LTC1562 Very Low Noise, Low Distortion Active RC Quad Universal Filter U FEATURES ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ DESCRIPTION Continuous Time—No Clock Four 2nd Order Filter Sections, 10kHz to 150kHz Center Frequency ±0.5% Typical Center Frequency Accuracy ±0.3% Typical Center Frequency Accuracy (A Grade) Wide Variety of Response Shapes Lowpass, Bandpass and Highpass Responses 103dB Typical S/N, ±5V Supply (Q = 1) 97dB Typical S/N, Single 5V Supply (Q = 1) 96dB Typical S/(N + THD) at ±5V Supply, 20kHz Input Rail-to-Rail Input and Output Voltages DC Accurate to 3mV (Typ) “Zero-Power” Shutdown Mode Single or Dual Supply, 5V to 10V Total Resistor-Programmable fO, Q, Gain U APPLICATIONS ■ ■ ■ ■ ■ ■ High Resolution Systems (14 Bits to 18 Bits) Antialiasing/Reconstruction Filters Data Communications, Equalizers Dual or I-and-Q Channels (Two Matched 4th Order Filters in One Package) Linear Phase Filtering Replacing LC Filter Modules The LTC®1562 is a low noise, low distortion continuous-time filter with rail-to-rail inputs and outputs, optimized for a center frequency (fO) of 10kHz to 150kHz. Unlike most monolithic filters, no clock is needed. Four independent 2nd order filter blocks can be cascaded in any combination, such as one 8th order or two 4th order filters. Each block’s response is programmed with three external resistors for center frequency, Q and gain, using simple design formulas. Each 2nd order block provides lowpass and bandpass outputs. Highpass response is available if an external capacitor replaces one of the resistors. Allpass, notch and elliptic responses can also be realized. The LTC1562 is designed for applications where dynamic range is important. For example, by cascading 2nd order sections in pairs, the user can configure the IC as a dual 4th order Butterworth lowpass filter with 94dB signal-to-noise ratio from a single 5V power supply. Low level signals can exploit the built-in gain capability of the LTC1562. Varying the gain of a section can achieve a dynamic range as high as 118dB with a ±5V supply. Other cutoff frequency ranges can be provided upon request. Please contact LTC Marketing. , LTC and LT are registered trademarks of Linear Technology Corporation. U TYPICAL APPLICATION Amplitude Response Dual 4th Order 100kHz Butterworth Lowpass Filter R21, 10k 3 5 5V 0.1µF 6 R23, 10k VIN1 2 8 9 RQ3, 5.62k 10 INV B INV C V1 B V1 C V2 B V2 C V– V + LTC1562 SHDN AGND V2 A V2 D V1 A V1 D INV A INV D RIN4, 10k –10 20 –20 19 RQ2, 13k 18 R22, 10k VOUT2 –5V 16 15 0.1µF VOUT1 13 12 R24, 10k 11 RQ4, 13k 1562 TA01 SCHEMATIC INCLUDES PIN NUMBERS FOR 20-PIN PACKAGE. PINS 4, 7, 14, 17 (NOT SHOWN) ALSO CONNECT TO V – SEE TYPICAL APPLICATIONS FOR OTHER CUTOFF FREQUENCIES DC ACCURATE, NONINVERTING, UNITY-GAIN, RAIL-TO-RAIL INPUT AND OUTPUTS. PEAK SNR ≈ 100dB WITH ±5V SUPPLIES GAIN (dB) 1 RQ1, 5.62k RIN3 10k 0 RIN2, 10k RIN1 10k VIN2 10 –30 –40 –50 –60 –70 –80 10k 100k FREQUENCY (Hz) 1M 1562 TA03b 1 W U U PACKAGE/ORDER INFORMATION LTC1562 U W W W ABSOLUTE MAXIMUM RATINGS (Note 1) Total Supply Voltage (V + to V –) .............................. 11V Maximum Input Voltage at Any Pin ....................(V – – 0.3V) ≤ V ≤ (V + + 0.3V) Operating Temperature Range LTC1562C................................................ 0°C to 70°C LTC1562I ............................................ – 40°C to 85°C Storage Temperature Range ................. – 65°C to 150°C Lead Temperature (Soldering, 10 sec).................. 300°C ORDER PART NUMBER TOP VIEW INV B V1 B V2 B V –* V+ SHDN V –* V2 A V1 A INV A 1 2 3 4 5 6 7 8 9 10 20 19 18 17 16 15 14 13 12 11 INV C V1 C V2 C V –* V– AGND V –* V2 D V1 D INV D LTC1562CG LTC1562ACG LTC1562IG LTC1562AIG G PACKAGE 20-LEAD PLASTIC SSOP *G PACKAGE PINS 4, 7, 14, 17 ARE SUBSTRATE/SHIELD CONNECTIONS AND MUST BE TIED TO V – TJMAX = 150°C, θJA = 136°C/W Consult factory for Military grade parts. ELECTRICAL CHARACTERISTICS VS = ±5V, outputs unloaded, TA = 25°C, SHDN pin to logic “low”, unless otherwise noted. AC specs are for a single 2nd order section, RIN = R2 = RQ =10k ±0.1%, fO = 100kHz, unless noted. SYMBOL PARAMETER VS Total Supply Voltage IS Supply Current VOS CONDITIONS MIN MAX 10.5 V 17.3 19 19.5 21.5 mA mA 23.5 25.5 mA mA 4.75 VS = ±2.375V, RL = 5k, CL = 30pF, Outputs at 0V VS = ±5V, RL = 5k, CL = 30pF, Outputs at 0V UNITS VS = ±2.375V, RL = 5k, CL = 30pF, Outputs at 0V VS = ±5V, RL = 5k, CL = 30pF, Outputs at 0V ● ● Output Voltage Swing VS = ±2.375V, RL = 5k, CL = 30pF VS = ±5V, RL = 5k, CL = 30pF ● ● DC Offset Magnitude, V2 Outputs (Lowpass Response) VS = ±2.375V, Input at AGND Voltage VS = ±5V, Input at AGND Voltage ● ● DC AGND Reference Point VS = Single 5V Supply 2.5 Center Frequency (f O) Error (Note 2) LTC1562 LTC1562A VS = ±5V, V2 Output Has RL = 5k, CL = 30pF VS = ±5V, V2 Output Has RL = 5k, CL = 30pF 0.5 0.3 1.0 0.6 % % + 0.05 + 0.1 dB + 0.2 + 0.5 dB HL LP Passband Gain (V2 Output) VS = ±2.375V, fIN = 10kHz, V2 Output Has RL = 5k, CL = 30pF ● HB BP Passband Gain (V1 Output) VS = ±2.375V, fIN = fO, V2 Output Has RL = 5k, CL = 30pF ● 2 TYP 4.0 9.3 4.6 9.8 3 3 0 VP-P VP-P 15 15 mV mV V LTC1562 ELECTRICAL CHARACTERISTICS VS = ±5V, outputs unloaded, TA = 25°C, SHDN pin to logic “low”, unless otherwise noted. AC specs are for a single 2nd order section, RIN = R2 = RQ =10k ±0.1%, fO = 100kHz, unless noted. SYMBOL THD PARAMETER CONDITIONS Q Error VS = ±2.375V, LP Output Has RL = 5k, CL = 30pF +3 % Wideband Output Noise, Lowpass Response (V2 Output) VS = ±2.375V, BW = 200kHz, Input AC GND VS = ±5V, BW = 200kHz, Input AC GND 24 24 µVRMS µVRMS Input-Referred Noise, Gain = 100 BW = 200kHz, f O = 100kHz, Q = 1, Input AC GND 4.5 µVRMS Total Harmonic Distortion, Lowpass Response (V2 Output) fIN = 20kHz, 2.8VP-P, V1 and V2 Outputs Have RL = 5k, CL = 30pF – 96 dB fIN = 100kHz, 2.8VP-P, V1 and V2 Outputs Have RL = 5k, CL = 30pF – 78 dB SHDN Pin to V + SHDN Pin to V +, VS = ±2.375V 1.5 1.0 Shutdown Supply Current MIN Shutdown-Input Logic Threshold TYP MAX UNITS µA µA 15 2.5 Shutdown-Input Bias Current SHDN Pin to 0V – 10 V µA – 20 Shutdown Delay SHDN Pin Steps from 0V to V + 20 µs Shutdown Recovery Delay SHDN Pin Steps from V + to 0V 100 µs 5 pA Inverting Input Bias Current, Each Biquad The ● denotes specifications that apply over the full operating temperature range. Note 1: Absolute Maximum Ratings are those values beyond which the life of a device may be impaired. Note 2: fO change from ±5V to ±2.375 supplies is – 0.15% typical, fO temperature coefficient, – 40°C to 85°C, is 25ppm/°C typical. U W TYPICAL PERFOR A CE CHARACTERISTICS fO Error vs Nominal fO (VS = ±5V) 1.50 1.25 1.25 Q = 2.5 0.25 0 – 0.25 – 0.50 – 0.75 Q=1 0 – 0.25 – 0.50 – 0.75 –1.00 –1.25 –1.25 50 60 70 80 90 100 110 120 130 140 150 NOMINAL fO (kHz) 1562 G01 Q = 10 0.25 –1.00 –1.50 25 Q = 2.5 0.50 –1.50 TA = 70°C TA = 25°C RIN = RQ Q=5 0.75 fO ERROR (%) fO ERROR (%) 0.50 30 1.00 Q=5 0.75 35 Q ERROR (%) 1.00 Q Error vs Nominal fO (VS = ±5V) fO Error vs Nominal fO (VS = ±2.5V) 1.50 Q=1 20 15 10 Q=5 Q = 2.5 5 0 Q=1 –5 50 60 70 80 90 100 110 120 130 140 150 NOMINAL fO (kHz) 1562 G02 50 60 70 80 90 100 110 120 130 140 150 NOMINAL fO (kHz) 1562 G03 3 LTC1562 U W TYPICAL PERFOR A CE CHARACTERISTICS Peak BP Gain vs Nominal fO (VS = ±5V) (Figure 3, V1 Output) Q Error vs Nominal fO (VS = ±2.5V) 35 3.0 2.5 RIN = RQ 15 Q = 2.5 10 2.0 PEAK BP GAIN (dB) Q=5 20 1.5 Q=5 1.0 Q = 2.5 0.5 5 Q=1 0 –5 – 0.5 50 60 70 80 90 100 110 120 130 140 150 NOMINAL fO (kHz) 1562 G04 55 50 50 45 45 Q = 2.5 