Instrumentation Amplifier Application Note ® Application Note May 27, 2009 AN1298.2 Table of Contents Introduction to the Instrumentation Amplifier................................................................................................................................... 2 Review of Standard Instrumentation Amplifier Design Techniques ................................................................................................ 2 Monolithic Instrumentation Amplifier Architecture ........................................................................................................................... 4 Introduction to Instrumentation Amplifier Product Family................................................................................................................ 4 Instrumentation Amplifier Specifications ......................................................................................................................................... 4 Instrumentation Amplifier Product Family Theory of Operation....................................................................................................... 6 Features of Instrumentation Amplifier Product Family .................................................................................................................... 7 Care and Feeding of Instrumentation Amplifiers ............................................................................................................................. 10 Application Circuits.......................................................................................................................................................................... 20 Pressure Sensor Interface Circuit ................................................................................................................................................... 21 Thermocouple Input with A/D Converter Output ............................................................................................................................. 22 Thermocouple Input with 4mA to 20mA Output Current ................................................................................................................. 23 RTD Input with A/D Converter Output ............................................................................................................................................. 24 Low Voltage High Side Current Sense............................................................................................................................................ 27 Multiplexed Low Voltage Current Sense ......................................................................................................................................... 30 Bi-Directional Current Sense........................................................................................................................................................... 32 1 CAUTION: These devices are sensitive to electrostatic discharge; follow proper IC Handling Procedures. 1-888-INTERSIL or 1-888-468-3774 | Intersil (and design) is a registered trademark of Intersil Americas Inc. Copyright Intersil Americas Inc. 2007, 2009. All Rights Reserved All other trademarks mentioned are the property of their respective owners. Application Note 1298 Introduction to the Instrumentation Amplifier An Instrumentation Amplifier is a confused animal – confused by its cousin, the op amp. This Application Note describes the Intersil bipolar and MOS input (see Table 1). Instrumentation Amplifiers, theory of operation, advantages, and typical application circuits. These devices are micropower Instrumentation Amplifiers which deliver rail-to-rail input amplification and rail-to-rail output swing on a single 2.4V to 5V supply. These Instrumentation Amplifiers deliver excellent DC and AC specifications while consuming only 60µA typical supply current. Because they provide an independent pair of feedback terminals to set the gain and to adjust output level, these Instrumentation Amplifiers achieve high common-mode rejection ratios regardless of the tolerance of the gain setting resistors. The ISL28271 and ISL28272 have an ENABLE pin to reduce power consumption, typically less than 5.0µA, while the Instrumentation Amplifier is disabled. • Its symbol looks like an op amp (see Figure 1) • It has many of the same basic properties and specifications as an op amp Offset Voltage, Input Bias Current, CMRR, PSRR, etc. • You can make an Instrumentation Amplifier from a simple op amp circuit. But the behavior of an Instrumentation Amplifier is profoundly different than an op amp! And it is very difficult to make a precision Instrumentation Amplifier from a simple op amp circuit – many have tried, but most have failed. An Instrumentation Amplifier provides a voltage subtraction block followed by a fixed gain block; i.e. V OUT = ( IN+ – IN- ) × Gain TABLE 1. Often, there is an optional output reference input which allows the output voltage to be shifted by a fixed voltage: MINIMUM CLOSED BW INPUT # OF STAGE AMPLIFIERS LOOP GAIN (kHz) ENABLE? PART EL8170 Bipolar 1 100 192 No EL8171 PMOS 1 10 450 No EL8172 PMOS 1 100 170 No EL8173 Bipolar 1 10 396 No ISL28270 Bipolar 2 100 240 No ISL28271 PMOS 2 10 180 Yes ISL28272 PMOS 2 100 100 Yes ISL28273 Bipolar 2 10 230 No ISL28470 Bipolar 2 100 240 No VOUT - VOUT GAIN In contrast, an op amp by definition only provides extremely high gain with provisions to apply negative feedback to establish a fixed gain or unique transfer function, H(s), such as an integrator or filter. Review of Standard Instrumentation Amplifier Design Techniques Difference Amplifier In its most basic topology, an Instrumentation Amplifier can be configured from a single op amp and four resistors as shown in Figure 4; this is often referred to as a