAP Automatic Threshold Detector Capacitor Port. DEFEAT Logic High: Defeats Noise Reduction. Logic Low: Normal Action. Connect to Ground if not used. MUTE Logic High: Mute. Logic Low: Normal Action. Connect to Ground if not used. NC Make no connection to this pin. DIGITAL GND Digital Logic Reference (Logic Zero). V- Negative Supply. R VCF C2 Right VCF Capacitor Port. R VCF C1 Right VCF Capacitor Port. R OUT Right Audio Output. L OUT Left Audio Output. ABSOLUTE MAXIMUM RATINGS ORDERING GUIDE Supply Voltage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . +18 V Audio Input Voltage . . . . . . . . . . . . . . . . . . Supply Voltage(s) Control Port Voltage (Pin 7) . . . . . . . . . . . . . . Positive Supply Default Override (Pin 14) . . . . . . . . . . . . . . . . . . . . . . . . . V+ Defeat (Pin 16) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V+ Mute Override (Pin 17) . . . . . . . . . . . . . . . . . . . . . . . . . . . V+ Operating Temperature Range . . . . . . . . . . . . -40C to +85C Storage Temperature Range . . . . . . . . . . . . -65C to +150C Junction Temperature (TJ) . . . . . . . . . . . . . . . . . . . . . +150C Lead Temperature (Soldering, 60 sec) . . . . . . . . . . . . . +300C ESD RATINGS Model SSM2000P SSM2000S Temperature Range -40C to +85C -40C to +85C Package Description 24-Pin Plastic DIP 24-Pin SOIC Package Option N-24 SOL-24 PACKAGE INFORMATION Package Type 24-Pin Plastic DIP (P) 24-Pin SOIC (S) 1 JA JC Units C/W C/W 883 (Human Body) Model . . . . . . . . . . . . . . . . . . . . . . 2.5 kV EIAJ Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 300 V 54 72 27 24 NOTE 1 JA is specified for the worst case conditions, i.e., JA is specified for device in socket for P-DIP packages; JA is specified for device soldered onto a circuit board for surface mount packages.. CAUTION ESD (electrostatic discharge) sensitive device. Electrostatic charges as high as 4000 V readily accumulate on the human body and test equipment and can discharge without detection. Although the SSM2000 features proprietary ESD protection circuitry, permanent damage may occur on devices subjected to high energy electrostatic discharges. Therefore, proper ESD precautions are recommended to avoid performance degradation or loss of functionality. WARNING! ESD SENSITIVE DEVICE REV. 0 -3- SSM2000 APPLICATIONS This applications example (Figure 4) demonstrates some of the additional functions that can be easily implemented through use of the VCA Control Port (Pin 7). These functions can be selected to operate simultaneously with the basic HUSH noise reduction function. The VCA detector output provides a signal which can be further processed using one quad op amp (OP482) to provide these various additional useful functions. The circuit shown in the figure below achieves: (1) Compression, (2) Volume Leveling, (3) DC Volume Adjustment. JA1 RCA LEFT IN JA2 RCA RIGHT IN C1 6.8F 1 C2 6.8F C3 0.001F 2 3 4 24 23 22 21 SIMPLIFIED THEORY OF OPERATION JA3 RCA LEFT OUT JA4 RCA RIGHT OUT The input audio signals are processed to extract information concerning the frequency distribution and amplitude of both the desired signals and the undesired noise. Left and right audio signals are passed through voltage controlled (low-pass) filter, and then through a voltage controlled amplifiers. Both the filters and amplifiers are low distortion, and add negligible noise of their own. The VCF has a user-definable cutoff range that is usually set from 1 kHz to 35 kHz. It can be set to cover other ranges as required by the application. The VCA can be adjusted from a loss to a gain, adding to the SSM2000's versatility. In accordance with proprietary patented algorithms, control signals are derived and applied to both the VCF and VCA, resulting in up to 25 dB noise suppression with the minimum loss of desired signals. VOLTAGE CONTROLLED FILTER VOLTAGE CONTROLLED AMPLIFIER C10 0.001F SSM2000 SW1-F SIDE CHAIN OPEN = OUT CLOSED = IN C4 0.0022F R1 1k C5 0.022F +5V 5 6 7 8 9 10 11 12 C9 3.3F 20 19 +5V 18 17 16 15 14 13 NC R4 3k NC R3 3k -5V RIN ROUT SW1-C MUTE OPEN = MUTE CLOSED = OFF LIN VOLTAGE CONTROLLED FILTER VOLTAGE CONTROLLED AMPLIFIER LOUT C7 2.2F C8 1F C11 0.22F +5V SW1-B BYPASS OPEN = BYPASS CLOSED = HUSH DEFEAT LOGIC INTERFACE LOGIC INTERFACE EXTERNAL VCA CONTROL PORT R2 510 C6 0.022F 10% MUTE DEFAULT THRESHOLD DIFFERENCE AMPLIFIER SW1-A OPEN = COMP CLOSED = LEVEL +5V R8 20k R10 5.1k R9 7.5k R16 1.5M 3 2 U1-A OP482 1 AUTO THRESHOLD NEGATIVE PEAK DETECTOR LEVEL THRESHOLD R7, 10k PT10V -5V R11 12k HIGH-PASS FILTER 1N4148 PEAK DETECTOR PEAK DETECTOR DIFFERENCE AMPLIFIER SW1-D OPEN = LEVEL CLOSED = COMP 13 SW1-E OPEN = LEVEL CLOSED = COMP R15 10M -5V Figure 5. SSM2000 System Block Diagram 14 12 R14 100k C12 1F NON-POLARIZED 10% POLY, 63V R18 30k R19 20k D2 1N4148 -5V U1-D OP482 R12 20k R13 20k 9 8 10 VOLUME CONTROL VR1, 10k LINEAR R17 20k 6 7 5 U1-C OP482 U1-B OP482 Figure 4. Typical Dual Supply Application and Test Circuit -4- REV. 0 Typical Performance Characteristics-SSM2000 10 TA = +25C VS = 4.25V (Pin 5, 20) DEF THRESHOLD = 0V (Pin 14) DEFEAT = ON (Pin 16) MUTE = OFF (Pin 17) RL = 100k BOTH CHANNELS 0 -10 -20 OUTPUT SIGNAL - dBu 1 -30 -40 -50 -60 -70 -80 -90 VTH = -1.2V VTH = 0V VTH = -2.0V TA = +25C VS = 4.25V (Pin 5, 20) DEF THRESHOLD = 0V (Pin 14) DEFEAT = OFF (Pin 16) MUTE = OFF (Pin 17) RL = 100k 0dBu = 0.775mVrms -10 0 THD + N - % 0.1 0.01 0.01 0.1 INPUT SIGNAL - Vrms 1 2 -100 -70 -60 -50 -40 -30 -20 INPUT SIGNAL - dBu @ 1kHz Figure 6. THD+N (%) vs. Input Signal Figure 9. Output vs. Input Characteristics (With Different Adaptive Threshold Voltage, Pin 17) 1 TA = +25C VS = 4.25V (Pin 5, 20) VIN = 300mVrms SINE WAVE (Pin 1, 2) DEF THRESHOLD = 0V (Pin 14) DEFEAT = ON (Pin 16) MUTE = OFF (Pin 17) RL = 100k BOTH CHANNELS 10 0 L, R CHANNEL OUTPUT - dB -10 -20 -30 -40 -50 -60 -70 -80 TA = +25C VS = 4.25V (Pin 5, 20) VIN = 300mVrms SINE WAVE (Pin 1, 2) DEF THRESHOLD = 0V (Pin 14) DEFEAT = OFF (Pin 16) MUTE = OFF (Pin 17) RL = 100k BOTH CHANNELS -0.5 0 0.5 1 1.5 2 VCA CONTROL VOLTAGE (Pin 7) - V 2.5 3 0.1 THD + N - % 0.01 0.001 20 100 1k FREQUENCY - Hz 10k 30k -90 -1 Figure 7. THD+N (%) vs. Frequency Figure 10a. Channel Output vs. VCA Control Voltage (Pin 7) 6 4 2 0 3.00 0.7 -1.6 -3.9 GAIN - dB -6.2 8kHz -8.5 -10.8 -13.1 -15.4 -17.7 100k 200k 50kHz 40kHz 20kHz GAIN - dB -2 TA = +25C -4 -6 -8 VS = 4.25V (Pin 5, 20) VIN = 300mVrms SINE WAVE (Pin 1, 2) DEF THRESHOLD = 0V (Pin 14) DEFEAT = ON (Pin 16) MUTE = OFF (Pin 17) RL = 100k BOTH CHANNELS TA = +25C VS = 4.25V (Pin 5, 20) VIN = 300mVrms SINE WAVE (Pin 1, 2) 6kHz DEF THRESHOLD = 0V (Pin 14) DEFEAT = OFF (Pin 16) 4kHz MUTE = OFF (Pin 17) RL = 100k 100 - 2kHz MIN BW 100 1k FREQUENCY - Hz 10k 80k -10 -12 20 100 1k 10k FREQUENCY - Hz -20.0 20 Figure 8. -3 dB Bandwidth Figure 