30 Q=1 25 35 Q = 2.5 30 25 20 15 15 10 Q=1 10 70 80 90 100 110 120 130 140 NOMINAL fO (kHz) Q = 2.5 0.5 Q=1 0 – 0.5 50 60 70 80 90 100 110 120 130 140 150 NOMINAL fO (kHz) 1562 G6 Distortion vs External Load Resistance (VS = ±5V, 25°C) (Figure 8) Q=5 40 20 60 1.0 THD (AMPLITUDE BELOW FUNDAMENTAL) (dB) 60 55 BP NOISE (µVRMS) 60 35 Q=5 1.5 BP Noise vs Nominal fO (VS = ±5V, 25°C) (Figure 3, V1 Output) (RIN = RQ) Q=5 Q = 10 RIN = RQ 1562 G5 LP Noise vs Nominal fO (VS = ±5V, 25°C) (Figure 3, V2 Output) (RIN = R2) 40 TA = 70°C TA = 25°C 2.0 Q=1 0 50 60 70 80 90 100 110 120 130 140 150 NOMINAL fO (kHz) 2.5 Q = 10 RIN = RQ Q = 10 25 Q ERROR (%) 3.0 TA = 70°C TA = 25°C PEAK BP GAIN (dB) TA = 70°C TA = 25°C 30 NOISE (µVRMS) Peak BP Gain vs Nominal fO (VS = ±2.5V) (Figure 3, V1 Output) 60 70 80 90 100 110 120 130 140 NOMINAL fO (kHz) 1562 G07 1562 G08 0 –10 –20 –30 2nd ORDER LOWPASS fO = 100kHz Q = 0.7 OUTPUT LEVEL 1VRMS (2.83VP-P) ± 5V SUPPLIES – 40 – 50 – 60 –70 – 80 fIN = 50kHz – 90 fIN = 20kHz –100 10k 2k 5k EXTERNAL LOAD RESISTANCE (Ω) 1k 1562 G09 U U U PIN FUNCTIONS Power Supply Pins: The V + and V – pins should be bypassed with 0.1µF capacitors to an adequate analog ground or ground plane. These capacitors should be connected as closely as possible to the supply pins. In the 20-lead SSOP package, the additional pins 4, 7, 14 and 17 are internally connected to V – (Pin 16) and should also be tied to the same point as Pin 16 for best shielding. Low noise linear supplies are recommended. Switching supplies are not recommended as they will lower the filter dynamic range. 4 Analog Ground (AGND): The AGND pin is the midpoint of an internal resistive voltage divider, developing a potential halfway between the V + and V – pins, with an equivalent series resistance nominally 7kΩ. This serves as an internal ground reference. Filter performance will reflect the quality of the analog signal ground and an analog ground plane surrounding the package is recommended. The analog ground plane should be connected to any digital ground at a single point. For dual supply operation, the AGND pin should be connected to the ground plane LTC1562 U U U PIN FUNCTIONS (Figure 1). For single supply operation, the AGND pin should be bypassed to the ground plane with at least a 0.1µF capacitor (at least 1µF for best AC performance) (Figure 2). ANALOG GROUND PLANE 1 20 2 19 3 18 4 17 V+ 5 0.1µF 6 15 7 14 8 13 9 12 10 11 LTC1562 SINGLE-POINT SYSTEM GROUND V– 0.1µF Shutdown (SHDN): When the SHDN input goes high or is open-circuited, the LTC1562 enters a “zero-power” shutdown state and only junction leakage currents flow. The AGND pin and the amplifier outputs (see Figure 3) assume a high impedance state and the amplifiers effectively disappear from the circuit. (If an input signal is applied to a complete filter circuit while the LTC1562 is in shutdown, some signal will normally flow to the output through passive components around the inactive op amps.) A small pull-up current source at the SHDN input defaults the LTC1562 to the shutdown state if the SHDN pin is left floating. Therefore, the user must connect the SHDN pin to a logic “low” (0V for ±5V supplies, V – for 5V total supply) for normal operation of the LTC1562. (This convention permits true “zero-power” shutdown since not even the driving logic must deliver current while the part is in shutdown.) With a single supply voltage, use V – for logic “low”— do not connect SHDN to the AGND pin. 16 DIGITAL GROUND PLANE (IF ANY) 1/4 LTC1562 *R1 AND C ARE PRECISION INTERNAL COMPONENTS 1562 F01 1 sR1C* Figure 1. Dual Supply Ground Plane Connection (Including Substrate Pins 4, 7, 14, 17) C – ANALOG GROUND PLANE 1 20 2 19 3 18 4 17 V+ 5 0.1µF 6 15 7 14 8 13 9 12 10 11 SINGLE-POINT SYSTEM GROUND LTC1562 + V2 16 INV R2 V1 RQ 1562 F01 ZIN 1µF + – V +/2 REFERENCE VIN RESPONSE RESPONSE ZIN TYPE AT V1 AT V2 R BANDPASS LOWPASS C HIGHPASS BANDPASS DIGITAL GROUND PLANE (IF ANY) 1562 F01 Figure 2. Single Supply Ground Plane Connection (Including Substrate Pins 4, 7, 14, 17) IN EACH CASE, fO = (100kHz) ( ( ) 10kΩ R2 ) Q = RQ 100kHz R2 fO Figure 3. Equivalent Circuit of a Single 2nd Order Section (Inside Dashed Line) Shown in Typical Connection. Form of ZIN Determines Response Types at the Two Outputs (See Table) 5 LTC1562 U U U PIN FUNCTIONS INV A, INV B, INV C, INV D: Each of the INV pins is a virtualground summing point for the corresponding 2nd order section. For each section, external components ZIN, R2, RQ connect to the INV pin as shown in Figure 3 and described further in the Applications Information. Note that the INV pins are sensitive internal nodes of the filter and will readily receive any unintended signals that are capacitively coupled into them. Capacitance to the INV nodes will also affect the frequency response of the filter sections. For these reasons, printed circuit connections to the INV pins must be kept as short as possible, less than one inch (2.5cm) total and surrounded by a ground plane. V1 A, V1 B, V1 C, V1 D: Output Pins. Provide a bandpass, highpass or other response depending on external circuitry (see Applications Information section). Each V1 pin also connects to the RQ resistor of the corresponding 2nd order filter section (see Figure 3 and Applications Information). Each output is designed to drive a nominal net load of 5kΩ and 30pF, which includes the loading due to the external RQ. Distortion performance improves when the outputs are loaded as lightly as possible. Some earlier literature refers to these outputs as “BP” rather than V1. V2 A, V2 B, V2 C, V2 D: Output Pins. Provide a lowpass, bandpass or other response depending on external circuitry (see Applications Information section). Each V2 pin also connects to the R2 resistor of the corresponding 2nd order filter section (see Figure 3 and Applications Information). Each output is designed to drive a nominal net load of 5kΩ and 30pF, which includes the loading due to the external R2. Distortion performance improves when the outputs are loaded as lightly as possible. Some earlier literature refers to these outputs as “LP” rather than V2. W BLOCK DIAGRA Overall Block Diagram Showing Four 3-Terminal 2nd Order Sections INV V1 V2 A V+ V1 V2 B – – C ∫ + V+ INV C ∫ + SHUTDOWN SWITCH V– R 2ND ORDER SECTIONS R SHUTDOWN SWITCH SHDN AGND D C + + ∫ V– – ∫ – C C 1562 BD INV 6 V1 V2 INV V1 V2 LTC1562 U W U U APPLICATIONS INFORMATION Functional Description Setting fO and Q The LTC1562 contains four matched, 2nd order, 3-terminal universal continuous-time filter blocks, each with a virtual-ground input node (INV) and two rail-to-rail outputs (V1, V2). In the most basic applications, one such block and three external resistors provide 2nd order lowpass and bandpass responses simultaneously (Figure 3, with a resistor for ZIN). The three external resistors set standard 2nd order filter parameters fO, Q and gain. A combination of internal precision components and external resistor R2 sets the center frequency fO of each 2nd order block. The LTC1562 is trimmed at manufacture so that fO will be 100kHz ±0.5% if the external resistor R2 is exactly 10k. Each of the four 2nd order sections in the LTC1562 can be programmed for a standard filter function (lowpass, bandpass or highpass) when configured as in Figure 3 with a resistor or capacitor for ZIN. These transfer functions all have the same denominator, a complex pole pair with center frequency ωO = 2πfO and quality parameter Q. (The numerators depend on the