Difference Amplifier. -VCC -VCC + IN- FIGURE 3. VOUT - IN+ (EQ. 2) VREF + + - V OUT = ( IN+ – IN- ) × Gain + V REF +VCC +VCC (EQ. 1) INSTRUMENTATION AMPLIFIER OP AMP FIGURE 1. R2 R1 + IN- R2 R1 + VOUT = (IN+ - IN-) * (1 + R2/R1) IN+ FIGURE 2. TWO OP AMP INSTRUMENTATION AMPLIFIER 2 AN1298.2 May 27, 2009 Application Note 1298 and a common voltage of 10V, the inputs to the op amp will be sitting at a voltage of 9.9V. This circuit would not be possible if the op amp was operated with VCC of +5V since the op amp input’s voltage would exceed the supply voltage. R2 R1 IN- - IN+ + VOUT R2 100k R3 R4 Vcm = 10V VREF VCC R1 1k - VOUT + R3 1k FIGURE 4. R4 100k In this configuration, the gain is set by resistors R1 and R2: Gain = R 2 ⁄ R 1 (EQ. 3) VREF V OUT = ( IN+ – IN- ) × Gain + V REF FIGURE 5. (EQ. 4) For the ability to reject a voltage that appears on both INand IN+ (i.e., common mode voltage), resistor values must match such that R1 = R3 and R2 = R4. The common mode rejection ratio (CMRR) is set by the matching ratio of R1:R3 and R2:R4. High common mode rejection ratio requires a very high degree of ratio matching. Two Amplifier Instrumentation Amplifier To provide a high input impedance, a two amplifier Instrumentation Amplifier can be used as in Figure 6. R2 It can be shown that the CMRR is: - CMRR = 20 × log 10 (x) (EQ. 5) IN- Where x = R 4 ⁄ ( R 3 + R 4 ) × ( R 1 + R 2 ) ⁄ R 1 – R 2 ⁄ R 1 (EQ. 6) IN+ Worse case CMRR occurs when the tolerance of R4 and R1 are at their maximum, and R2 and R3 are at their minimum value. The following table shows the relationship between resistor tolerance and CMRR for gains of 1, 10, and 100. TABLE 2. RESISTOR R1 CMRR TOLERANCE GAIN =1 GAIN = 10 GAIN =100 ±5% -20.4dB -15.6dB -14.8dB ±1% -34.1dB -28.9dB -28.1dB ±0.1% -54.0dB -48.8dB -48.0dB ±0.01% -74.0dB -68.8dB -68.0dB The Difference Amplifier has the advantage of simplicity and the ability to operate with high common mode voltage on its inputs, IN+ and IN-. However, the input resistance is set by the resistor values R3 and R4, and does not provide high input resistance as is common in most Instrumentation Amplifier circuits. + R4 R3 - VOUT + FIGURE 6. TWO AMPLIFIER INSTRUMENTATION AMPLIFIER In this configuration, the gain is set by resistors R3 and R4: Gain = 1 + R 4 ⁄ R 3 (EQ. 7) V OUT = ( IN+ – IN- ) × Gain (EQ. 8) The ability to reject a voltage that appears on both IN- and IN+ (i.e., common mode voltage), depends on matched resistor values such that, R1 = R3 and R2 = R4. The common mode rejection ratio (CMRR) is set by the matching ratio of R1:R3 and R2:R4, and, high CMRR requires a very high degree of ratio matching. For example, with 10V of common mode voltage, resistor tolerance’s must be at least ±0.01% to achieve 12-bit accuracy (72dB). Classic Three Amplifier Instrumentation Amplifier By adding a third op amp, the “Classic Three Amplifier Instrumentation Amplifier” can be configured as shown in Figure 7. Additionally, the REF input must be driven by a very low source impedance since the CMRR will be degraded by any source resistance that contributes to the value of R4 and causes increased mismatch between R2 and R4. Also note that the common mode voltage will bias internal nodes at a voltage that is set by the ratio of R3 and R4, or the gain of the circuit. For example, in Figure 5, for a gain of 100 3 AN1298.2 May 27, 2009 Application Note 1298 IN- R1 + Introduction to Instrumentation Amplifier Product Family R2 - V2 R5 - Rg VOUT VIN V1 + R6 - IN+ INVOUT VOUT V4 R3 IN+ R4 VREF + FIGURE 7. CLASSIC THREE AMPLIFIER INSTRUMENTATION AMPLIFIER V3 FB+ FB- Rf Rg Usually, resistors R1 through R6 are equal value resistors of R and the gain: Gain = ( 1 + 2 × R ⁄ R gain ) V OUT = ( IN+ – IN- ) × Gain + V REF (EQ. 9) FIGURE 8. TWO AMPLIFIER INSTRUMENTATION AMPLIFIER (EQ. 10) This Application Note describes the Intersil Instrumentation Amplifier Product Family, which includes the following features: With this circuit, the Gain can be set with a single resistor, RGAIN and the input impedance is very high. However, the common mode rejection ratio, CMRR, just like the Difference Amplifier topology, is still set by the resistor matching between R1, R2, R3, and R4. Extremely low tolerance resistors or precision resistor trimming is required to achieve high CMRR. The equations and Table shown for the Difference Amplifier apply directly to the Classic Three Amplifier Instrumentation Amplifier configuration. Monolithic Instrumentation Amplifier Architecture Each of the three basic Instrumentation Amplifier architectures that have been already discussed have been implemented in standard integrated circuit packages. To achieve a high CMRR, extensive resistor trimming is required with lasers or other suitable techniques. While each of these devices provide adequate specifications for a precision Instrumentation Amplifier, each device has its own compromise based on operating voltage range, supply current, common mode operating range, input impedance, etc. These instrumentation amplifiers use one external resistor to set the gain; while this may seem to be an advantage, there are considerations which make the single resistor configuration undesirable from a design viewpoint. The temperature coefficient (TC) of the external resistor will be a direct gain drift. Also, an external filter can not be applied to the feedback network because it is internal to the device. 4 1. Bipolar transistor inputs for low voltage noise 2. PMOS transistor inputs for low input bias current 3. Micropower operation requiring only 60μA supply current 4. Rail-to-rail inputs and rail-to-rail output swing 5. Single supply operation from 2.4V to 5V supply 6. An independent pair of feedback terminals to set the gain and to adjust output level allow these Instrumentation Amplifier to achieve high CMRR (>104dB) regardless of the tolerance of the gain setting resistors. 