10b. VFC Characteristics (Simulated Noise Level 90 mV rms with Different Frequency) Applied through a 3-Pole HP filter REV. 0 -5- SSM2000-Typical Performance Characteristics 10 VC = 0V 0 -10 CHANNEL OUTPUT - dB -20 -30 -40 -50 -60 -70 -80 -90 20 100 1k FREQUENCY - Hz VC = 2V -70 VC = 2.5V -80 20 TA = +25C VS = 4.25V (Pin 5, 20) VIN = 300mVrms SINE WAVE (Pin 1, 2) DEF THRESHOLD = 0V (Pin 14) DEFEAT = OFF (Pin 16) MUTE = OFF (Pin 17) RL = 100k VC = 1V PSRR+ - dB -20 VC = 0.5V -30 TA = +25C VS = 4.25V (Pin 5, 20) V+ RIPPLE = 425mVp-p DEF THRESHOLD = 0V (Pin 14) DEFEAT = ON (Pin 16) MUTE = OFF (Pin 17) NO DECOUPLING CAPACITOR ON V + -40 -50 VC = 1.5V -60 10k 30k 100 1k FREQUENCY - Hz 10k 30k Figure 11. Channel Output vs. Frequency for Different VCA Control Port Voltages Figure 14. PSRR+ vs. Frequency -40 TA = +25C VS = 4.25V (Pin 5, 20) DEF THRESHOLD = 0V (Pin 14) DEFEAT = ON (Pin 16) MUTE = OFF (Pin 17) RL = 100k VIN = 300mVrms SINE WAVE (Pin 1 or 2) UNSELECTED INPUT = 0V (Pin 1 or 2) -20 TA = +25C VS = 4.25V (Pin 5, 20) V - RIPPLE = 425mVp-p DEF THRESHOLD = 0V (Pin 14) DEFEAT = ON (Pin 16) MUTE = OFF (Pin 17) NO DECOUPLING CAPACITOR ON V - -45 CHANNEL SEPARATION - dB -30 -50 -40 PSRR- - dB -55 -50 -60 -60 -65 -70 -70 20 100 1k FREQUENCY - Hz 10k 20k -80 20 100 1k FREQUENCY - Hz 10k 30k Figure 12. Channel Separation vs. Frequency Figure 15. PSRR- vs. Frequency -50 -55 TA = +25C VS = 4.25V (Pin 5, 20) SUPPLY CURRENT- mA 14 13 TA = +25C 12 11 10 9 8 7 6 5 4 4 6 8 10 12 14 SUPPLY VOLTAGE - V 16 18 20 MUTED OUTPUT LEVEL - dB -60 -65 -70 -75 -80 -85 -90 -95 VIN = 300mVrms SINE WAVE (Pin 1, 2) DEF THRESHOLD = 0V (Pin 14) DEFEAT = OFF (Pin 16) MUTE = ON (Pin 17) RL = 100k BOTH CHANNELS -100 20 100 1k FREQUENCY - Hz 10k 30k Figure 13. Muted Output Level vs. Frequency Figure 16. Supply Current vs. Supply Voltage -6- REV. 0 SSM2000 APPLICATIONS INFORMATION Introduction The SSM2000 is a dual channel audio noise reduction IC which reduces noise through a combination of variable filtering and downward expansion in conjunction with a unique adaptive noise threshold detector. These two techniques yield an overall noise reduction of up to 25 dB on AM and FM radio, open reel and cassette tape, CD, Dolby B encoded programming, broadcast studio-transmitter links, telephone lines, and other audio sources without the need for any additional manual adjustment. The HUSH Noise Reduction System as implemented in the SSM2000 has been shown to substantially reduce noise in PC multimedia, intercom systems, teleconferencing systems, mobile communications, automotive audio, home stereos and televisions, and other consumer and professional audio applications while preserving full signal fidelity and transparency. A brief explanation of the nature of audio signals is useful before discussing the details of the SSM2000's operation. First, audio signals have both amplitude and frequency content. Music and voice are created by changing both the amplitude and frequency of sound waves as shown in Figure 17. Second, the highest audio signal amplitudes typically occur at low frequencies (100 Hz-1 kHz) and taper off exponentially as frequency increases. Third, when an audio signal is recorded, audio noise is also generated. This noise is the white noise "hiss" or waterfall sound that is easily heard on taped material. Audio noise for a given source and bandwidth does not, in general, change with frequency or amplitude. However, noise levels do change between different sources, because of differences in recording equipment, media, and the surrounding environment. Fourth, psychoacoustic effects mask noise that occurs at or near the frequency of the audio signal. Finally, audio noise is usually considered to be most objectionable in the 3 kHz-8 kHz bandwidth region. NORMAL AUDIO SYSTEM FREQUENCY AND AMPLITUDE ENVELOPE AMPLITUDE occurs above the highest signal frequency. This noise reduction method utilizes a Voltage Controlled Filter (VCF) and is the basic method of operation in the DNR(R) system, which provides about 10 dB of noise reduction. The HUSH system shown in Figure 34 combines elements of both of these techniques to achieve 25 dB of noise reduction, and also has significant improvements. 10 SSM2000 VCF DETECTOR NOISE THRESHOLD DETECTOR VCA DETECTOR 16 PRE-DETECTOR 8 SIGNAL CONDITIONING 9 x3 OUT AMP VCF VCA HUSH DEFEAT 24,23 1,2 L/R IN L/R OUT 3,22 4,21 7 17 VCF FILTER CAP VCA EXTERNAL CONTROL PORT MUTE Figure 18. SSM2000 Block Diagram (One Audio Channel Shown) Recalling that the noise floor changes with different audio sources due to recording equipment, media, and the environment, we realize that fixed threshold approaches cannot yield optimal results. The HUSH Noise Reduction System incorporates an automatic noise threshold detector that senses these changes and adapts the VCA and VCF to become more or less aggressive depending on the amplitude of the noise floor. To determine the amplitude of the noise floor the SSM2000 assumes that the averaged amplitude during short periods of no audio is equal to the noise floor. This assumption works well in audio applications, but does have implications for evaluation and testing that are discussed in the "Evaluating the SSM2000" section. Variable Low-Pass Filtering MUSIC'S FREQUENCY AND AMPLITUDE ENVELOPE NOISE FREQUENCY The audio signal is first passed through a single-pole low-pass Voltage Controlled Filter (VCF). Both the Left and Right VCFs are controlled by a detector which places their cutoff frequencies just beyond the highest audio signal frequency. Since the highest audio signal frequency constantly changes, the VCF's cutoff frequency must also change in concert with the audio signal to avoid attenuating desired signal. For example, with signal levels below the filter threshold (presumed to be mostly noise), the VCF shuts down to about 1 kHz, providing noise reduction in the critical 3 kHz- 8 kHz band, as Figure 19 illustrates. The VCF progressively "opens up" as higher frequency amplitudes are detected at the inputs. The VCF's cutoff reaches 20 kHz when the high frequency signal amplitude is 30 dB above the threshold. At this point the VCF is acoustically transparent. The VCF's cutoff frequency range is between 1 kHz to 35 kHz. The minimum range of the VCF is limited to 1 kHz for two reasons: one, to avoid high frequency loss at the leading edge of transients, because the lower the minimum cutoff the longer it takes the VCF to slew "open," and two, noise is most objectionable at mid and high range frequencies. Once again, if the VCF characteristics are to be measured, then please refer to the section on "Evaluating the SSM2000." DNR is a registered trademark of National Semiconductor Corporation. Figure 17. A Frequency-Amplitude Plot of the Audio Envelope An analog noise