response type as described below.) External resistors R2 and RQ set fO and Q as follows: However, lowpass/bandpass filtering is only one specific application for the 2nd order building blocks in the LTC1562. Highpass response results if the external impedance ZIN in Figure 3 becomes a capacitor CIN (whose value sets only gain, not critical frequencies) as described below. Responses with zeroes are available through other connections (see Notches and Elliptic Responses). Moreover, the virtual-ground input gives each 2nd order section the built-in capability for analog operations such as gain (preamplification), summing and weighting of multiple inputs, handling input voltages beyond the power supplies or accepting current or charge signals directly. These Operational FilterTM frequency-selective building blocks are nearly as versatile as operational amplifiers. The user who is not copying exactly one of the Typical Applications schematics shown later in this data sheet is urged to read carefully the next few sections through at least Signal Swings, for orientation about the LTC1562, before attempting to design custom application circuits. Also available free from LTC, and recommended for designing custom filters, is the general-purpose analog filter design software FilterCADTM for Windows®. This software includes tools for finding the necessary f0, Q and gain parameters to meet target filter specifications such as frequency response. fO = Q = 10kΩ = 100kHz 2πC (R1)R2 R2 1 RQ = (R1)R2 RQ (10kΩ)R2 ( ) = RQ 100kHz R2 fO R1 = 10k and C = 159pF are internal to the LTC1562 while R2 and RQ are external. A typical design procedure proceeds from the desired fO and Q as follows, using finite-tolerance fixed resistors. First find the ideal R2 value for the desired fO: 2 100kHz R2 Ideal = 10kΩ fO ( ) ( ) Then select a practical R2 value from the available finitetolerance resistors. Use the actual R2 value to find the desired RQ, which also will be approximated with finite tolerance: RQ = Q (10kΩ)R2 The fO range is approximately 10kHz to 150kHz, limited mainly by the magnitudes of the external resistors required. As shown above, R2 varies with the inverse square of fO. This relationship desensitizes fO to R2’s Operational Filter and FilterCAD are trademarks of Linear Technology Corporation. Windows is a registered trademark of Microsoft Corporation. 7 LTC1562 U U W U APPLICATIONS INFORMATION Basic Lowpass tolerance (by a factor of 2 incrementally), but it also implies that R2 has a wider range than fO. (RQ and RIN also tend to scale with R2.) At high fO these resistors fall below 5k, heavily loading the outputs of the LTC1562 and leading to increased THD and other effects. At the other extreme, a lower fO limit of 10kHz reflects an arbitrary upper resistor limit of 1MΩ. The LTC1562’s MOS input circuitry can accommodate higher resistor values than this, but junction leakage current from the input protection circuitry may cause DC errors. When ZIN of Figure 3 is a resistor of value RIN, a standard 2nd order lowpass transfer function results from VIN to V2 (Figure 5): 2 – HL ω O V2(s) = HLP (s) = 2 VIN(s) s2 + ω O / Q s + ω O ( The DC gain magnitude is HL = R2/RIN. (Note that the transfer function includes a sign inversion.) Parameters ωO (= 2πfO) and Q are set by R2 and RQ as above. For a 2nd order lowpass response the gain magnitude becomes QHL The 2nd order transfer functions HLP(s), HBP(s) and HHP(s) (below) are all inverting so that, for example, at DC the lowpass gain is – HL. If two such sections are cascaded, these phase inversions cancel. Thus, the filter in the application schematic on the first page of this data sheet is a dual DC preserving, noninverting, rail-to-rail lowpass filter, approximating two “straight wires with frequency selectivity.” RIN VIN RQ 2nd ORDER fL fO Figure 5. Basic Lowpass Configuration fH fO ; fO = fL fH fH – fL 2 1 –1 fL = fO + + 1 2Q 2Q 2 1 1 fH = fO + + 1 2Q 2Q Q= 1562 F05 HIGHPASS RESPONSE HP HL 0.707 HL fP f (LOG SCALE) HP HH 0.707 HH fC fC 2 1 1 fC = fO 1 – + 1– 2 + 1 2Q2 2Q 1 2 1 1 fC = fO 1 – + 1– + 1 2 2Q2 2Q –1 2Q2 1 HP = HL 1 1 1– Q 4Q2 1 fP = fO 1 – 2Q2 1 HP = HH 1 1 1– Q 4Q2 Figure 4. Characteristics of Standard 2nd Order Filter Responses 8 fP f (LOG SCALE) f (LOG SCALE) fP = fO 1 – V2 1/4 LTC1562 GAIN (V/V) 0.707 HB V1 LOWPASS RESPONSE GAIN (V/V) GAIN (V/V) BANDPASS RESPONSE R2 VOUT INV Figure 4 shows further details of 2nd order lowpass, bandpass and highpass responses. Configurations to obtain these responses appear in the next three sections. HB ) –1 LTC1562 U U W U APPLICATIONS INFORMATION at frequency fO, and for Q > 0.707, a gain peak occurs at a frequency below fO, as shown in Figure 4. Basic Bandpass There are two different ways to obtain a bandpass function in Figure 3, both of which give the following transfer function form: HBP (s) = ( ) – HB ω O / Q s ( ) CIN VIN 2 s2 + ω O / Q s + ω O RQ CIN RIN VIN VIN RQ RQ R2 R2 VOUT V1 R2 VOUT ωO = 2πfO and Q are set by R2 and RQ as described previously in Setting fO and Q. When ZIN is a resistor of value RIN, a bandpass response results at the V1 output (Figure 6a) with a gain parameter HB = RQ/RIN. Alternatively, a capacitor of value CIN gives a bandpass response at the V2 output (Figure 6b), with the same HBP(s) expression, and the gain parameter now HB = (RQ/10kΩ)(CIN/159pF). This transfer function has a gain magnitude of HB (its peak value) when the frequency equals fO and has a phase shift of 180° at that frequency. Q measures the sharpness of the peak (the ratio of fO to – 3dB bandwidth) in a 2nd order bandpass function, as illustrated in Figure 4. INV Parameters ωO = 2πfO and Q are set by R2 and RQ as above. The highpass gain parameter is HH = CIN/159pF. For a 2nd order highpass response the gain magnitude at frequency fO is QHH, and approaches HH at high frequencies (f >> fO). For Q > 0.707, a gain peak occurs at a frequency above fO as shown in Figure 4. The transfer function includes a sign inversion. VOUT INV V2 V1 V2 2nd ORDER 2nd ORDER 1/4 LTC1562 1/4 LTC1562 1562 F06 (a) Resistive Input (b) Capacitive Input Figure 6. Basic Bandpass Configurations Basic Highpass When ZIN of Figure 3 is a capacitor of value CIN, a highpass response appears at the V1 output (Figure 7). – HHs2 V1(s) = HHP (s) = 2 VIN(s) s2 + ω O / Q s + ω O ( ) INV V1 V2 2nd ORDER 1/4 LTC1562 1562 F07 Figure 7. Basic Highpass Configuration Signal Swings The V1 and V2 outputs are capable of swinging to within roughly 100mV of each power supply rail. As with any analog filter, the signal swings in each 2nd order section must be scaled so that no output overloads (saturates), even if it is not used as a signal output. (Filter literature often calls this the “dynamics” issue.) When an unused output has a larger swing than the output of interest, the section’s gain or input amplitude must be scaled down to avoid overdriving the unused output. The LTC1562 can still be used with high performance in such situations as long as this constraint is followed. For an LTC1562 section as in Figure 3, the magnitudes of the two outputs V2 and V1, at a frequency ω = 2πf, have the ratio, | V2( jω )| (100kHz) = | V1( jω )| f regardless of the details of ZIN. Therefore, an input frequency above or below 100kHz produces larger output amplitude at V1 or V2, respectively. This relationship can guide the choice of filter design for maximum dynamic range in situations (such as bandpass responses) where there is more than one way to achieve the desired frequency response with an LTC1562 section. 