7. Internal loop compensation to provide optimum bandwidth trade-off as shown in Table 1 8. The ISL28271 and ISL28272 have an ENABLE pin to reduce the supply current to a typical of less than 5µA and tri-state the output stage to a high impedance state. Instrumentation Amplifier Specifications Many of the Instrumentation Amplifier specifications are very similar to the standard specifications for operational amplifiers. However, the unique architecture of the Intersil Instrumentation Amplifiers make some of these specifications differ slightly. Table 3 summarizes the Specifications and Features of the Instrumentation Amplifier Product Family. AN1298.2 May 27, 2009 Application Note 1298 TABLE 3. PARAMETERS EL8170 Input Stage Minimum Gain Gain Set Supply Current: Enabled per Channel Supply Current: Shutdown - ISL28270 ISL28470 EL8173 ISL28273 EL8171 ISL28271 EL8172 ISL28272 UNITS Bipolar Bipolar PMOS PMOS 100 10 10 100 2 Ext R 2 Ext R 2 Ext R 2 Ext R 65 65 65 60 65 60 µA - - 4 - 4 µA - - Minimum VCC 2.4 2.4 2.4 2.4 VDC Maximum VCC 5.5 5.5 5.5 5.5 VDC Input Offset Voltage 200 150 150 1000 600 1500 600 300 500 μV Offset Drift 0.24 0.7 0.7 2.5 0.7 1.5 0.7 0.14 0.7 µV/°C Input Bias Current, Maximum 3000 2000 2500 2000 2500 50 30 50 30 pA 25 30 25 30 pA Input Offset Current, Maximum 2000 2000 Input Bias Current Cancellation Yes Yes Bandwidth (-3dB) at AV = 10 - 396 Bandwidth (-3dB) at AV = 100 192 240 240 Slew Rate (Typ) 0.55 0.5 0.5 265 450 0.55 180 0.6 0.55 0.5 170 100 kHz 0.55 0.5 V/µs Rail-to-Rail Input Yes Yes Yes Yes Rail-to-Rail Output Yes Yes Yes Yes ±26 ±26 Output Current Limit, V+ = 5V ±26 Output in Shutdown Mode Gain Accuracy ±29 ±29 ±26 - ±29 - kHz mA - HiZ - HiZ ±0.15 0.08 ±0.2 -0.19 ±0.35 ±0.5 ±0.5 ±0.1 ±0.12 CMRR (Typ) 114 110 110 106 110 PSRR (Typ) 106 110 110 90 95 90 100 eN at 1kHz 58 60 60 220 210 220 240 80 78 nv/√Hz 3.6 3.5 14 10 10 6 µVP-P 100 100 dB 100 dB eN 0.1Hz to 10Hz 3.5 Input Protection - Diodes to Rails Yes Yes Yes Yes Input Protection - Diodes across Inputs Yes No No No Max Input Diode Current 5 5 5 5 SO8 SO8 SO8 SO8 -40 to +85 -40 to +85 -40 to +85 -40 to +85 Yes Yes Yes Yes Package Operating Temp. Range RoHS Compliant 5 % mA °C AN1298.2 May 27, 2009 Application Note 1298 +Ven I I I Re Va V1 IN- Re Vb Ix1 Q1 Ix2 Q2 IN+ V2 V3 I2 I1 I Q3 FB+ I3 Q4 V4 FB- I4 V5 V6 VOUT I5 I6 Ry GAIN = A Ry FIGURE 9. SIMPLIFIED SCHEMATIC Instrumentation Amplifier Product Family Theory of Operation Each of the features specifications of the Intersil Instrumentation Amplifier Product Family will be discussed in more detail in a future section of this Application Note, but first, let’s study the internal operation of this unique Instrumentation Amplifier Product Family. V5 = I5 × Ry = 2 × Ry × I + ( V1 – V2 ) × Ry ⁄ Re + ( V4 – V3 ) × Ry ⁄ Re (EQ. 20) V6 = I6 × Ry = 2 × Ry × I + ( V2 – V1 ) × Ry ⁄ Re + ( V3 – V4 ) × Ry ⁄ Re (EQ. 21) V OUT = A × ( V 5 – V 6 ) (EQ. 22) where A is the gain of the output stage A simplified schematic is shown in Figure 9. ( V 2 + V be2 ) – ( V 1 + V be1 ) I x1 = ----------------------------------------------------------------------, and since V be1 = V be2 Re V2 – V1 I x1 = -------------------(EQ. 11) Re Assume Ry/Re = 1 (i.e., Re and Ry are equal value). V OUT = A × [ 2 × R y × I + ( V 1 – V 2 ) + ( V 4 – V 3 ) – [ 2 × R y × I + ( V 2 – V 1 ) + ( V 3 – V 4 ) ]] (EQ. 23) V OUT = A × [ ( V 1 – V 2 ) + ( V 4 – V 3 ) + ( V 1 – V 2 ) + ( V 4 – V 3 ) ] Assuming β high transistors: (EQ. 24) I 1 = I + I x1 = I + ( V 2 – V 1 ) ⁄ R e (EQ. 12) V OUT = 2 × A × [ ( V 1 – V 2 ) + ( V 4 – V 3 ) ] (EQ. 25) V OUT ⁄ ( 2 × A ) = [ ( V 1 – V 2 ) + ( V 4 – V 3 ) ] (EQ. 26) Since A is very large: I 2 = I – I x1 = I – ( V 2 – V 1 ) ⁄ R e V OUT ⁄ ( 2 × A ) ⇒ 0 (EQ. 27) Similarly for Q3 and Q4: 0 = ( V1 – V2 ) + ( V4 – V3 ) (EQ. 28) I 3 = I + I x2 = I + ( V 4 – V 3 ) ⁄ R e (EQ. 14) Let VIN = V2 – V1, and V3 = FB+, V4 = FB- I 4 = I – I x2 = I – ( V 4 – V 3 ) ⁄ R e (EQ. 15) (EQ. 13) Summing currents: 0 = -V IN + ( FB- – FB+ ) (EQ. 29) V IN + FB- – FB+ (EQ. 30) or I5 = I2 + I3 = I – ( V2 – V1 ) ⁄ Re + I + ( V4 – V3 ) ⁄ Re (EQ. 16) I5 = 2 × I + ( V1 – V2 ) ⁄ Re + ( V4 – V3 ) ⁄ Re (EQ. 17) I6 = I1 + I4 = I + ( V2 – V1 ) ⁄ Re + I – ( V4 – V3 ) ⁄ Re (EQ. 18) I6 = 2 × I + ( V2 – V1 ) ⁄ Re + ( V3 – V4 ) ⁄ Re (EQ. 19) 6 IN+ – IN- = FB- – FB+ (EQ. 31) As you can see from Equation 31, negative feedback is applied around the amplifier so that the voltage applied to the feedback terminals (FB+ - FB-) must be equal to the voltage applied to the input terminals (IN+ - IN-). AN1298.2 May 27, 2009 Application Note 1298 For the standard data sheet connection: V2 VIN The input terminals (IN+ and IN-) and feedback terminals (FB+ and FB-) are single differential pair devices aided by an Input Range Enhancement Circuit to increase the headroom of operation of the common-mode input voltage. As a result, the input common-mode voltage range for all these Instrumentation Amplifiers is rail-to-rail. The parts are able to handle input voltages that are at or slightly beyond the supply and ground making these in-amps well suited for single 5V or 3.3V low voltage supply systems. There is no need then to move the common-mode input voltage of the these Instrumentation Amplifiers to achieve symmetrical input voltage. IN+ V1 INVOUT VOUT V4 FB+ V3 Rf FB- The use of a bipolar transistor input stage vs. the MOSFET input stage allows the user to choose low bias current, high input resistance. Rg Rail-to-rail operation for both the inputs and outputs is an important and unique feature. The rail-to-rail inputs allow the input voltages to be slightly below the VS- rail (typically Ground) to slightly above the VS+ rail. FIGURE 10. TWO AMPLIFIER INSTRUMENTATION AMPLIFIER FB+ = 0V FB- = V OUT × R