reduction system must first distinguish between the desired source material and the undesirable noise. It must then attenuate the noise while leaving the source material unaffected. One approach to noise reduction is to assume that signal below a predetermined amplitude is noise, and to then attenuate the noise by using a Voltage Controlled Amplifier (VCA). A variation of this noise reduction method is found in Dolby B cassette tape systems. This method achieves about 10 dB of improvement in signal-to-noise ratio (SNR). This system incorporates a high-frequency compressor on the recording side, and a high frequency expander during playback. Another noise reduction technique senses and reduces noise by measuring the frequency content of the audio signal and then filters noise that REV. 0 -7- SSM2000 a poor quality connection is very near the gate's switching level. BYPASS FREQUENCY RESPONSE 0.5 dB fc = 1kHz to 35kHz AUDIO SIGNAL 6dB/OCTAVE SLOPE MASKED NOISE FREQUENCY FILTERED NOISE Figure 19. A Frequency Amplitude Plot Illustrating the Noise Reducing Effect of the VCF The SSM2000 has been designed to minimize control feedthrough. Control feedthrough may cause an audible output as the internal control lines of the VCAs and VCFs are changed rapidly. This feedthrough is the cause of many of the unpleasant artifacts prevalent among noise reduction systems and is often due to parasitic capacitance and mismatches within the IC. This specification is important for the VCA and VCF, but the VCF is the most susceptible because it operates at constant dc gain. The solution incorporated into the SSM2000 to reduce control feedthrough has been to convert from single-ended to full differential at the signal input and convert back again at the output buffer. Because the audio path through the VCA and VCF are fully differential, control feedthrough is determined purely by mismatches with no systematic errors. The slight penalty to be paid for a fully differential system, besides extra complexity, is that the external VCF capacitors (1 nF recommended) now require two pins each instead of one. Careful attention to VCF capacitor matching and layout symmetry reduces control feedthrough to better than 40 dB through the signal path. Downward Expander The Mute function (Pin 17) can override the VCA controls, as shown in Figure 5. When mute is active (HIGH), it pulls the VCA to maximum attenuation. An 85 dB professional quality mute under worse case conditions can be expected over most of the audible frequency range as shown in Figure 13. Mute overrides both the internal VCA control coming from the VCA Detector and the external VCA control port (Pin 7). The external VCA control port is additive in nature to the internal VCA control signals; therefore, noise reduction and volume control may occur simultaneously in the SSM2000. The VCA control port allows the gain of the VCA to changed externally at approximately 22 mV/dB, where 150 mV is equal to 0 dB, as shown in Figure 10a. 30 20 10 0 OUTPUT IN dB AMPLITUDE -10 -20 -30 -40 -50 -60 -70 -70 -60 -50 -40 -30 -20 -10 INPUT IN dB 0 10 20 30 Figure 20. Downward Expander Transfer function VCF Detector After the audio signal passes through the VCF, it is then differentially sent to the VCA. The VCA is characterized by a downward expander transfer function as shown in Figure 20. Attenuation begins at output levels below the internal threshold at an effective rate of 2.2 dB/decade. Therefore, audible noise is attenuated when source material is not present. Conversely, when audio source levels are at or above the threshold, the VCA is set to unity gain. Because of well documented psychoacoustic effects, it is valid to assume that at high audio amplitudes audio noise is being masked by the audio material. The control information that is required for the downward expander to function properly is provided by the internal VCA detector side-chain circuitry common to both the L and R channels. Both the VCF and VCA detectors are amplitude detectors and identical in every way. The applied input signals to the detectors must be preconditioned for the detector circuitry to give the information that is required by the L and R VCFs and VCAs. Figure 21 is a schematic of the recommended preconditioning filters for use with the VCA and VCF detectors. VCF DETECTOR INPUT 2.2nF 8 22nF 22nF 1k 510 SSM2000 SUM OUT 9 ADDITIONAL FIXED-FREQUENCY FILTER IF NECESSARY 2.2F VCA Previous downward expander circuits have often used a higher DETECTOR 10 ratio than 2.2 dB/decade to provide impressive noise reduction INPUT figures (50 dB or more) in the absence of signal information. However, in most applications, such severe noise gating is toFigure 21. The Recommended Signal Preconditioning tally pointless, and actually exacerbates the effects of VCA conFilters for the VCA and VCF Detectors trol feedthrough, since control feedthrough is not attenuated as The VCF detector is fed by 3 x (L+R)/2 averaged input signal the VCA shuts down. For this reason, the maximum VCA atprocessed by a three-pole high pass filter with a -15 dB point at tenuation in the SSM2000 is limited to 15 dB. In addition, the 10 kHz. In Figure 22, the response of the VCF's 3-pole filter is "smooth" roll-off of the downward expander eliminates any shown. The VCF preconditioning filter performs two funcaudible artifacts resulting from this type of noise reduction. In tions: first, it eliminates the large amplitude, low frequency aucontrast, many noise reduction system employ gates and comdio which would otherwise mask the high frequency signals; parators that result in dramatic shifts in attenuation. In these and second, it becomes increasingly sensitive throughout most types of noise reduction systems noise can be heard switching in of the VCF's frequency range of interest (660 Hz to 20 kHz), and out during periods of low volume audio near that attenuacompensating for the effect of most audio signals which typition discontinuity. Telephone systems with noise gating are parcally decrease in amplitude as frequency increases. ticularly prone to this type of behavior, because the noise level of -8- REV. 0 SSM2000 0 -10 VCA -20 -30 -40 -50 VCF -60 -70 1 10 100 1k FREQUENCY - Hz 10k 100k DET CAP, the emitter diode of Q3 is turned off. During this condition, an internal 1.1 A current source sets the release time by discharging the VCF DET CAP. The release time constant of the VCF Detector is about 10 times faster than the release time constant of the Auto Threshold Detector. This ratio should not be decreased, because the output of the VCF Detector is negative peak detected to arrive at the Auto Threshold level (for more information please refer to the Auto Noise Threshold section). The output of the VCF detector (Pin 11) is multiplied by 