9 LTC1562 U U W U APPLICATIONS INFORMATION Because 2nd order sections with Q ≥ 1 have response peaks near fO, the gain ratio above implies some rules of thumb: fO < 100kHz ⇒ V2 tends to have the larger swing fO > 100kHz ⇒ V1 tends to have the larger swing. The following situations are convenient because the relative swing issue does not arise. The unused output’s swing is naturally the smaller of the two in these cases: Lowpass response (resistor input, V2 output, Figure 5) with fO < 100kHz Bandpass response (capacitor input, V2 output, Figure 6b) with fO < 100kHz Bandpass response (resistor input, V1 output, Figure 6a) with fO > 100kHz Highpass response (capacitor input, V1 output, Figure 7) with fO > 100kHz The LTC1562-2, a higher frequency derivative of the LTC1562, has a design center fO of 200kHz compared to 100kHz in the LTC1562. The rules summarized above apply to the LTC1562-2 but with 200kHz replacing the 100kHz limits. Thus, an LTC1562-2 lowpass filter section with fO below 200kHz automatically satisfies the desirable condition of the unused output carrying the smaller signal swing. RIN 10k VIN RQ 6.98k INV R2 10k V1 V2 2nd ORDER 1/4 LTC1562 CL 30pF VOUT RL (EXTERNAL LOAD RESISTANCE) 1562 F08 Figure 8. 100kHz, Q = 0.7 Lowpass Circuit for Distortion vs Loading Test Low Level or Wide Range Input Signals The LTC1562 contains a built-in capability for low noise amplification of low level signals. The ZIN impedance in each 2nd order section controls the block’s gain. When set for unity passband gain, a 2nd order section can deliver an output signal more than 100dB above the noise level. If low 10 level inputs require further dynamic range, reducing the value of ZIN boosts the signal gain while reducing the input referred noise. This feature can increase the SNR for low level signals. Varying or switching ZIN is also an efficient way to effect automatic gain control (AGC). From a system viewpoint, this technique boosts the ratio of maximum signal to minimum noise, for a typical 2nd order lowpass response (Q = 1, fO = 100kHz), to 118dB. Input Voltages Beyond the Power Supplies Properly used, the LTC1562 can accommodate input voltage excursions well beyond its supply voltage. This requires care in design but can be useful, for example, when large out-of-band interference is to be removed from a smaller desired signal. The flexibility for different input voltages arises because the INV inputs are at virtual ground potential, like the inverting input of an op amp with negative feedback. The LTC1562 fundamentally responds to input current and the external voltage VIN appears only across the external impedance ZIN in Figure 3. To accept beyond-the-supply input voltages, it is important to keep the LTC1562 powered on, not in shutdown mode, and to avoid saturating the V1 or V2 output of the 2nd order section that receives the input. If any of these conditions is violated, the INV input will depart from a virtual ground, leading to an overload condition whose recovery timing depends on circuit details. In the event that this overload drives the INV input beyond the supply voltages, the LTC1562 could be damaged. The most subtle part of preventing overload is to consider the possible input signals or spectra and take care that none of them can drive either V1 or V2 to the supply limits. Note that neither output can be allowed to saturate, even if it is not used as the signal output. If necessary the passband gain can be reduced (by increasing the impedance of ZIN in Figure 3) to reduce output swings. The final issue to be addressed with beyond-the-supply inputs is current and voltage limits. Current entering the virtual ground INV input flows eventually through the output circuitry that drives V1 and V2. The input current magnitude (VIN/ZIN in Figure 3) should be limited by design to less than 1mA for good distortion performance. On the other hand, the input voltage VIN appears across the LTC1562 U U W U APPLICATIONS INFORMATION external component ZIN, usually a resistor or capacitor. This component must of course be rated to sustain the magnitude of voltage imposed on it. Lowpass “T” Input Circuit The virtual ground INV input in the Operational Filter block provides a means for adding an “extra” lowpass pole to any resistor-input application (such as the basic lowpass, Figure 5, or bandpass, Figure 6a). The resistor that would otherwise form ZIN is split into two parts and a capacitor to ground added, forming an R-C-R “T” network (Figure 9). This adds an extra, independent real pole at a frequency: fP = 1 2πRPCT where CT is the new external capacitor and RP is the parallel combination of the two input resistors RINA and RINB. This pair of resistors must normally have a prescribed series total value RIN to set the filter’s gain as described above. The parallel value RP can however be set arbitrarily (to RIN/4 or less) which allows choosing a convenient standard capacitor value for CT and fine tuning the new pole with RP. RINA A practical limitation of this technique is that the CT capacitor values that tend to be required (hundreds or thousands of pF) can destabilize the op amp in Figure 3 if RINB is too small, leading to AC errors such as Q enhancement. For this reason, when RINA and RINB are unequal, preferably the larger of the two should be placed in the RINB position. Highpass “T” Input Circuit A method similar to the preceding technique adds an “extra” highpass pole to any capacitor-input application (such as the bandpass of Figure 6b or the highpass of Figure 7). This method splits the input capacitance CIN into two series parts CINA and CINB, with a resistor RT to ground between them (Figure 10). This adds an extra 1st order highpass corner with a zero at DC and a pole at the frequency: fP = 1 2πRTCP where CP = CINA + CINB is the parallel combination of the two capacitors. At the same time, the total series capacitance CIN will control the filter’s gain parameter (HH in Basic Highpass). For a given series value CIN, the parallel value CP can still be set arbitrarily (to 4CIN or greater). CINA RINB VIN CINB VIN CT RQ INV V1 RT R2 V2 RQ INV 2nd ORDER V1 R2 V2 2nd ORDER 1/4 LTC1562 1/4 LTC1562 1562 F09 1562 F10 Figure 9. Lowpass “T” Input Circuit Figure 10. Highpass “T” Input Circuit The procedure therefore is to begin with the target extra pole frequency fP. Determine the series value RIN from the gain requirement. Select a capacitor value CT such that RP = 1/(2πfPCT) is no greater than RIN/4, and then choose RINA and RINB that will simultaneously have the parallel value RP and the series value RIN. Such RINA and RINB can be found directly from the expression: The procedure then is to begin with the target corner (pole) frequency fP. Determine the series value CIN from the gain requirement (for example, CIN = HH(159pF) for a highpass). Select a resistor value RT such that CP = 1/(2πRTfP) is at least 4CIN, and select CINA and CINB that will simultaneously have the parallel value CP and the series value CIN. Such CINA and CINB can be found directly from the expression: ( 2 1 1 RIN ± RIN – 4RINRP 2 2 ) ( 2 1 1 CP ± CP – 4CINCP 2 2 ) 11 LTC1562 U W U U APPLICATIONS INFORMATION This procedure can be iterated, adjusting the value of RT, to find convenient values for CINA and CINB since resistor values are generally available in finer increments than capacitor values. Different “fO” Measures Standard 2nd order filter algebra, as