g ⁄ ( R g + R f ) The conventional technique to achieve a rail-to-rail input stage is to use two separate input stages, as shown in Figure 12. One input stage (Q1 and Q2) provides common mode input range to the top rail (VS+), and the other input stage (Q3 and Q4) provides common mode input range to the bottom rail. V IN = FB- – FB+ V IN = V OUT × R g ⁄ ( R g + R f ) – 0 V OUT = V IN × ( 1 + R f ⁄ R g ) (EQ. 32) Features of Instrumentation Amplifier Product Family A simplified schematic and block diagram is shown in Figure 11 to illustrate the rail-to-rail operation for both the input stage and the output stage. The same schematic applies to the PMOS input devices when the PNP transistors (Q1 to Q4) are replaced with P-Channel MOSFETs for ultra-low input bias current. VS+ INPUT RANGE ENHANCEMENT CIRCUIT Ven = VS+ + 2V I I I Re Vb Va Q1 IN- I Re Q2 IBC IN+ FB+ IBC Q3 FB- Q4 IBC IBC Q5 P-Channel OUT Ry Ry Q6 N-Channel VSIBC => INPUT BIAS CURRENT CANCELLATION FIGURE 11. SIMPLIFIED SCHEMATIC 7 AN1298.2 May 27, 2009 Application Note 1298 VS+ I Q3 Q4 TRANSISTION CIRCUIT +IN Q1 -IN TO OUTPUT STAGE INPUT OFFSET VOLTAGE (µV) 250 VDD = 5.5V 200 -40°C 150 100 +25°C 50 0 -50 -100 +85°C -150 -200 -250 -0.5 Q2 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 0 COMMON-MODE INPUT VOLTAGE (V) I FIGURE 14. TYPICAL RAIL-TO-RAIL INPUT AMPLIFIER FIGURE 12. 2 AMPLIFIER INSTRUMENTATION AMPLIFIER Unless the input stages transistors are exactly matched, changes in offset voltage and input bias current will result as the common mode input range transitions between the two input stages. In contrast, the Product Family uses a single input stage for the IN inputs and a single input stage for the FB inputs. An Input Range Enhancement Circuit (IREC) provides a bias voltage that is approximately 2V above the VS+ rail which is used to bias the I current sources shown in the Block Diagram. Since there is a single input stage, there is no input stage transition point to create shifts in offset voltage and bias current as the input common mode voltage changes. The effectiveness of the Single Input Stage and IREC circuit technique is evident as shown in the following Figures for the offset voltage of a EL8170 (Figure 13) and a typical rail-torail input amplifier (Figure 14). In addition to shifts in offset voltage as the input common mode voltage changes, the input bias current will change dramatically as the input stages transition from a PNP transistor input stage to a NPN transistor input stage. The following graphs compare the input bias current over the common mode input range for the EL8170 (Figure 15) and a typical rail-to-rail input amplifier (Figure 16). AVERAGE INPUT BIAS CURRENT (pA) VS- 1500 1000 VS = 3.3V VS = 5.0V 500 0 VS = 2.9V -500 -0.5 0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 COMMON-MODE INPUT VOLTAGE (V) 250 200 +85°C INPUT BIAS CURRENT vs COMMON-MODE INPUT VOLTAGE 3 150 +25°C 100 50 -45°C 0 -0.5 0 0.5 1.0 1.5 2.0 2.5 COMMON-MODE INPUT VOLTAGE (V) FIGURE 13. EL8170 3.0 3.5 IIB - INPUT BIAS CURRENT - (nA) INPUT OFFSET VOLTAGE (µV) FIGURE 15. EL8170 2 VDD = 5V TA = +25°C 1 0 -1 -2 -3 -4 -0.5 0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 VIC - COMMON-MODE INPUT VOLTAGE (V) FIGURE 16. TYPICAL RAIL-TO-RAIL INPUT AMPLIFIER 8 AN1298.2 May 27, 2009 Application Note 1298 The PNP input stage transistors are biased with an adequate amount of current for speed, and consequently, their base current increases. In order to keep the input bias current low, an Input Bias Current Cancellation Circuit is used to apply and equal but opposite compensation current to the inputs. This compensation current subtracts from the base currents, and the resulting input bias current is reduced to typically around 500pA. This is shown in Figure 17 for the IN+ and INinputs, where the FB+ and FB- are identical for proper matching between stages. The compensation current, (Icomp) is derived from the IBC circuit and is equal to the base current of Q1 and Q2. Ven = VS+ + 2V Re Vb Va Q1 IN- Q2 IN+ Icomp Icomp Q7 Q8 IBC INPUT BIAS CURRENT CANCELLATION VS- FIGURE 17. INPUT BIAS CURRENT CANCELLATION CIRCUIT Since the feedback terminals are differential inputs, they can be used in applications such as current sources for a true Kelvin sense of the feedback voltage. In addition, a complex network can be placed in the feedback path for frequency shaping and filter circuits. The basic Instrumentation Amplifier configuration is shown in Figure 19: Input Bias Current Cancellation Circuit is typically active from 10mV above the negative rail (VS-) up to the positive rail (VS+). V2 Not only does the Input Bias Current Cancellation compensation circuit keep the input bias current very small, it also maintains a very small input bias current variation over a wide operating range as shown in Figure 18 for +25°C to +85°C. AVERAGE INPUT BIAS CURRENT (pA) Another unique feature built into the ISL28271 and ISL28272 is the ability to tri-state the output stage to a high impedance state when the part is disabled via the ENABLE pin. This allows several outputs to be wired together for a multiplexer function. This feature will be shown in the Applications section. Because the Instrumentation Amplifier product family provides an independent pair of feedback terminals to set the gain and to adjust output level, these Instrumentation Amplifiers achieve high CMRR regardless of the tolerance of the gain setting resistors. The FB+ pin can be used as a REF terminal to center or to adjust the output voltage. Because the FB+ pin is a high impedance input, an economical resistor divider can be used to set the voltage at the REF terminal without degrading or affecting the CMRR performance. Any voltage applied to the REF terminal will shift the output voltage by VREF times the closed loop gain, which is set by resistors RF and RG. I I The operating voltage range of the Instrumentation Amplifier product family is from 2.4V to 5.5V making it ideally suited for operation on 3.3V or 5V power supplies. Also, it will operate with a single 4.2 lithium ion battery. Additionally, they are well suited for battery operation since the supply current is only 66µA maximum. 