13 (A1 of Figure 24) and then the noise threshold is subtracted (A2 of Figure 24) to arrive at the actual control voltage for the L and R VCFs. D1 and Q4 are used to set limits on the output of the detector circuitry to ensure that the detector remains responsive to pulsed high frequency audio. VCA Detector OUTPUT IN dB Figure 22. Preconditioning Filter Responses for the VCF DET IN and VCA DET IN Once the preconditioned VCF signal is applied to the input of the VCF detector, the log(x) of the signal is taken. This log function essentially removes the effect of large signal peaks and smoothes out the response of the VCF control. After the log(x) is taken the signal is rectified. The signal is rectified for the following two reasons: first, the following peak detector's output will have less ripple, because twice as many peaks are available; and second, the most negative potential of the rectified signal is equivalent to the noise floor. Following the rectifier, the signal is positive peak detected, as shown in Figure 23. RECTIFIED AND LOG (x) OF VCA DET IN AND VCF DET IN Q3 D1 TO AUTO THRESHOLD AND DIFFERENCE AMPLIFIER 1.1A 11 The VCF and VCA detectors are identical, therefore refer to the previous section on the VCF detector for detailed information of the internal operation of these detectors. The VCA detector is used to detect the audio-band signal amplitude (20 Hz-20 kHz). Usually the lower frequencies 50 Hz to 2 kHz contain the highest audio amplitudes. Therefore, the VCA's preconditioning filter must allow low frequency signals to be presented to the VCA detector. A single pole filter is used to accomplished this function. This filter is formed by a 2.2 F capacitor and the 6 k of internal input impedance of Pin 10 as shown in the schematic, Figure 21. Adaptive Noise Threshold Q4 VCF DET CAP 1F VEE VEE Figure 23. VCA DET IN and VCF DET IN's Positive Peak Detector Circuitry The attack time of the VCF control is set internally and cannot be decreased, however the release time constant is directly proportional to the value of the capacitor VCF DET CAP. Signals above the potential at Pin 11 cause the emitter diode of Q3 to turn on, thus rapidly charging the VCF DET CAP (Pin 11). When the audio signal has dropped below the potential at VCF The threshold level chosen for both the variable filter and downward expander is of prime importance in differentiating between signal and noise. In an automotive environment for example, the audio sources are generally AM and FM radio, tape, and CD. Setting the noise threshold at a value suitable to improve a noisy FM station could easily wipe out most of the dynamic range of a CD. FM station threshold setting is compounded by the vast variations in signal strength in any given location, and the fact that many FM receivers will revert to monaural operation with a greatly improved signal to noise ratio when signal strengths become weak. It is also unreasonable to expect the driver of an automobile to fiddle with panel controls in order to improve the threshold tracking. The patented adaptive noise threshold in the SSM2000 solves these problems, maintaining the transparency of the noise reduction system under most operation conditions while not introducing cumbersome end-user controls. 20k 20k SSM2000 VCF LOG (RECTIFIER) AND FAST PEAK DETECTOR 600 7.2k VEE A1 VCC A2 35nA 20k 20k 20k 20k VCF CONTROL 8 20k 20k VEE VCA LOG (RECTIFIER) AND FAST PEAK DETECTOR + OP AMP 20k A3 20k TO VCA CONTROL VEE 11 13 14 15 10 12 VCF TC 1F DEF ATCAP 0.22F VCA TC 3.3F Figure 24. Schematic Diagram of the Adaptive Noise Threshold and Related Circuitry REV. 0 -9- SSM2000 Noise is most objectionable at high frequencies (3 kHz-8 kHz). Therefore, only the VCF detector output signal is used to determine the adaptive noise threshold. Figures 25a-c, are a series of circuits which illustrate how the noise threshold is derived. It is important to remember that the signal that is applied to the noise threshold detector circuitry has already been rectified and averaged. Hence, the lowest potential over a set period of time corresponds to the noise floor. Node A corresponds to the output of the VCF Detector, and Node B is proportional to the adaptive noise threshold. Figure 25a illustrates the condition where the potential at Node A is above the maximum possible potential for Node B. The maximum noise threshold is set by the potential placed on Pin 14. If the potential at Node B rises to a diode drop above Pin 14, then Q1's emitter-base diode turns on and clamps Node B. This is represented by the current flow I2. However, if Node B has not yet risen to the maximum noise threshold level, then both Q1 and Q2 are OFF and the 35 nA current source is charging C1 (A.T. CAP). The auto threshold capacitor should be a ceramic or equivalent low leakage capacitor, because the charging current could otherwise be of similar amplitude to the capacitor leakage current. VOLTS A MAX B I1 IS CHARGING C1 Q1 CLAMPS NODE B TO A DIODE DROP ABOVE THE POTENTIAL AT PIN 14 MIN MINIMUM NOISE THRESHOLD LEVEL 35nA B Figure 25c illustrates the condition where the potential at Node A is below the minimum potential for Node B. In this case the internal minimum noise potential causes a diode to turn ON. This clamps the Node A potential to the minimum noise threshold level. I1 represents the current flow in this condition. In addition, the 35 nA flows through Q2's emitter-base diode as shown by I2. VOLTS I 1 AND I 2 FLOW CLAMPING NODE B TO THE MINIMUM NOISE THRESHOLD LEVEL MAX MINIMUM NOISE THRESHOLD LEVEL I1 20k B C A I2 20k Q2 Q1 MIN A C 15 35nA B VEE t MAXIMUM NOISE THRESHOLD 14 LEVEL C1 0.22F Figure 25c. Condition Where the Noise Level Is Below the Minimum Noise Threshold Level Setting I1 I2 Simply subtracting the noise threshold from the average VCF HF control signal plus noise threshold and the average VCA control signal plus noise threshold will yield the final VCF and VCA control signal. This operation is accomplished with two internal difference amplifiers. Figures 26a-b shows the response of the detector that controls the VCF bandwidth and VCA gain respectively. Both L IN and R IN pins receive a 10 kHz tone burst. The lower trace of Figure 26a shows the control voltage to the VCF (Pin 11) and the lower trace of Figure 26b shows the control voltage to the VCA (Pin 12). Note the quick rise and slow fall times. This allows fast adaptation to changed input signal conditions, while avoiding pumping effects and other sonic artifacts. 