in Figure 4 and the various transfer-function expressions in this data sheet, uses a center frequency parameter fO (or ωO, which is 2πfO). fO can also be measured in practical ways, including: • The frequency where a bandpass response has 180° phase shift • The frequency where a bandpass response has peak gain • The geometric mean of the – 3.01dB gain frequencies in a bandpass (√ƒLƒH in Figure 4) An ideal mathematical 2nd order response yields exactly the same frequency by these three measures. However, real 2nd order filters with finite-bandwidth circuitry show small differences between the practical fO measures, which may be important in critical applications. The issue is chiefly of concern in high-Q bandpass applications where, as the data below illustrate, the different f0 measurements tend to converge anyway for the LTC1562. At low Q the bandpass peak is not sharply defined and the 12 – 3dB frequencies fL and fH are widely separated from this peak. The LTC1562’s fO is trimmed in production to give an accurate 180° phase shift in the configuration of Figure 6a with resistor values setting f0 = 100kHz and Q = 1. Table 1 below shows typical differences between fO values measured via the bandpass 180° criterion and fO values measured using the two other methods listed above (Figure 6a, RIN = RQ). Table 1 fO (BP 180°) Q=1 BP-PEAK fO Q=1 √ƒLƒH fO Q=5 BP-PEAK fO Q=5 √ƒLƒH fO 60kHz + 0.3% + 0.3% + 0.05% + 0.05% 100kHz + 0.6% + 0.6% + 0.1% + 0.1% 140kHz + 0.8% + 0.8% + 0.15% + 0.15% LTC1562 Demo Board The LTC1562 demo board is assembled with an LTC1562 or LTC1562A in a 20-pin SSOP package and power supply decoupling capacitors. Jumpers on the board configure the LTC1562 for dual or single supply operation and power shutdown. Pads for surface mount resistors and capacitors are provided to build application-specific filters. Also provided are terminals for inputs, outputs and power supplies. LTC1562 U TYPICAL APPLICATIONS (Basic) Quad 3rd Order Butterworth Lowpass Filter, Gain = – 1 RIN1B VIN1 CIN1 VOUT2 1 RQ1 R21 2 3 5 5V 0.1µF 6 R23 8 9 RIN3A RIN3B VIN3 RQ3 CIN3 10 INV B INV C V1 B V1 C V2 B V2 C V + LTC1562 V– SHDN AGND V2 A V2 D V1 A V1 D INV A VOUT3 INV D RIN2B 20 19 RQ2 18 R22 16 10 RIN2A VIN2 CIN2 –10 –5V 15 0.1µF 13 R24 –20 –30 –40 12 11 f– 3dB = 100kHz 0 GAIN (dB) VOUT1 RIN1A Amplitude Response RQ4 RIN4B RIN4A VIN4 CIN4 VOUT4 1562 TA05a –50 –60 10k SCHEMATIC INCLUDES PIN NUMBERS FOR 20-PIN PACKAGE. PINS 4, 7, 14, 17 (NOT SHOWN) ALSO CONNECT TO V – 100k FREQUENCY (Hz) 1M 1562 TA05b Quad 3rd Order Butterworth Lowpass Filters f – 3dB 20kHz f – 3dB 40kHz f – 3dB 60kHz f – 3dB 80kHz f – 3dB 100kHz f – 3dB 120kHz f – 3dB 140kHz CIN RINA RINB RQ R2 220pF 44.2k 205k 249k 249k 1000pF 4.32k 57.6k 61.9k 61.9k 1000pF 3.16k 24.3k 27.4k 27.4k 1000pF 2.43k 13.0k 15.4k 15.4k 1000pF 1.96k 8.06k 10.0k 10.0k 1000pF 1.87k 5.11k 6.98k 6.98k 1000pF 1.69k 3.4k 5.11k 5.11k All four sections have identical RINA, RINB and CIN values. All resistor values are ±1% 13 LTC1562 U TYPICAL APPLICATIONS (Basic) Dual 4th Order Lowpass Filters Amplitude Response 10 RIN2 VIN2 1 RQ1 R21 2 3 5 5V 0.1µF 6 8 RIN3 VIN1 R23 9 RQ3 10 INV B INV C V1 B V1 C V2 C V2 B V + LTC1562 V– AGND SHDN V2 D V2 A V1 D V1 A INV D INV A 0 20 19 RQ2 –10 18 R22 –20 VOUT2 –5V 16 15 GAIN (dB) RIN1 0.1µF VOUT1 13 –30 –40 –50 12 R24 –60 11 RQ4 –70 RIN4 BUTTERWORTH f – 3dB = 100kHz –80 10k 1562 TA03a 100k FREQUENCY (Hz) SCHEMATIC INCLUDES PIN NUMBERS FOR 20-PIN PACKAGE. PINS 4, 7, 14, 17 (NOT SHOWN) ALSO CONNECT TO V – 1M 1562 TA03b Quick Design Formulas for Some Popular Response Types: Butterworth (Maximally Flat Passband) for fC 10kHz to 140kHz R21, R23, RIN1, RIN3 = 10k 100kHz ƒC RQ1, RQ3 = 5.412k 100kHz ƒC R22, R24, RIN2, RIN4 = 10k 100kHz ƒC RQ2, RQ4 = 13.07k 100kHz ƒC Chebyshev (Equiripple Passband) for fC 20kHz to 120kHz 2 14.24k 100kHz ƒC 7.26k 100kHz ƒC 7.097k 100kHz ƒC 17.53k 100kHz ƒC 2 Bessel (Good Transient Response) for fC 10kHz to 70kHz 2 3.951k 100kHz ƒC 5.066k 100kHz ƒC 4.966k 100kHz ƒC 3.679k 100kHz ƒC 2 2 2 Notes: fC is the cutoff frequency: For Butterworth and Bessel, response is 3dB down at fC. For Chebyshev filters with ± 0.1dB passband ripple up to 0.95 fC, use LTC1562 “A” grade. Example: Butterworth response, fC = 50kHz. from the formulas above, R21 = R23 = RIN1 = RIN3 = 10k(100kHz/50kHz)2 = 40k. RQ1 = RQ3 = 5.412k(100kHz/50kHz) = 10.82k. R22 = R24 = RIN2 = RIN4 = 10k(100kHz/50kHz)2 = 40k. 1562 TA03 TABLE RQ2 = RQ4 = 13.07k(100kHz/50kHz) = 26.14k. Use nearest 1% values. 14 LTC1562 U TYPICAL APPLICATIONS (Basic) 8th Order Lowpass Filters Amplitude Response 10 RIN2 VIN 1 RQ1 R21 2 3 5 5V 0.1µF 6 R23 8 9 RQ3 10 INV B INV C V1 B V1 C V2 B V2 C V– V + LTC1562 SHDN AGND V2 A V2 D V1 A V1 D INV D INV A 19 RQ2 18 R22 16 15 13 –10 –20 –5V 0.1µF R24 –30 –40 –50 –60 12 11 CHEBYSHEV fC = 100kHz 0 20 GAIN (dB) RIN1 –70 RQ4 –80 –90 10k RIN4 VOUT RIN3 100k FREQUENCY (Hz) 500k 1562 TA04b 1562 TA04a SCHEMATIC INCLUDES PIN NUMBERS FOR 20-PIN PACKAGE. PINS 4, 7, 14, 17 (NOT SHOWN) ALSO CONNECT TO V – Quick Design Formulas for Some Popular Response Types: Butterworth (Maximally Flat Passband) for fC 10kHz to 140kHz R21 = RIN1 = 10k 100kHz ƒC RQ1 = 6.01k 100kHz ƒC R22 = RIN2 = 10k 100kHz ƒC RQ2 = 9k 100kHz ƒC R23 = RIN3 = 10k 100kHz ƒC RQ3 = 5.1k 100kHz ƒC R24 = RIN4 = 10k 100kHz ƒC RQ4 = 25.63k 100kHz ƒC Chebyshev (Equiripple Passband) for fC 20kHz to 120kHz 2 Bessel (Good Transient Response) for fC 10kHz to 70kHz 2 R21 = 7.51k 100kHz , RIN1 = 2.2R21* ƒC RQ1 = 119.3k 2 R22 = RIN2 = 14.99k RQ2 = 279.9k 2 R23 = RIN3 = 7.15k 100kHz ƒC 100kHz ƒC RQ1 = 3.63k 100kHz ƒC R22 = RIN2 = 2.07k 100kHz ƒC RQ2 = 5.58k 100kHz ƒC R23 = RIN3 = 2.96k 100kHz ƒC RQ3 = 3.05k 100kHz ƒC R24 = RIN4 = 3.14k 100kHz ƒC RQ4 = 2.84k 100kHz ƒC 2 100kHz 100kHz ƒC + 2440kHz ƒC 100kHz ƒC RQ3 = 118.1k 2 100kHz 100kHz ƒC ƒC + 560kHz R21 = RIN1 = 2.61k 2 100kHz 100kHz ƒC ƒC + 530kHz 2 R24 = 26.7k 100kHz R24* , RIN4 = ƒC 2.2 RQ4 = 8.75k 100kHz ƒC 2 2 2 2 Notes: fC is the cutoff frequency: For Butterworth and Bessel, response is 3dB down at fC. For Chebyshev filters with ± 0.1dB passband ripple up to 0.95 fC, use LTC1562 “A” grade. *The resistor values marked with an asterisk (*) in the Chebyshev formulas (R21 and R24) should be rounded to the nearest standard finite-tolerance value before computing the values dependent on them (RIN1 and RIN4 respectively). Example: Chebyshev response, fC = 100kHz. The formulas above give R21 = 7.51k, nearest standard 1% value 7.50k. Using this 1% value gives RIN1 = 16.5k, already a standard 1% value. RQ1 = 18.075k, nearest 1% value 18.2k. R22 = RIN2 = 14.99k, nearest 1% value 15k. RQ2 = 11.02k, nearest 1% value 11k. R23 = RIN3 = 7.15k, already a standard 1% value. RQ3 = 18.75k, nearest 1% value 18.7k. R24 = 26.7k, already a standard 1% value. This gives RIN4 = 12.14k, nearest 1% value 12.1k. RQ4 = 8.75k, nearest 1% value 8.66k. 1562 TA04 TABLE 15 LTC1562 U TYPICAL APPLICATIONS (Basic) 8th Order Bandpass Filter, Single 5V Supply, Center Frequency – 3dB Bandwidth = 10 Amplitude Response 10 RIN2 CIN1 0 VIN RQ1 R21 2 3 5 5V 0.1µF 6 R23 8 9 RQ3 CIN3 10 INV B INV C V1 B V1 C V2 B V2 C V + LTC1562 V– AGND SHDN V2 D V2 A V1 D V1 A INV D INV A 20 19 RQ2 18 R22 –10 16 15 fCENTER = 80kHz – 20 GAIN (dB) 1 1µF R24 13 – 30 – 40 – 50 – 60 12 –70 RQ4 11 – 80 – 90 VOUT RIN4 40 48 56 64 72 80 88 96 104 112 120 FREQUENCY (kHz) 1562 TA07a 1562 TA07b SCHEMATIC INCLUDES PIN NUMBERS FOR 20-PIN PACKAGE. PINS 4, 7, 14, 17 (NOT SHOWN) ALSO CONNECT TO V – Quick Design