1500 VIN V1 INVOUT VOUT V4 V3 VS = 3.3V IN+ FB+ FB- Rf 1000 500 +85°C 0 +25°C Rg FIGURE 19. TYPICAL RAIL-TO-RAIL INSTRUMENTATION AMPLIFIER -500 -1000 -0.5 0 0.5 1.0 1.5 2.0 2.5 COMMON-MODE INPUT VOLTAGE (V) FIGURE 18. 9 3.0 3.5 The gain of this circuit is set by the ratio of Rf and Rg such that: V OUT = V IN × ( 1 + R f ⁄ R g ) (EQ. 33) AN1298.2 May 27, 2009 Application Note 1298 In this configuration, adjustable gain is possible with external resistors for gains from unity up to 10,000. Two external gain setting resistors are used to minimize temperature coefficient (TC) mismatch as is common with a single gain setting resistor. Notice that resistor value mismatches only effect the gain, and CMRR is not degraded by resistor mismatches as is the case with the other basic Instrumentation Amplifier configurations discussed previously. In this case: V REF = V CC × R 2 ⁄ ( R 1 + R 2 ) (EQ. 37) The feedback terminals can also be used to apply a reference voltage to shift the output voltage as shown in Figure 22 with Rg connected to VREF instead of ground. IN+ VIN INVOUT VOUT The feedback terminals can be used to apply a reference voltage to shift the input voltage. These are a high impedance reference input that is not affected by gain. The basic circuit is shown in Figure 20: VREF FB+ FB- Rf IN+ VIN Rg INVOUT VOUT VREF FB+ FIGURE 22. Rf FB- V IN = FB- – FB+ FB- = V REF + R g × ( V OUT – V REF ) ⁄ ( R f + R g ) V IN = V REF + R g × ( V OUT – V REF ) ⁄ ( R f + R g ) – V REF Rg V OUT = V IN × ( 1 + R f ⁄ R g ) + V REF FIGURE 20. BASIC CIRCUIT If we go back to the equations derived previously: V IN = FB- – FB+ (EQ. 34) V IN = V OUT × R g ⁄ ( R f + R g ) – V REF (EQ. 35) V OUT = ( V IN + V REF ) × ( 1 + R f ⁄ R g ) (EQ. 36) Since the FB+ is a high input impedance, a simple resistor divider could be used to set the VREF voltage as shown in Figure 21. IN+ VIN INVOUT VOUT VCC Rf VREF FB+ Rf FB- Rg Rg (EQ. 38) Since the current in Rg must flow into VREF, the driving point impedance of VREF will effect the accuracy of this configuration. Therefore, VREF should be a low impedance from an op amp, voltage regulator, or voltage reference. Alternately, if a resistor divider is used to obtain VREF, the Thevenin resistance of the divider network must be much lower than the values of Rf and Rg, or the Thevenin resistance must be included in the value of Rg and VREF. However, the CMRR is not affected by the reference voltage or its source resistance. Care and Feeding of Instrumentation Amplifiers As in any low voltage, high accuracy measurement system, extreme care must be taken with any of the Instrumentation Amplifiers with respect to grounding scheme, Kelvin sense connections, guarding and shielding, and interface to the digital world. If the PCB connections are made incorrectly, the most perfect measurement circuit can still have errors resulting from poor grounding considerations and not understanding the impact of Ohm’s Law. Any analog or mixed signal PCB must have a well thought-out grounding scheme with multiple ground planes or traces. There must be no heavy DC current or AC current in the analog ground planes that connect system measurement points. FIGURE 21. 10 AN1298.2 May 27, 2009 Application Note 1298 be sure there is no digital noise introduced into the Analog Ground, the two grounds are tied together at only one point at the A/D Converter. Furthermore, a 0Ω resistor can be used to connect the two grounds; this ensures a separate Net for each ground so the PCB layout software or layout person does not arbitrarily connect the two grounds. The use of a 0Ω resistor is cheap insurance against a noisy and inaccurate analog system! This Thermocouple Circuit will be discussed in more detail in the Applications section. A one point measurement system must be established to prevent high currents from interfering with the basic measurement. This is shown in Figure 23 for interfacing a thermocouple to an A/D Converter. The “High quality measurement Ground” must only connect to the critical ground points in the analog front-end; this ground must make a single point connection to the A/D converter at its Analog Ground pin (AGND). There must be no other connections such as digital grounds or power supply returns to the “High quality measurement Ground” except a single connection at the A/D Converter pins (AGND and DGND). To +5V 7 VS+ +5V R1 R INPUT FILTER J-TYPE THERMOCOUPLE (51.7µV/C) R2 R IBIAS +5V RETURN LM35DM (10m/C) OPEN TC BIAS R4 R 3 VS- 4 IN+ C1 C (Vtc) 2 IN- V OUT 6 (Vcjc) 8 R3 FB+ R 5 VOUT = (Vtc + Vcjc) * (1 + Rf/Rg) Rf 191k, 1% FB- GAIN = 1 + Rf/Rg GAIN = 1 + 191k/1k GAIN = 192 R5 191k, 1% R6 1k, 1% A/D CONVERTER s VIN ISL6007DIB825 Rg 1k, 1% VOUT (2.5V) VREF GND HIGH QUALITY MEASUREMENT GROUND AGND s DGND CONNECT AGND AND DGND AT ONE POINT AT ADC d FIGURE 23. Rs 1.2V DC/DC CONVERTER OUTPUT PROCESSOR LOAD 10A, MAX 0.005Ω 10k 0.1µF +5V 10k 7 VS+ EL8171, EL8173, ISL28273, ISL28271 3 IN+ 2 INVOUT 6 VOUT = 0V to + 2.5V 8 FB+ 5 FB- V S Rf 48.7k, 0.1% 4 GAIN = 50 Rg 1k, 0.1% FIGURE 24. 