20k 20k Q2 Q1 A VEE MAXIMUM NOISE THRESHOLD 14 LEVEL 15 C1 0.22F t Figure 25a. Condition Where the Actual Noise Threshold Is Above the Maximum Noise Threshold Level Setting (Pin 14) Figure 25b illustrates the condition where the potential at Node A is between the maximum and minimum potentials for Node B. When Node A falls below Node B, then the emitter-base diode of Q2 turns ON causing Node B to follow Node A. Current I2 illustrates how the discharge current from C1 and the 35 nA current source are directed through Q2. Q2 shuts OFF the moment that Node A rises above Node B. This forces the 35 nA current source to begin charging C1 at a constant rate set by the value of C1 at Pin 15. VOLTS I2 FLOWS DISCHARGING C1 CAUSING NODE B TO VOLTAGE FOLLOW NODE A MAX 20k 20k Q2 Q1 I1 IS CHARGING C1 A MIN VEE MAXIMUM NOISE THRESHOLD 14 LEVEL 15 100 90 10 0% 100mV 500mV 100ms MINIMUM NOISE THRESHOLD LEVEL 35nA B Figure 26a. VCF Control Voltage for a Tone Burst I2 I1 100 90 A B C1 0.22F 10 0% t Figure 25b. Condition Where the Noise Level Is Between the Maximum and Minimum Threshold Settings 100mV 500mV 100ms Figure 26b. VCA Control Voltage for a Tone Burst -10- REV. 0 SSM2000 Signal Chain Position Figure 27 illustrates a common audio signal path. The SSM2000 is normally placed just before the volume control and power amplifier. The reason the SSM2000 should be placed as far down the signal chain as possible is that noise that is introduced by the previous components will be operated on by the noise reduction system. On the other hand, the SSM2000 works best at the 300 mV rms level and therefore the IC should be placed before any significant volume control. To operate the SSM2000 in a post-volume control position, please refer to the Optional Application Circuits-Using the SSM2000 in a post-volume control application section. TUNER SSM2000 VCC 1F 0.1F 5 V+ ACOM V- 20 GND* 19 0.1F 1F VEE 6 Figure 28a. The Dual Power Rail Connections for the SSM2000 Powering the SSM2000--Single Supply CASSETTE PREAMP, MIXING, EQ... R SSM2000 L VOLUME + POWER AMP CD LP MIC... Figure 27. The Position of the SSM2000 in a Common Audio Signal Chain Powering the SSM2000--Dual Supply To operate HUSH in the single supply mode, a false ground potential should be generated and attached to ACOM, as shown in Figure 28b. This false ground is usually set at the midpoint between the power rails. The false ground that will be created for the HUSH chip must be able to both sink and source current up to 10 mA. A low cost OP292 will perform this function very well. Once again, a 1 F aluminum electrolytic capacitor and a 0.1 F ceramic capacitor should be use to decouple both the false ground potential and positive supply from the negative supply. It may be helpful to read the preceding section "Powering HUSH--Dual Supply" for information on power supply issues that affect both dual and single supply applications. Mute (Pin 17) and Bypass (Pin 16) potentials should be referenced to DGND. Other inputs are either capacity coupled or referenced to ACOM. SSM2000 VCC 10k 1F 2 1 3 1F 10k 5 6 The SSM2000 has a supply voltage range from +7 V to +18 V. A well regulated (< 20 mV of ripple) 9 V supply is acceptable, even though this is value is exactly equal to the absolute maximum rating of 18 V. The HUSH has been found to be operational down to a supply range of only 4.5 V. However, for guaranteed performance the supplies must be within the specified supply ranges. The power supplies of choice from a performance perspective are quality linear supplies or batteries, because of their low noise. If a switching supply is required, or the power supply lines are noisy, then adequate filtering of the power supply lines should be provided. It is common to place a power supply filter at the power supply and another at the entry point of power to the audio system, but only one filter is necessary if these two locations are separated by less than a foot or so. In normal dual supply applications ACOM should be connected to the system ground. If ACOM is not attached to the system ground, then please refer to the section on Powering HUSH-- single supply. V+ and V- should be connected to their respective power rails. V+ and V- will need to be properly decoupled, because the IR drops in the power rails during large signal swings in the HUSH chip can cause additional distortion. As shown in Figure 28a, it is recommended that a aluminum electrolytic bulk storage cap of 1 F be placed in the vicinity of the HUSH chip and a 0.1 F ceramic capacitor be placed directly at the HUSH chip. Both power rails should be decoupled this manner. V+ ACOM V- 20 GND 19 MUTE BYPASS 1/2 OP292 Figure 28b. Powering the SSM2000--Single Supply DETAILED PIN DESCRIPTIONS Left Audio IN and Right Audio IN (Pins 1, 2) L IN and R IN are the inputs for the Left and Right channels. These are each single ended inputs with an input impedance of 10 k. The input driver for SSM2000 should be able to handle this load level without significant harmonic distortion. These inputs should be ac coupled with at least the minimum capacitor value shown in the example circuits, which is 6.8 F. This sets up a high pass filter with a corner frequency of less than 3 Hz. Choosing a lower value capacitor than 6.8 F may result in the loss of low frequency audio signal. On the other hand choosing a large value will improve low frequency channel separation. REV. 0 -11- SSM2000 VCF Capacitor Port (L Pins 3, 4--R Pins 21, 22) VCA Level Control Input Port (Pin 10) The capacitors connected to these ports control the low-pass filter range of the VCF. Choosing the recommended value, 1 nF, results in a VCF range where the lowest possible low-pass corner frequency is about 660 Hz. There is a 20 dB/decade decrease in amplitude after the corner frequency regardless of the capacitor value at this port. Raising the value to 10 nF lowers the VCF range by a decade to 66 Hz. This will permanently attenuate the high frequency audio signals regardless of the audio frequency spectrum applied to the input of HUSH. In other words, for VCF capacitor values higher than 1 nF, high frequency loss will occur. On the other hand, lowering the value of the VCF capacitor to 0.1 nF will result in a VCF range of 10 kHz to 100 kHz. Of course, most of this range is outside of audible frequencies and is therefore not very useful. V+ (Pin 5) This pin is the input to the VCA detector, which outputs a voltage that will control the VCA in the audio signal path. A singlepole filter is recommended between the SUM OUT (Pin 9) and VCA IN (Pin 10). VCF Time Constant (Pin 11) A 1 F capacitor is normally connected to this point. The purpose of which is to control the rate of change of the VCF's low pass corner frequency. Raising the value of this capacitor from the recommended 1 F will tend to increase "breathing" (noise and high frequency signal that slowly fades in and out). Lowering the value of this capacitor from the recommended 1 F will tend to increase "pumping" (noise that noticeably cuts in and out). The final capacitor values are, of