Formulas for Center Frequency fC (Recommended Range 40kHz to 140kHz): R21 = R23 = 10.6k 100kHz ƒC R22 = R24 = 9.7k 100kHz ƒC CIN1 = CIN3 = 159pF 10k RQ1 2 RQ1 = RQ3 = 164.6k 100kHz 100kHz ƒC ƒC + 319kHz RQ2 = RQ4 = 143.2k 100kHz 100kHz ƒC + 294kHz ƒC 2 RIN2 = RIN4 = R22RQ1CIN1 100kHz (10k)(10.6pF) ƒC + 286kHz Notes: RQ1, R22 and CIN1 should be rounded to the nearest standard finite-tolerance value before using these values in the later formulas. Example: Center frequency fC of 80kHz. The formulas give R21 = R23 = 16.56k, nearest standard 1% value 16.5k. RQ1 = RQ3 = 51.56k, nearest 1% value 51.1k. R22 = R24 = 15.15k, nearest 1% value 15k. RQ2 = RQ4 = 47.86k, nearest 1% value 47.5k. CIN1 = CIN2 = 31.11pF using 51.1k for RQ1, nearest standard 5% capacitor value 33pF. This and the 1% value R22 = 15k also go into the calculation for RIN2 = RIN4 = 65.20k, nearest 1% value 64.9k. 1562 TA07 TABLE 16 LTC1562 U TYPICAL APPLICATIONS (Basic) 8th Order Bandpass Filter, Single 5V Supply, Center Frequency – 1dB Bandwidth = 10 Amplitude Response RIN2 VIN 1 RQ1 R21 2 3 5 5V 0.1µF 6 R23 8 9 RQ3 10 INV B INV C V1 B V1 C V2 B V2 C – V + LTC1562 V AGND SHDN V2 A V2 D V1 A V1 D INV D INV A 0 20 19 18 – 20 R22 16 15 13 1µF R24 – 30 – 40 – 50 – 60 12 11 fCENTER = 100kHz –10 RQ2 GAIN (dB) RIN1 10 –70 RQ4 – 80 VOUT RIN4 RIN3 – 90 60 68 76 84 92 100 108 116 124 132 140 FREQUENCY (kHz) 1562 TA06a 1562 TA06b SCHEMATIC INCLUDES PIN NUMBERS FOR 20-PIN PACKAGE. PINS 4, 7, 14, 17 (NOT SHOWN) ALSO CONNECT TO V – Quick Design Formulas for a Center Frequency fC (Recommended Range 50kHz to 120kHz): R21 = R23 = 11.7k 100kHz ƒC R22 = R24 = 8.66k 100kHz ƒC RIN1 = RIN3 = ƒC + 1736kHz R21 2.56 100kHz RQ1 = RQ3 = 215.5k 100kHz 100kHz ƒC + 229kHz ƒC RIN2 = RIN4 = ƒC + 634kHz RQ2 14.36 100kHz RQ2 = RQ4 = 286.2k 100kHz 100kHz ƒC + 351kHz ƒC 2 2 Notes: R21 and RQ2 should be rounded to the nearest standard finite-tolerance value before using these values in the later formulas. For fC < 100kHz, the maximum peak-to-peak passband input level is (fC /100kHz)5V. Use LTC1562A for minimum variation of passband gain. Example: Center frequency fC of 100kHz. The formulas give R21 = R23 = 11.7k, nearest standard 1% value 11.5k. This value gives RIN1 = RIN3 = 82.46k, nearest 1% value 82.5k. RQ1 = RQ3 = 65.5k, nearest 1% value 64.9k. R22 = R24 = 8.66k, already a standard 1% value. This gives RIN2 = RIN4 = 32.4k (again already a standard 1% value). RQ2 = RQ4 = 63.45k, nearest 1% value 63.4k. If LTC1562A is used, resistor tolerances tighter than 1% will further 1562 TA06 TABLE improve the passband gain accuracy. 17 LTC1562 U TYPICAL APPLICATIONS (Basic) 8th Order Bandpass (High Frequency) Filter Center Frequency – 3dB Bandwidth = , Gain = 10 10 Amplitude Response RIN2 VIN 1 RQ1 2 R21 3 5 V+ 0.1µF 6 R23 8 9 RQ3 10 INV B INV C V1 B V1 C V2 B V2 C – V + LTC1562 V AGND SHDN V2 A V2 D V1 A V1 D INV D INV A 19 RQ2 18 R22 10 0 16 15 fCENTER = 100kHz 20 20 GAIN (dB) RIN1 30 V– 0.1µF R24 13 –10 – 20 – 30 – 40 12 – 50 RQ4 11 – 60 – 70 RIN4 VOUT 40 RIN3 60 80 100 120 140 FREQUENCY (kHz) 1562 TA08a 160 180 1562 TA08b SCHEMATIC INCLUDES PIN NUMBERS FOR 20-PIN PACKAGE. PINS 4, 7, 14, 17 (NOT SHOWN) ALSO CONNECT TO V – 8th Order Bandpass Filter f – 3dB BW = CENTER, Gain = 10 10 fCENTER 80kHz fCENTER 90kHz fCENTER 100kHz fCENTER 110kHz fCENTER 120kHz fCENTER 130kHz fCENTER 140kHz 4.64k 46.4k 12.4k 5.23k 52.3k 15.4k 6.34k 42.2k 10.0k 5.11k 38.3k 8.25k 5.11k 34.8k 6.98k 5.49k 32.4k 5.9k 5.62k 30.1k 5.11k 46.4k 46.4k 12.4k 52.3k 52.3k 15.4k 42.2k 42.2k 10.0k 38.3k 38.3k 8.25k 34.8k 34.8k 6.98k 32.4k 32.4k 5.90k 30.1k 30.1k 5.11k Side B RIN1 RQ1 R21 Sides A, C, D RIN2, RIN3, RIN4 RQ2, RQ3, RQ4 R22, R23, R24 All resistor values are ±1% 18 LTC1562 U TYPICAL APPLICATIONS (Basic) 8th Order Wideband Bandpass Filter fCENTER = 50kHz, – 3dB BW 40kHz to 60kHz Amplitude Response RIN2 69.8k CIN1 22pF 10 0 RQ1 59k 2 R21 56.2k 3 5 V+ 0.1µF 6 8 R23 63.4k 9 RQ3 82.5k 10 INV B INV C V1 B V1 C V2 B V2 C 20 19 RQ2 48.7k 18 R22 34.8k – 16 V + LTC1562 V 15 AGND SHDN 13 V2 D V2 A 12 V1 D V1 A 11 INV D INV A – 10 GAIN (dB) 1 VIN V– 1µF – 20 – 30 – 40 R24 28.7k – 50 RQ4 100k – 60 20 CIN3 27pF 100 FREQUENCY (kHz) VOUT 1562 TA09b CIN4 47pF 1562 TA09a SCHEMATIC INCLUDES PIN NUMBERS FOR 20-PIN PACKAGE. PINS 4, 7, 14, 17 (NOT SHOWN) ALSO CONNECT TO V – 8th Order Highpass 0.05dB Ripple Chebyshev Filter fCUTOFF = 30kHz Amplitude Response 10 CIN1 150pF RQ1, 10.2k R21, 35.7k 2 3 5 5V 0.1µF 6 8 CIN3 150pF R23, 107k 9 RQ3, 54.9k 10 INV B INV C V1 B V1 C V2 C V2 B V + LTC1562 V– SHDN AGND V2 A V2 D V1 A V1 D INV D INV A 0 20 19 RQ2, 22.1k 18 R22, 66.5k 16 15 –10 CIN2 150pF –20 GAIN (dB) 1 CIN –5V 0.1µF 13 –30 –40 –50 –60 12 R24, 127k 11 RQ4, 98.9k CIN4 150pF –70 –80 –90 1562 TA10a VOUT 1k SCHEMATIC INCLUDES PIN NUMBERS FOR 20-PIN PACKAGE. PINS 4, 7, 14, 17 (NOT SHOWN) ALSO CONNECT TO V – 10k 100k FREQUENCY (Hz) 1M 1562 TA10b TOTAL OUTPUT NOISE = 40µVRMS Amplitude Response 2nd Order 30kHz Highpass Cascaded with 6th Order 138kHz Lowpass RIN2, 5.23k 1 VIN RQ1, 30.1k R21, 110k 2 3 5 5V 0.1µF 6 8 R23, 5.23k 9 RQ3, 14k 10 RIN3, 8.06k INV B INV C V1 B V1 C V2 B V2 C 20 10 20 19 18 – 16 V + LTC1562 V 15 AGND SHDN 13 V2 D V2 A 12 V1 D V1 A 11 INV D INV A RQ2, 5.11k 0 R22, 5.23k –10 GAIN (dB) CIN1 150pF –5V 0.1µF –40 –60 RQ4, 3.74k SCHEMATIC INCLUDES PIN NUMBERS FOR 20-PIN PACKAGE. PINS 4, 7, 14, 17 (NOT SHOWN) ALSO CONNECT TO V – –30 –50 R24, 5.23k RIN4, 3.4k –20 –70 VOUT 1562 TA11a –80 10 100 FREQUENCY (kHz) 400 1562 TA11b ALL RESISTORS = 1% METAL FILM 19 LTC1562 U U W U APPLICATIONS INFORMATION Notches and Elliptic Responses fN. The two signals then cancel out at frequency fN. The notch depth (the completeness of cancellation) will be infinite to the extent that the two paths have matching gains. Three practical circuit methods are presented here, with different features and advantages. The basic (essentially all-pole) LTC1562 circuit techniques described so far will serve many applications. However, the sharpest-cutoff lowpass, highpass and bandpass filters include notches (imaginary zero pairs) in the stopbands. A notch, or band-reject, filter has zero gain at a frequency fN. Notches are also occasionally used by themselves to reject a narrow band of frequencies. A number of circuit methods will give notch responses from an Operational Filter block. Each method exhibits an inputoutput transfer function that is a standard 2nd order bandreject response: HBR (s) = Examples and design procedures for practical filters using these techniques appear in a series of articles1 attached to this data sheet on the Linear Technology web site (www.linear-tech.com). Also available free is the analog filter design software, FilterCAD for Windows, recommended for designing filters not shown in the Typical Applications schematics in this data sheet. – HN s2 + ω N2 ( Elementary Feedforward Notches ) A “textbook” method to get a 180° phase difference at frequency fN for a notch is to dedicate a bandpass 2nd order section (described earlier under Basic Bandpass), which gives 180° phase shift at the section’s center frequency fO (Figure 11, with CIN1 = 0), so that fN = fO. The bandpass section of Figure 6a, at its center frequency fO, has a phase shift of 180° and a gain magnitude of HB = RQ /RIN. A notch results in Figure 11 if the paths summed into virtual ground have the same gains at the 180° frequency (then IO = 0). This requires a constraint on the resistor values: s2 + ω O / Q s + ω 2O with parameters ωN = 2πfN