11 AN1298.2 May 27, 2009 Application Note 1298 Instrumentation Amplifiers can be used in high accuracy current sense applications as shown in the circuit in Figure 24. Notice the use of a Kelvin connection shown on the current sense resistor Rs as indicated by the slanted connections to the resistor. To avoid errors caused by IR drops, the connections must be made directly at the leads of the 0.005Ω current sense resistor. Just 1mΩ of contact resistance or PCB trace resistance will cause a 20% error in the current reading. Guarding and driven guards is a PCB layout technique to reduce errors caused by PCB leakage currents and improve high frequency CMRR. This can be done by surrounding high impedance input leads with traces that are driven by a low source impedance voltage that is equal to the common mode voltage. At any point in a circuit where dissimilar metals come in contact a small thermocouple voltage is developed. Fortunately, the copper lead frame of a surface mount device is the same copper material as PCB etch, and the thermocouple effect is minimized. However, there are many other places where thermocouples can be generated; for example, across a connector finger, across relay contacts, or even across a resistor! Yes, a poorly constructed resistor can show many µV/°C of thermocouple voltage. It has been found that external components (resistors, contacts, sockets, etc.) can create thermocouple voltages that exceed 10µV/°C. It must be recognized that thermocouple voltages are developed by the difference in temperature between the two ends of dissimilar metal junctions, and not the absolute ambient temperature. If both ends of the metal junctions are isothermal (i.e., at the same temperature) there is no thermocouple voltage developed. Therefore, the first rule to avoid thermocouple effects is to eliminate hot spots on a PCB (e.g., linear voltage regulators). If hot spots cannot be avoided, then the two ends of metal junctions must be oriented so they are on isothermal lines on the PCB. The second rule to minimize thermocouple effects is to balance the number of junctions in a loop so that the error voltages are cancelled or become a common mode voltage that is reduced by the CMRR of the op amps in the signal chain. If the number of junctions are not balanced, then it may be necessary to create a junction by adding a series resistor that has no effect on circuit operation but balances the number of junctions. Unknown to most design engineers is the danger of internal clipping when operating an instrumentation amplifier on a single supply. Unfortunately, the internal nodes are invisible to the user and impossible to measure; manufacturer’s data sheets often ignore the issue, or they have obscure “typical characteristics” graphs or misleading paragraphs that attempt to explain the phenomena. Since the Instrumentation Amplifiers operate in a current summing 12 mode as explained in the “Instrumentation Amplifier Product Family Theory of Operation” on page 6 there is no possibility of internal clipping. If we review the classic three amplifier instrumentation amplifier configuration shown in Figure 25, the effect of internal clipping can be clearly shown. VCC INA1 VIN Rg 100 IN+ A2 VO1 + - 10k R1 50k 10k - R2 50k VOUT + 10k + A3 VO2 10k VREF Vcm FIGURE 25. Simple circuit analysis shows that the two internal voltage VO1 and VO2 are: V O1 = V IN × ( 1 + R 1 ⁄ R g ) + V cm (EQ. 39) V O2 = V cm – V IN × R 1 ⁄ R g (EQ. 40) Two clipping conditions will occur if the effects of VIN and Vcm are not considered: 1. VO1 cannot exceed the maximum output voltage for A1 which is the supply voltage (VCC) and the saturation voltage of A1’s output stage. V IN × ( 1 + R 1 ⁄ R g ) + V cm < V CC + V sat (EQ. 41) 2. VO2 cannot go below Ground + the saturation voltage of A2’s output stage. V cm > V IN × R 1 ⁄ R g + V sat (EQ. 42) In reality, this places a such a severe restriction on single supply operation that it makes this circuit almost impossible to use as a general purpose single supply instrumentation amplifier. For example, A2 output stage saturation voltage prevents even 0V of common mode voltage! To overcome this issue, modern monolithic IC instrumentation amplifiers add PNP level shift transistors (Q1 and Q2) to raise the input voltages off Ground as shown in the circuit in Figure 26. AN1298.2 May 27, 2009 Application Note 1298 A1 VO1 + Q1 there is no possibility of internal clipping. As long as the total of the common mode voltage plus the input signal is between 0V and the VS+ supply voltage there will be no internal clipping. The output voltage will swing within its railto-rail output specification of 10mV to either rail for a 100kΩ load. There is no restriction on differential input voltage or common mode voltage provided the output voltage does not exceed its full scale range due to input voltage level, gain, CMRR, and VREF level. - IN- VCC VIN Rg 100 R1 50k R2 50k - Q2 + IN+ VO2 A2 Vcm FIGURE 26. The two internal voltage VO1 and VO2 are now: V O1 = V IN × ( 1 + R 1 ⁄ R g ) + V cm + 0.7V (EQ. 43) V O2 = V cm – V IN × R 1 ⁄ ( R g + 0.7V ) (EQ. 44) Now, the danger of internal clipping situation has been improved for A2 but made worse for A1 since an additional 0.7V has been added to VO1. For example, the maximum common mode voltage is only 1.5V for this instrumentation amplifier operating on a supply voltage of 5V ±5% with a gain of 250 and a 10mV input signal! If you doubt the validity of these statements, check vendor data sheets for analog devices that exhibit these characteristics. All input and feedback terminals of the Instrumentation Amplifiers have internal ESD protection diodes to both positive (VS+) and negative supply (VS-) rails, limiting the input voltage to within one diode drop beyond the supply rails. The EL8170, EL8172, ISL28270 and ISL28470 have additional back-to-back diodes across the input terminals and also across the feedback terminals. If overdriving the inputs is necessary, the external input current must never exceed 5mA. On the other hand, the EL8171, EL8172, ISL28271, ISL28272 and ISL28273 have no diode clamps to limit the differential voltage on the input terminals allowing higher differential input