course, subjective and should be chosen based on one's own evaluation within the system environment. VCA Time Constant (Pin 12) V+ is the supply pin. Please refer to the sections "Powering the SSM2000--Dual Supply" and "Powering the SSM2000-- Single Supply" for more information. ACOM (Pin 6) ACOM is the internal ground reference for the audio circuitry. In single supply applications this pin is normally connected to a low impedance potential that is one half the positive supply. "Powering the SSM2000--Dual Supply" and "Powering the SSM2000--Single Supply" for more information. VCA Control Port (Pin 7) The function of this pin is similar to the VCF TC pin. Once again, a 3.3 F capacitor is placed on this pin, which controls the rate of change of the VCA's amplitude. Higher value capacitors will result in "breathing" and lower values will cause "pumping." DO NOT CONNECT/Adaptive Noise Threshold Override (Pin 13) As shown in Figure 5, this port connects into the VCA circuitry. The purpose of this port is to allow external control of the VCA amplitude setting or volume in addition to the normal internal VCA detector control. Every 22 mV applied to the port results in a 1 dB decrease in amplitude, with 0 dB corresponding to about 150 mV. Figure 10 illustrates the transfer characteristic of the VCA control port. VCF Level Control Input Port (Pin 8) Connecting this pin to any low-impedance potential including ground will disable the adaptive noise threshold. Pin 13's internal connection is shown in Figure 24. It is possible to set up an artificial noise threshold using this pin. Default Noise Threshold (Pin 14) This pin is the input to the VCF control and noise threshold detection circuitry. A three-pole filter is recommended between the SUM OUT (Pin 9) and VCF IN (Pin 8). SUM OUT (Pin 9) This pin sets up the maximum noise threshold possible. As can be seen in Figure 25a-c, the emitter voltage of Q1 can be no higher than a diode drop above the potential set at Pin 14. A maximum noise threshold level prevents HUSH from attempting to filter audio signal in the presence of very high noise. The minimum noise threshold detection level has already been set internally by Q3. By setting the lower limit, HUSH is prevented from attenuating already very low levels of noise which would reduce the responsiveness of HUSH. Auto Threshold CAP (Pin 15) SUM OUT is the buffered combination of the Left and Right inputs. Figure 29 shows the internal configuration of SUM OUT. It is this combined Left and Right audio signal that will be used by the VCF and VCA detectors to determine audio amplitude and frequency content. SSM2000 L IN 1 TO L VCF 20k 18 The recommended 0.22 F capacitor connected to this pin controls the rate of change of the adaptive noise threshold level. As shown in Figures 25a-c, if Q1 and Q2 are turned OFF, then a 35 nA constant current source begins to charge the Auto Threshold CAP. This rising potential across the AT CAP represents the detected noise level. Because the capacitor's charging current only 35 nA, a low-leakage ceramic or equivalent capacitor is required to hold the charge. DEFEAT (Pin 16) 20k R IN 2 470pF TO R VCF X3 SUM OUT 9 3*(L+R)/2 Figure 29. The Internal Circuit for SUM OUT Applying a +5 V volt signal to this pin will defeat the HUSH noise reduction system. Potentials applied to this pin should be reference to GND. Standard TTL levels are recommended when using DEFEAT. Activating DEFEAT causes the VCF detector and VCA detector to send their maximum control signal levels to the VCF and VCA respectively. If DEFEAT is being controlled by noisy digital logic lines, then use adequate filtering to avoid digital noise interfering with audio signals present in the SSM2000. -12- REV. 0 SSM2000 Mute (Pin 17) Applying a +5 V voltage signal to this pin will mute the SSM2000. Potentials applied to this pin should be reference to GND. Standard TTL levels are recommended when using MUTE. Activating MUTE causes maximum attenuation in the VCA on both the L and R channels. If MUTE is being controlled by noisy digital logic lines, then use adequate filtering to avoid digital noise interfering with audio signals present in the SSM2000. DO NOT CONNECT/Detector Bandwidth Limit (Pin 18) * Compression--reduces dynamic range to improve audibility of softer sounds in noisy environments such as when playing CDs in cars. * Volume vs. Road Speed--a speedometer signal can be used to increase audio levels as road speed increases. * Compression vs. Road Speed--see above. Automatic Leveling Pin 18 is connected internally to both the left and right channels though 20 k resistors, as shown in Figure 29. This node allows access to the input of the x3 gain buffer, whose output is SUM OUT (Pin 9). Do not connect this node to ground or another potential. Connecting a 470 pF capacitor to this pin will limit the range of the input signal that runs through the VCF and VCA detectors to 35 kHz. This is an inexpensive way to remove fixed-frequency tones of frequencies of greater than 35 kHz. Please refer the Precautions, Limitations, and Observations section under Pre-filtering Out-of-Audio Band Noise for more information. GND (Pin 19) Automatic leveling is similar to compression, however instead of attempting to attenuate all large amplitude audio signals, the leveling circuit attenuates audio signal when the overall signal amplitude has increased for a considerable amount of time, >1 sec. Attenuating loud TV commercials whose average volume levels are significantly above the program material is an application were this feature is particularly useful. The circuit shown in Figure 30 is an example of HUSH automatic leveling. Automatic leveling is a signal amplitude operation, therefore, the input of the circuit is derived from Pin 12, VCA TC. This input voltage is gained and level shifted. The volume control is "counter intuitive," because the higher the gain the lower the resulting volume, as the VCA control port attenuates when positive voltage is applied. The level threshold potentiometer sets the volume level, which is a gain range of 1.5 dB to -80 dB. After the op amp, the signal passes through a positive peak detector, and is averaged by a long time constant RC. The signal is now representative of the long term average amplitude of the original audio that is applied to the L/R input of the SSM2000. This signal is then passed to the VCA control port. Audio Compression GND is normally connected to V- in single-supply applications and ACOM in dual-supply applications. Please read the MUTE and DEFEAT sections for more information. V- (Pin 20) This pin should be connected to the most negative power potential, typically chassis ground in single-supply applications and the negative power rail in dual-supply applications. R OUT