and HN set by component values as described below. (ω0 = 2πf0 and Q are set for the Operational Filter block by its R2 and RQ resistors as described earlier in Setting f0 and Q). Characteristically, the gain magnitude |HBR(j2πf)| has the value HN(fN2/f02) at DC (f = 0) and HN at high frequencies (f >> fN), so in addition to the notch, the gain changes by a factor: High Frequency Gain DC Gain = ƒ 2O RIN2 R = Q1 RFF 2 RIN1 ƒN2 The common principle in the following circuit methods is to add a signal to a filtered replica of itself having equal gain and 180° phase difference at the desired notch frequency 1Nello Sevastopoulos, et al. “How to Design High Order Filters with Stopband Notches Using the LTC1562 Quad Operational Filter.” Attached to this data sheet, available on the LTC web site (www.linear-tech.com). CIN1 RIN1 VIN RQ1 INV V1 R21 RIN2 V2 IO VIRTUAL GROUND 2nd ORDER 1/4 LTC1562 RFF2 RGAIN – + VOUT 1562 F11 Figure 11. Feedforward Notch Configuration for fN ≥ fO 20 LTC1562 U W U U APPLICATIONS INFORMATION Note that the depth of the notch depends on the accuracy of this resistor ratioing. The virtual-ground summing point in Figure 11 may be from an op amp as shown, or in a practical cascaded filter, the INV input of another Operational Filter block. The transfer function in Figure 11 with CIN1 = 0 is a “pure” notch (fN = f0) of the HBR(s) form above, and the parameters are: ƒN = ƒ O R HN = GAIN RFF 2 Feedforward Notches for fN > f0 When CIN1 ≠ 0 in Figure 11, the notch frequency fN is above the center frequency f0 and the response has a lowpass shape as well as a notch (Figure 13). CIN1 contributes phase lead, which increases the notch frequency above the center frequency of the 2nd order Operational Filter section. The resistor constraint from the previous section also applies here and the HBR(s) parameters become: 1 ƒN = ƒ O Because fN = f0 in this case, the gain magnitude both at DC and at high frequencies (f >> fN) is the same, HN (assuming that the op amp in Figure 11 adds no significant frequency response). Figure 12 shows this. Such a notch is inefficient as a cascaded part of a highpass, lowpass or bandpass filter (the most common uses for notches). Variations of Figure 11 can add a highpass or lowpass shape to the notch, without using more Operational Filter blocks. The key to doing so is to decouple the notch frequency fN from the center frequency f0 of the Operational Filter (this is shown in Figures 13 and 15). The next two sections summarize two variations of Figure 11 with this highpass/ lowpass shaping, and the remaining section shows a different approach to building notches. 1– R ƒ2 HN = GAIN O RFF 2 ƒN2 C is the internal capacitor value in the Operational Filter (in the LTC1562, 159pF). The configuration of Figure 11 is most useful for a stopband notch in a lowpass filter or as an upper stopband notch in a bandpass filter, since the two resistors RIN2 and RFF2 can replace the input resistor RIN of either a lowpass section (Figure 5) or a resistor-input bandpass section (Figure 6a) built from a second Operational Filter. The configuration is 0 20 DC GAIN = HN –20 – 40 GAIN (dB) GAIN (dB) 0 – 60 –100 fN = fO = 100kHz HN = 1 Q=1 10 100 FREQUENCY (kHz) 1000 AN54 • TA18 Figure 12. Notch Response with fN = fO ( ) fN2 fO2 HIGH FREQ GAIN = HN –20 – 40 – 80 RIN1CIN1 RQ1C – 60 fO = 100kHz fN = 200kHz Q=1 DC GAIN = 0dB 10 100 FREQUENCY (kHz) 1000 1562 F13 Figure 13. Notch Response with fN > fO 21 LTC1562 U W U U APPLICATIONS INFORMATION Feedforward Notches for fN < f0 Just as feedforward around an inverting bandpass section yields a notch at the section’s f0 (Figure 11 with CIN1 = 0), feedforward around an inverting lowpass section causes a notch at zero frequency (which is to say, a highpass response). Moreover, and this is what makes it useful, introducing a capacitor for phase lead moves the notch frequency up from DC, exactly as CIN1 in Figure 11 moves the notch frequency up from the center frequency f0. In Figure 14, the inverting lowpass output (V2) of the Operational Filter is summed, at a virtual ground, with a fedforward input signal. Capacitor CIN1 shifts the resulting notch frequency, fN, up from zero, giving a low frequency notch with a highpass shape (Figure 15). The HBR(s) response parameters are now: RIN2 R C = Q1 IN1 RFF 2 R1C R1 and C are the internal precision components (in the LTC1562, 10k and 159pF respectively) as described above in Setting f0 and Q. The configuration of Figure 14 is most useful as a lower stopband notch in a bandpass filter, because the resistors RIN2 and RFF2 can replace the input resistor RIN of a bandpass section made from a second Operational Filter, as in Figure 6a. The configuration is robust against tolerances in the CIN1 value when fN approaches f0 (for f0/fN ≤ 1.4, as a rule of thumb) which is attractive in narrow transition-band filters, because of the relative cost of high accuracy capacitors. Further application details appear in Part 2 of the series of articles.1 20 ( ) R1 C R21 ƒN = ƒ O 1 – RQ1 CIN1 RIN1 HN = HIGH FREQ GAIN = HN fN2 0 DC GAIN = HN 2 fO GAIN (dB) robust against tolerances in the CIN1 value when fN approaches f0 (for fN/f0 ≤ 1.4, as a rule of thumb) which is attractive in narrow transition-band filters, because of the relative cost of high accuracy capacitors. Further application details appear in Part 1 of the series of articles.1 –20 – 40 RGAIN RFF 2 fO = 100kHz fN = 50kHz Q=1 HIGH FREQ GAIN = 0dB – 60 10k 100k FREQUENCY (Hz) 1562 F15 The constraint required for exact cancellation of the two paths (i.e., for infinite notch depth) becomes: Figure 15. Notch Response with fN < f0 CIN1 RIN1 VIN RQ1 INV V1 R21 RIN2 V2 IO VIRTUAL GROUND 2nd ORDER 1/4 LTC1562 RFF2 RGAIN – + VOUT 1562 F14 Figure 14. Feedforward Notch Configuration for fN < fO 22 1M LTC1562 U U W U APPLICATIONS INFORMATION R R21 DC Gain = GAIN RIN1 RN R-C Universal Notches A different way to get 180° phase shift for a notch is to use the built-in 90° phase difference between the two Operational Filter outputs along with a further 90° from an external capacitor. This method achieves deep notches independent of component matching, unlike the previous techniques, and it is convenient for cascaded highpass as well as lowpass and bandpass filters. The V2 output of an Operational Filter is a time-integrated version of V1 (see Figure 3), and therefore lags V1 by 90° over a wide range of frequencies. In Figure 16, a notch response occurs when a 2nd order section drives a virtualground input through two paths, one through a capacitor and one through a resistor. Again, the virtual ground may come from an op amp as shown, or from another Operational Filter’s INV input. Capacitor CN adds a further 90° to the 90° difference between V1 and V2, producing a wideband 180° phase difference, but frequency-dependent amplitude ratio, between currents IR and IC. At the frequency where IR and IC have equal magnitude, IO becomes zero and a notch occurs. This gives a net transfer function from VIN to VOUT in the form of HBR(s) as above, with parameters: ƒN = 1 ƒ O2 ƒN2 = High Frequency Gain DC Gain = RN CN R21C R1 and C are the internal precision components (in the LTC1562, 10k and 159pF respectively) as described above in Setting f0 and Q. Unlike the notch methods of Figures 11 and 14, notch depth from Figure 16 is inherent, not derived from component matching. Errors in the RN or CN values alter the notch frequency, fN, rather than the degree of cancellation at fN. Also, the notch frequency, fN, is independent of the section’s center frequency f0, so fN can freely be equal to, higher than or lower than f0 (Figures 12, 13 or 15, respectively) without changing the configuration. The chief drawback of Figure 16 compared to the previous methods is a very practical one—the CN capacitor value directly scales HN (and therefore the high frequency gain). Capacitor values are generally not available in increments or tolerances as fine as those of resistors, and this configuration lacks the property of the previous two configurations that sensitivity to the capacitor value falls as fN approaches f0. 