voltages at lower gain applications. It is recommended however, that the input terminals of these devices are not overdriven beyond 1V to avoid offset drift. An external series resistor may be used as an external protection to limit excessive external voltage and current from damaging the inputs. A 20k resistor can be used to protect the inputs against 100V transients on the inputs. If the overvoltage condition is continuous, the 20k resistor must be rated at 1W for adequate power dissipation. Since the Instrumentation Amplifiers operate in a current summing mode as explained in the “Instrumentation Amplifier Product Family Theory of Operation” on page 6, Ven = VS+ + 2V VS+ I I Re VS+ VS+ IN- I I Re VS+ FB- IN+ VS- VS- VS- VSEL8170 ONLY EL8170 ONLY INPUT BIAS CURRENT CANCELLATION 5mA MAXIMUM INTO ANY PROTECTION DIODE! INPUT BIAS CURRENT CANCELLATION VS- FIGURE 27. INPUT PROTECTION DIODES 13 AN1298.2 May 27, 2009 Application Note 1298 20k VIN 20k 3 2 TRANSFORMER COUPLED SOURCE IN+ VIN INVOUT 8 5 3 6 IN+ 2 IN- VOUT VOUT 8 FB+ Rf FB- 6 VOUT FB+ 5 Rf FB- Rg Rg FIGURE 28. FIGURE 30. Input bias current from the IN+ and IN- inputs of the Instrumentation Amplifiers must find a DC path to their home (i.e., Ground). While it seems obvious to the casual user, this is an often ignored principle when designing with an Instrumentation Amplifier, and results in many telephone calls to the Applications Engineer. Many voltage sources do not provide a DC path to Ground such as thermocouples, microphones, transformer coupled circuits, and AC coupled circuits. Without a DC return path, the input bias current will accumulate on any stray capacitance on the inputs until they are clamped to the rails by the protection diodes. The output of the Instrumentation Amplifier will slowly increase or decrease until it saturates into the VS+ or VS- rail. AC COUPLED SOURCE VIN 3 2 47k IN+ INVOUT 6 47k 8 5 VOUT FB+ FB- Rf Rg A DC return path as shown in the following circuits must be supplied to provide a return path for the input bias current. 3 VIN 2 FIGURE 31. IN+ INVOUT 10k 8 5 6 VOUT FB+ Rf FB- Rg An error budget can be calculated by summing the factors which contribute to the output voltage error. Most of the error sources are referred to the input and multiplied by the Gain to get an output voltage error term as shown in the following. Offset voltage: Normally instrumentation amplifiers have two offset voltage specifications - an input offset voltage (VOSI) and output offset voltage (VOSO) specification such that the input offset voltage is multiplied by the gain, and the output offset voltage exhibits unity gain to the output voltage. Therefore, the output voltage error from offset voltage is: V OUT = Gain × V OSI + V OSO FIGURE 29. Due to the unique architecture of the Instrumentation Amplifiers, there is only one offset voltage specification required. The input offset voltage (VOSI) is the amount of voltage applied the inputs terminals such that the voltage across the FB is zero, or input offset voltage will be the difference between the IN terminals and the FB terminals: ( IN+ – IN+ ) = ( FB+ – FB- ) + V OS 14 (EQ. 45) (EQ. 46) AN1298.2 May 27, 2009 Application Note 1298 V IB = R S × I OS (EQ. 48) 0.1Hz to 10Hz Noise: The error introduced by voltage can be modeled the same as an input offset voltage, Vn. If noise is required in a wider bandwidth than 0.1Hz to 10Hz, the noise can be calculated by evaluating the Input Noise Voltage Density (en) over the desired bandwidth. Multiplying the rms noise by six will give a good approximation for the peak-to-peak noise. V OUT ( I OS ) = Gain × R S × I OS (EQ. 49) V OUT ( V n ) = Gain × V n V OUT ( V OSI ) = Gain × V OSI (EQ. 47) Offset bias current: Similar to an Op Amp circuit, the input resistance creates an error source that can be modeled the same as offset voltage such that: Gain Error: Gain error results from two factors. The first is the basic gain deviation from the ideal gain equation, Gain = (1 + Rf/Rg); for the EL8173 this error (E.g.) is typically ±0.2%. Second is the tolerance (ERf and ERg) of the Rf and Rg resistors which set the Gain. Common Mode Rejection Ratio: The error introduced by common mode voltage can be modeled the same as an input offset voltage, VCMR. CMRR = 20 × log 10 V CMR = V CMV × 10 ( V CMR ⁄ V CMV ) (EQ. 50) ( CMRR ⁄ 20 ) V OUT = V IN × ( 1 + R f ⁄ R g ) × [ 1 – ( ER f + ER g + Eg ) ] (EQ. 51) V OUT ( CMRR ) = Gain × V CMR Rs PROCESSOR LOAD 10A, MAX 0.005Ω 10k 0.1µF (EQ. 54) Temperature Drift: The effect of operating over the expected temperature range must be included in all these calculations based on the data sheet specifications. (EQ. 52) 1.2V DC/DC CONVERTER OUTPUT (EQ. 53) +5V 10k EL8171, EL8173, ISL28273, ISL28271 7 VS+ 3 IN+ 2 INVOUT 6 VOUT = 0V to + 2.5V 8 FB+ Rf 48.7k, 0.1% 5 FB- V S 4 GAIN = 50 Rg 1k, 0.1% FIGURE 32. TABLE 4. ERROR BUDGET CALCULATION ERROR SOURCE SPECIFIED VALUE REFEREED TO OUTPUT % FS ERROR Offset voltage 400µV 20mV 0.8% Input Offset Current 0.5nA 0.25mV 0.01% CMRR 104dB 0.24mV 0.01% 0.1Hz to 10Hz Noise 10µV 0.5mV 0.02% Gain Error 0.2% 0.2% Rf, Rg Tolerance 0.1% 0.2% Total Error 15 1.24% AN1298.2 May 27, 2009 Application Note 1298 Example of an Error Budget Calculation Consider the circuit shown in Figure 32 for a CPU core voltage current monitor circuit operating at +25°C. The importance of an Error Budget as shown in Table 4 is that it is shows the overall accuracy which can be expected and which factors are determining the overall accuracy of the circuit. In this circuit, the Offset voltage is the factor which is driving the Total Error; if tighter accuracy is required for the application, the offset term could be removed by hardware calibration with a digital potentiometer or software calibration. The Total Error could be reduced to 0.5% just by