and L OUT (Pin 23 and Pin 24) The outputs of HUSH are buffered and can therefore drive moderate loads. For loads (e.g., speakers or headphones) which exceed 2 k or 300 pF additional buffering may be necessary. OPTIONAL APPLICATION CIRCUITS The SSM2000 has been designed to facilitate numerous additional functions by adding a modest amount of external circuitry. The direct access to the VCA gain control port plus access to the amplitude and frequency detectors allow construction of valuable systems features such as: * DC Volume Control--one dc level controls both audio channels without routing audio signals to the front panel. * Automatic Volume Leveling--constrains long term average volume levels to within a few dB, avoiding constant resetting of volume as one tuner through various stations or switches from radio to tape, etc. +5V 5 7 6 Audio compression can be thought of as fast automatic leveling. The purpose of audio compression is to attenuate larger amplitude signals, such that signals above the adaptive threshold will be scaled by 0.2 dB/dB, while the signals below the externally set level threshold will remain essentially unchanged. A shorter time constant and slight gain change are the only differences between the audio compression circuit and the automatic leveling circuit. Figure 30 illustrates the audio compression circuit. The OP275 is an excellent audio amplifier and is suggested for this application when the power supply rails are greater than or equal to 5 V. For applications where the power supplies are separated by less than 10 V, then a single-supply op amp like the OP279 is recommended. OFF LEVEL COMPRESS 100k SSM2000 19 12 20 SP3T OFF LEVEL 3.3F 10M 0.1F 20k 7.5k COMPRESS 1/2 OP275 3k 27k -5V -5V +5V 10k POT LEVEL THRESHOLD +5V 5k POT VOLUME CONTROL Figure 30. An Automatic Leveling/Compression Circuit REV. 0 -13- SSM2000 Volume Control vs. Road Speed Excess External Noise In automotive applications the external volume control port (Pin 7) can also be used to increase the volume as road noise increases. The digital speedometer information available in most vehicles may be filtered or D/A converted and sent to the SSM2000 to control the volume. Figure 31 illustrates a SSM2000 audio volume control system using digital speedometer data. SPEEDOMETER DATA MPH VREFH +5V 5 Even though the SSM2000 is a noise reduction IC, excessively high noise floor amplitudes may result in no noise reduction. Common causes of this condition are high audio volume or very poor reception. In this case the VCF and the VCA will remain at their maximum settings and the audio signal will simply pass through HUSH. Prefiltering Out-of-Audio Band Tones D/A CONVERTER VCA 7 CONTROL PORT VREFL SSM2000 6 19 20 -5V Figure 31. Volume Control vs. Road Speed Block Diagram Using the SSM2000 in a Post-Volume Control Application The SSM2000 is capable of operation for limited reasonable changes in input levels, because of the internal adaptive noise threshold detector. However, this detector is limited in range for optimal performance and therefore external circuitry is required if the prevolume control line level is to be changed by more than -20 dB. Figure 32 illustrates a circuit that alters the range of the noise threshold. +5V 5 14 The most common cause of SSM2000 operational error is due to large amplitude fixed-frequency tones. When SSM2000 detector's are exposed to large fixed-frequency tones, they recognizes these types of tones as signal and automatically adjust the VCA and VCF to allow this frequency to pass through to the output along with the underlying noise. The following is a list of typical fixed-frequency tones which might cause this type of HUSH behavior: switching power supply noise (20 kHz- 200 kHz), sampling frequency in digitized music (44 kHz), PC monitor scan frequency (30 kHz), microprocessor operating frequency or submultiple, and FM stereo subcarrier frequency (19 kHz). Figure 33 is a scope photo of a spectrum analyzer output showing the 19 kHz FM stereo subcarrier frequency. So, if your application will encounter any of these types of noise sources, then a preconditioning filter for the internal detector is necessary. 120k 1.3k 18k 1.2k 20k 10k +5V SSM2000 6 19 10dB/DIV 12 20 3.3F 1/2 OP292 750k 22F -5V 1/2 OP292 20Hz 20kHz -5V Figure 33. Scope Photo of the 19 kHz FM Subcarrier Frequency Figure 32. A Schematic to Use the SSM2000 in a PostVolume Control Application PRECAUTIONS, LIMITATIONS, AND OBSERVATIONS Operating Temperature Under light loading conditions HUSH runs relatively cool, typically only 4C over ambient room temperature. If HUSH is hot, then check to make sure that the capacitive and resistive loading requirements are not being exceeded. Operating Signal Amplitude and Frequency Range The output signal range of the SSM2000 is set internally (3.67 V p-p @ 0.1 THD) and will not increase even if the power supply rails are raised. The frequency range is set by the input ac coupling capacitors (< 10 Hz) and by the VCF which is internally limited to 35 kHz. Figures 34 and 35 illustrate two different approaches to filtering these undesired tones. If the interfering tone frequency is less than 20 kHz (e.g., TV-15.625 kHz), a Twin-T Hi-Q notch filter as shown in Figure 34 should be used. The notch filter will attenuate the tone while having a minimal effect on the audio band signal. For interfering tones above 20 kHz, a general 3-pole low-pass filter is recommended, such as the Butterworth filter shown in Figure 35. The input of the filter should always be connected to the output, SUM OUT. Filtering the SUM OUT signal will not affect audio quality because the SUM OUT signal is not part of the main audio path. However, overaggressive filtering may remove noise and signal that is necessary to set the VCA control, VCF control, and noise threshold levels. The OP292 or OP275 shown in the example filter circuit are high performance, low cost amplifiers suitable for this application. -14- REV. 0 SSM2000 VCF DETECTOR 8 INPUT R = 100k C = 100pF R SUM OUT 9 2.2nF 22nF 1k 22nF 510 EVALUATING THE SSM2000 2 1 3 C R 2C 1/2 OP275 1k fc = 10k 1 2RC SSM2000 VCA DETECTOR 10 INPUT C R 2 2.2F Because the SSM2000 is a dynamic system, it will respond to test signals that are applied during the evaluation procedures. Without thorough understanding of how the HUSH noise reduction system works, attempts to apply standard test procedures can produce misleading results. The following paragraphs will present evaluation techniques to avoid many of these common evaluation pitfalls. The SSM2000 has been designed as a stereo system. As a result the input to the detector circuits is the sum of both the left and right inputs. Monaural signals that used during evaluation should be applied to both inputs, because the