2π RN CNR1C R C HN = GAIN N RIN1 C RIN1 VIN RQ1 INV V1 R21 RN IR VIRTUAL GROUND V2 2nd ORDER 1/4 LTC1562 IO CN IC RGAIN – + VOUT 1562 F16 Figure 16. The R-C Universal Notch Configuration for an Operational Filter Block 23 LTC1562 U TYPICAL APPLICATIONS (Advanced) 8th Order 50kHz Lowpass Elliptic Filter with 100dB Stopband Attenuation CIN2 24pF Amplitude Response RIN2 37.4k 1 VIN RQ1 30.1k 2 R21 31.6k 3 5 5V 0.1µF 6 8 R23 31.6k 9 RQ3 34k 10 INVB INVC V1B V1C V2B V2C V+ V– LTC1562 SHDN AGND V2A V2D V1A V1D INVA INVD 20 20 19 RQ2 13k 0 18 R22 57.6k –20 16 GAIN (dB) RIN1 48.7k – 5V 0.1µF 15 13 12 – 60 –80 R24 32.4k –100 11 RQ4 11.5k –120 VOUT RIN4 32.4k RIN3 31.6k – 40 10 100 FREQUENCY (kHz) 500 1562 TA12b CIN3 18pF CIN4 10pF 1562 TA12a USES THREE R-C UNIVERSAL NOTCHES AT fN = 133kHz, 167kHz, 222kHz. DETAILED DESCRIPTION IN LINEAR TECHNOLOGY DESIGN NOTE 195. WIDEBAND OUTPUT NOISE 60µVRMS SCHEMATIC INCLUDES PIN NUMBERS FOR 20-PIN PACKAGE. PINS 4, 7, 14, 17 (NOT SHOWN) ALSO CONNECT TO V – 8th Order 100kHz Elliptic Bandpass Filter RFF2 301k RIN2 93.1k 1 VIN CIN1 5.6pF RQ1 86.6k R21 10.7k 2 3 5 5V 0.1µF R23 10k RQ3 71.5k RIN3 294k 6 8 9 10 INVC INVB V1C V1B V2C V2B V+ LTC1562 SHDN V– AGND V2A V2D V1A V1D INVA CIN3 18pF INVD Amplitude Response 20 19 18 10 RQ2 84.5k 0 –10 R22 10k –20 16 15 – 5V 0.1µF 13 12 R24 9.53k 11 RQ4 82.5k GAIN (dB) RIN1 95.3k – 30 – 40 – 50 – 60 –70 – 80 – 90 RIN4 95.3k VOUT RFF4 332k 25 100 FREQUENCY (kHz) 175 1562 TA13b SCHEMATIC INCLUDES PIN NUMBERS FOR 20-PIN PACKAGE. PINS 4, 7, 14, 17 (NOT SHOWN) ALSO CONNECT TO V – 24 1562 F13a LTC1562 U TYPICAL APPLICATIONS (Advanced) 9th Order 22kHz Lowpass Elliptic Filter RIN2 249k CIN2 33pF RIN1A 140k RIN1B 69.8k 1 VIN CIN1 390pF RQ1 95.3k R21 324k 2 3 5 V+ 0.1µF V– R23 196k RQ3 392k 6 8 9 10 CIN3 27pF INVB INVC V1B V1C V2B V2C V+ SHDN LTC1562 V– AGND V2A V2D V1A V1D INVA INVD TO PIN 10 20 19 RQ2 182k 18 R22 226k RIN3 536k 16 V– 0.1µF 15 13 R24 649k 12 RQ4 66.5k 11 RIN4 301k CIN4 56pF VOUT 1562 F14a SCHEMATIC INCLUDES PIN NUMBERS FOR 20-PIN PACKAGE. PINS 4, 7, 14, 17 (NOT SHOWN) ALSO CONNECT TO V – Noise + THD vs Frequency 10 – 40 0 – 45 –10 – 50 – 20 – 55 NOISE + THD (dB) GAIN (dB) Amplitude Response – 30 – 40 – 50 – 60 –70 VIN = 1.65VRMS = 4.6VP-P VS = ± 5V – 60 – 65 –70 –75 – 80 – 80 – 85 – 90 – 90 5 10 FREQUENCY (kHz) 50 1 10 20 FREQUENCY (kHz) 1562 TA14b 1562 TA14c 25 LTC1562 U TYPICAL APPLICATIONS (Advanced) Dual 5th Order Lowpass “Elliptic” Filter RIN2 CIN2 Amplitude Response RIN1B 1 VIN1 CIN1 RQ1 2 R21 3 5 5V 0.1µF R21 8 RQ1 RIN1A 6 9 RIN1B 10 VIN2 INVC INVB V1B V1C V2B V2C + V– V LTC1562 SHDN AGND V2A V2D V1A V1D INVA INVD CIN1 VOUT1 20 19 RQ2 18 R22 20 fC = 100kHz 0 –20 16 – 5V GAIN (dB) RIN1A 0.1µF 15 13 R22 12 RQ2 – 40 – 60 – 80 –100 11 –120 VOUT2 CIN2 RIN2 10 100 FREQUENCY (kHz) 1000 1562 TA15b 1562 TA15a SCHEMATIC INCLUDES PIN NUMBERS FOR 20-PIN PACKAGE. PINS 4, 7, 14, 17 (NOT SHOWN) ALSO CONNECT TO V – fC (Hz) RIN1A RIN1B CIN1 RQ1 R21 RIN2 CIN2 RQ2 R22 100k 5.9k 7.5k 680pF 28k 7.5k 6.34k 68pF 9.31k 11.3k 75k 8.06k 15.4k 560pF 36.5k 13.3k 11.3k 68pF 12.7k 20k 50k 16.9k 35.7k 390pF 56.2k 30.1k 25.5k 68pF 18.7k 44.2k Construction and Instrumentation Cautions 100dB rejections at hundreds of kilohertz require electrically clean, compact construction, with good grounding and supply decoupling, and minimal parasitic capacitances in critical paths (such as Operational Filter INV inputs). In a circuit with 5k resistances trying for 100dB rejection at 100kHz, a stray coupling of 0.003pF around the signal path can preclude the 100dB. (By comparison, the stray capacitance between two adjacent pins of an IC can be 1pF or more.) Also, high quality supply bypass capacitors of 0.1µF near the chip provide good decoupling from a clean, low inductance power source. But several inches of wire (i.e., a few microhenrys of inductance) from the power supplies, unless decoupled by substantial 26 capacitance (≥ 10µF) near the chip, can cause a high-Q LC resonance in the hundreds of kHz in the chip’s supplies or ground reference, impairing stopband rejection and other specifications at those frequencies. In demanding filter circuits we have often found that a compact, carefully laid out printed circuit board with good ground plane makes a difference of 20dB in both stopband rejection and distortion performance. Highly selective circuits can even exhibit these issues at frequencies well below 100kHz. Finally, equipment to measure filter performance can itself introduce distortion or noise floors; checking for these limits with a wire replacing the filter is a prudent routine procedure. LTC1562 U PACKAGE DESCRIPTION Dimensions in inches (millimeters) unless otherwise noted. G Package 20-Lead Plastic SSOP (0.209) (LTC DWG # 05-08-1640) 0.278 – 0.289* (7.07 – 7.33) 20 19 18 17 16 15 14 13 12 11 0.301 – 0.311 (7.65 – 7.90) 1 2 3 4 5 6 7 8 9 10 0.205 – 0.212** (5.20 – 5.38) 0.068 – 0.078 (1.73 – 1.99) 0° – 8° 0.005 – 0.009 (0.13 – 0.22) 0.022 – 0.037 (0.55 – 0.95) *DIMENSIONS DO NOT INCLUDE MOLD FLASH. MOLD FLASH SHALL NOT EXCEED 0.006" (0.152mm) PER SIDE **DIMENSIONS DO NOT INCLUDE INTERLEAD FLASH. INTERLEAD FLASH SHALL NOT EXCEED 0.010" (0.254mm) PER SIDE 0.0256 (0.65) BSC 0.010 – 0.015 (0.25 – 0.38) Information furnished by Linear Technology Corporation is believed to be accurate and reliable. However, no responsibility is assumed for its use. Linear Technology Corporation makes no representation that the interconnection of its circuits as described herein will not infringe on existing patent rights. 0.002 – 0.008 (0.05 – 0.21) G20 SSOP 0595 27 LTC1562 U TYPICAL APPLICATION Amplitude Response 20 GAIN (dB) 0 Dual 4th Order 12dB Gaussian Lowpass Filter RIN2 RIN1 1 VIN2 RQ1 R21 2 3 5 5V 0.1µF 6 8 R23 RIN3 RQ3 VIN1 9 10 INV B INV C V1 B V1 C V2 B V2 C V + LTC1562 V– SHDN AGND V2 A V2 D V1 A V1 D INV D INV A – 20 fC = 64kHz fC = 32kHz – 40 fC = 16kHz 20 19 RQ2 18 R22 10 1 1µF 100 12 R24 11 RQ4 300 FREQUENCY (kHz) 1562 TA16b VOUT1 13 RIN4 – 80 VOUT2 16 15 – 60 4-Level Eye Diagram fC = 16kHz, Data Clock = 32kHz 1562 TA16a SCHEMATIC INCLUDES PIN NUMBERS FOR 20-PIN PACKAGE. PINS 4, 7, 14, 17 (NOT SHOWN) ALSO CONNECT TO V – 1V/DIV 1562 TA16c 10µs/DIV fC (Hz) RIN1 = RIN3 R21 = R23 RQ1 = RQ3 RIN2 = RIN4 R22 = R24 RQ2 = RQ4 16k 105k 105k 34k 340k 340k 34k 32k 26.1k 26.1k 16.9k 84.5k 84.5k 16.9k 64k 8.45k 6.49k 8.45k 16.2k 21k 8.45k RELATED PARTS PART NUMBER DESCRIPTION COMMENTS LTC1068, LTC1068-X Quad 2-Pole Switched Capacitor Building Block Family Clock-Tuned LTC1560-1 5-Pole Elliptic Lowpass, fC = 1MHz/0.5MHz No External Components, SO8 LTC1562-2 Quad 2-Pole Active RC, 20kHz to 300kHz Same Pinout as the LTC1562 28 Linear Technology Corporation 1562f LT/TP 0199 4K • PRINTED IN USA 1630 McCarthy Blvd., Milpitas, CA 95035-7417 (408)432-1900 ● FAX: (408) 434-0507 ● www.linear-tech.com LINEAR TECHNOLOGY CORPORATION 1998