detectors' input requires both the left and right signal. Otherwise, the control levels from the VCA detector, VCF detector, and the Adaptive Noise Threshold will be about one-half the necessary amplitude for proper evaluation results. The SSM2000 has exceptional tolerance for varying line levels with noise reducing capability between line levels of 100 mV rms to 1 V rms. However, the IC has been optimized for an industry standard input line level of 300 mV rms. Audio line levels not equal to 300 mV rms will exhibit slightly degraded specifications (e.g., distortion). If the line levels are outside of the 100 mV rms to 1 V rms range, then audio output may appear to have no noise reduction or to be muted. The circuit shown in Figure 32 will enable noise reduction for very low line level audio. THD measurement instruments typically sweep a sine wave across the audio spectrum. The normal SSM2000 response to this type of signal results in the automatic noise threshold detector slowly rising to its maximum value; the VCA detector at maximum value, the VCF detector will initially be closed at low frequencies and will gradually open as the frequency is increased. When the testing signal reaches about 500 Hz, it will begin to be attenuated by the "closed" VCF. This attenuation will reach at maximum of about 1 dB at 660 Hz for recommended line levels, at which point the VCF detector begins to open up. However, all these conditions and characteristics are not representative of a normal music signal. To properly measure THD, it is necessary to activate HUSH Defeat. THD measurements in DEFEAT mode give accurate results because it does not remove the VCA and VCF from the signal chain, rather it sets there values to unity (VCA) and maximum frequency (VCF). The bandwidth of the SSM2000, as stated before, is limited to 35 kHz because of the VCF capacitors. This bandwidth limitation must be taken into account when the slew rate of the SSM2000 is measured. The simplest method to measure slew rate is to remove the VCF capacitors from the circuit, and take the slew rate measurement while in DEFEAT mode. If the VCF capacitors are not removed, then the audio signal is limited by the VCF's bandwidth and not by the overall slew rate. One standard of the SSM2000's audio quality is the characteristics of the attack and release times of the VCA and VCF. The attack and release times of the VCA and VCF are controlled by 1.1 A current sources which charge the VCA Time Constant capacitor and VCF Time Constant capacitor. The attack and released times are typically measured from the VCA's and VCF's fully closed position (e.g., VCA--maximum attenuation, Figure 34. A Hi-Q Twin-T Notch Filter with the Center Frequency at 15.625 kHz VCF DETECTOR 8 INPUT 2.2nF 22nF 1k 0.056F 2 1k SUM OUT 9 22nF 510 SSM2000 1/2 OP275 1 1k 1k 3 C 0.033F 2.2F 0.022F VCA DETECTOR 10 INPUT Figure 35. A Three-Pole Butterworth Filter Channel-to-Channel Separation The SSM2000 has a 60 dB of channel separation specification at 1 kHz. As shown in Figure 36, the Left input to the R VCF node and the Right input to the L VCF node are tied together with 50 k of resistance, and tied to ground with 2.5 k. If the Left and Right ac coupling capacitors were not present this would results in a dc separation of around -27 dB. However, the inputs of the SSM2000 are usually ac coupled; therefore the low frequency channel separation is a function of the value of the ac coupling capacitors and the theoretical dc separation. The higher the value of the ac coupling capacitor, the better the channel separation will be. For example, using 10 F ac coupling capacitors will result in 32 dB of channel separation at around 6 Hz-8 Hz, improving at a rate of 20 dB/decade, till it reaches a maximum -60 dB. Therefore, to achieve 60 dB of channel separation at 75 Hz and above, 47 F capacitors are recommended. 2.5k TO L VCF 6.8F L 20k TO DETECTORS 6.8F R 10k TO R VCF 2.5k 20k 470pF 10k Figure 36. The SSM2000's Input Impedance Network REV. 0 -15- SSM2000 VCF--minimum bandwidth) to their fully open positions. If the VCA and VCF are fully opened then at least 1.5 seconds of no signal is required for the VCA and VCF detectors to return to the closed position. A series of 10 kHz pulses interspersed with 1.5 seconds of no signal will accomplish this requirement. The system's attack and release times will be dominated by VCA since the attack and release times of the VCA are slower than the VCF's. In addition, monitoring Pins 11 (VCF) and 12 (VCA) will ensure that the VCA and VCF are either fully opened or fully closed. As stated before, the most common cause of a failure to reduce noise is fixed-frequency tones that are being detected by the output. A solution for this condition is detailed in the "Precautions, Limitations, and Observations" section. In single-supply applications, ACOM should be driven with a well regulated reference capable of both sinking and souring current. Zener-resistor references are unacceptable, because of their inability to sink significant amounts of current. By following these simple suggestions, the common pitfalls to a successful implementation of the HUSH noise reduction system can be avoided. OUTLINE DIMENSIONS Dimensions shown in inches and (mm). 24-Pin Plastic DIP (N-24) 1.275 (32.30) 1.125 (28.60) 24 1 13 12 PIN 1 0.210 (5.33) MAX 0.200 (5.05) 0.125 (3.18) 0.022 (0.558) 0.014 (0.356) 0.100 (2.54) BSC 0.070 (1.77) 0.045 (1.15) 0.325 (8.25) 0.300 (7.62) 0.195 (4.95) 0.115 (2.93) 0.060 (1.52) 0.015 (0.38) 0.150 (3.81) MIN SEATING PLANE 0.280 (7.11) 0.240 (6.10) 0.015 (0.381) 0.008 (0.204) 24 Pin SOIC (SOL-24) 0.6141 (15.60) 0.5985 (15.20) 24 13 1 12 0.4193 (10.65) 0.3937 (10.00) 0.2992 (7.60) 0.2914 (7.40) PIN 1 0.1043 (2.65) 0.0926 (2.35) 0.0291 (0.74) x 45 0.0098 (0.25) 0.0118 (0.30) 0.0040 (0.10) 0.0500 (1.27) BSC 8 0.0192 (0.49) 0 SEATING 0.0125 (0.32) 0.0138 (0.35) PLANE 0.0091 (0.23) 0.0500 (1.27) 0.0157 (0.40) IMPORTANT NOTICE--LICENSING The SSM2000 is manufactured under a license granted to Analog Devices, Inc., by Rocktron Corp. and embodies technology covered by the following patents: 4696044, 4893099, 5124657, 5263091, 5402498, and 5493617. Foreign patents are pending. Rocktron Corp. grants a royalty-free license to purchasers of the SSM2000 for use in end equipment, and no further rights are implied. The use of the HUSH logo is encouraged on products, packaging, promotional and descriptive literature. Rocktron Corp. reserves the right to approve use of the HUSH logo and may require submission of a schematic illustrating the use of the SSM2000; a physical sample of the end product may also be required. Applications for approval of the use of the HUSH name and logo may be obtained by contacting ADI. -16- REV. 0 PRINTED IN U.S.A. C2130-10-5/96