decreasing the offset voltage term by a factor of 10. within a few millivolts of the supply rails. At a 100kΩ load, the PMOS sources current and pulls the output up to 4mV below the positive supply, while the NMOS sinks current and pulls the output down to 4mV above the negative supply, or ground in the case of a single supply operation. As the load current is increased, the maximum output voltage will decrease as a result of the voltage drop caused by the sourced load current times the top MOSFET ON resistance. Likewise, the minimum output voltage will increase as a result of the voltage drop caused by the sink load current times FET ON resistance. The current sinking and sourcing capability is internally limited to about 26mA with a 5V supply. Because of the independent pair of feedback terminals provided by Intersil’s Instrumentation Amplifiers, the CMRR is not degraded by any resistor mismatches. Hence, unlike a three op amp and especially a two op amp instrumentation amplifier, the Intersil solution will reduce the cost of external components by allowing the use of 1% or more tolerance resistors without sacrificing CMRR performance. The CMRR will be greater than 100dB regardless of the tolerance of the resistors used. Care must be taken with excessive load capacitance, CLOAD. As shown in the following graphs, excessive load capacitance will cause excessive peaking in the frequency response. The result will be ringing in the output voltage under transient conditions, and potentially oscillations resulting from unstable operation. If the Instrumentation Amplifiers are used in applications where there may be large load capacitance (cable driving, filters, FET gates, etc.), a suitable buffer should be used on the output of the Instrumentation Amplifier. The effects of loading the rail-to-rail output stage must also be considered too since the output stage exhibits an “ON” state resistance. A pair of complementary MOSFET devices with approximately 50Ω ON resistance drives the output VOUT to Noise calculations for the Instrumentation Amplifiers are very similar to those for an op amp circuit. The noise model is shown in the following where the Input Noise Voltage and Input Noise Current noise sources are lumped into the IN+ terminal. 25 30 VS = ±2.5V 25 MAGNITUDE (dB) MAGNITUDE (dB) 20 VS = 15 VS = 10 A = 10 V RL = 1kΩ C = 10pF 5 RL/R = 9.08Ω F G RF = 178kΩ RG = 19.6kΩ 0 100 1k 10k 100k FREQUENCY (Hz) 1M FIGURE 33. EL8171 FREQUENCY RESPONSE vs SUPPLY VOLTAGE 20 CL = 27pF 15 A = 10 10 VV = 5V S RL = 10kΩ 5 RF/RG = 9.08Ω RF = 178kΩ RG = 19.6kΩ 0 100 1k 10k 100k FREQUENCY (Hz) 1M FIGURE 34. EL8171, EL8172 FREQUENCY RESPONSE vs CLOAD 50 30 CL = 47pF CL = 20 CL = 27pF 15 A = 10 10 VV = 5V S RL = 10kΩ R 5 F/RG = 9.08Ω RF = 178kΩ RG = 19.6kΩ 0 100 1k 10k 100k FREQUENCY (Hz) CL = 1M CL = 40 CL = 820pF 35 30 FIGURE 35. EL8171 FREQUENCY RESPONSE vs CLOAD 16 45 MAGNITUDE (dB) 25 MAGNITUDE (dB) CL = CL = 47pF AV = 100 VS = 5V RL = 10kΩ RF/RG = 99.02Ω RF = 221kΩ RG = 2.23kΩ 25 100 1k CL = 390pF 10k 100k FREQUENCY (Hz) 1M FIGURE 36. EL8172 FREQUENCY RESPONSE vs CLOAD AN1298.2 May 27, 2009 Application Note 1298 en(V) en(I1) en(I2) en(Rs) en(Rfg) V V V V V IN+ IN- Rs VOUT VOUT FB+ Rf FB- Rg NOISE MODEL FIGURE 37. Where: en(V) is the Input Noise Voltage over the desired bandwidth en(I1) is the voltage noise generated by the Input Noise Current over the desired bandwidth and the source resistance (Rs): The 1.57 term in the equations is the noise equivalent bandwidth representing a 1st order roll-off equivalent as if there is a brick wall filter at 1.57*Fh. If we have a brick wall filter that cuts off right (infinitely steep) at Fh then this term is 1. e n ( I1 ) = I IN × R s 1st order = 1.57 2nd order = 1.11 3rd order = 1.05 4th order = 1.025 (EQ. 55) en(I2) is the voltage noise generated by the Input Noise Current over the desired bandwidth and the feedback and gain resistors (Rf || Rg): e n ( I2 ) = I IN × R f || R g (EQ. 56) en(Rs) is the thermal noise over the desired bandwidth of Rs. en(Rfg) is the thermal noise over the desired bandwidth of Rf || Rg Rs is the source resistance 2 2 2 2 2 2 e n = ( 1 + R f ⁄ R g ) × ( e n ( V ) + e n ( I1 ) + e n ( I2 ) + e n ( I2 ) + e n ( Rs ) + e n ( Rfg ) ) (EQ. 59) The peak-to-peak output noise is typically 6 times the rms value (rule of thumb). Rf and Rg are the gain setting resistors To calculate rms noise, N over a desired bandwidth: N = N O ( F C × ln ( F h ⁄ F l ) + 1.57 × F h – F l ) To determine the total rms output noise from all the sources, the rms summation is taken multiplied by the gain. en ( P – P ) = 6 × en (EQ. 60) (EQ. 57) eno(pp) = 6 * eno where: NO is the specified noise density in nV/√Hz Fc is the corner frequency Fh is the upper frequency of interest Fl is the lower frequency of interest To calculate resistor thermal noise over a desired bandwidth: Nr = 4kTR × ( 1.57 × F h – F l ) (EQ. 58) where: R is the resistor value k is Boltzman’s Constant, 1.39*10-23 T is temperature in Kelvins Fh is the upper frequency of interest Fl is the lower frequency of interest 17 AN1298.2 May 27, 2009 Application Note 1298 +5V Rb1 1k Rb3 1k +5V R1 7.5k 3 R2 7.5k 2 IN+ 7 VS+ INVOUT Rb2 1k Rb4 1k 8 5 6 VOUT + 0V to 2.5V FB+ FB- VS- Rf 100k 4 Rg 100k FIGURE 38. Example of Noise Calculation: Consider the circuit shown in Figure 38 for a bridge amplifier operating in a 0.5Hz to 100Hz bandwidth with a full scale output voltage of 2.V. eN = 50nv/√Hz From the EL8170 data sheet specifications Fc = 100Hz From the EL8170 data sheet performance curves iN = 0.1pA/Hz From the EL8170 data sheet specifications Fc = 50Hz From the EL8170 data sheet performance curves Rs = 8kΩ Balanced bridge Thevenin resistance + 7.5kΩ resistor Rfg = 990Ω Rf || Rg = 90.9k || 1k e n ( V ) = 50nV ⁄ Hz × 100Hz × L n ( 100Hz ⁄ 0.5Hz ) + 1.57 × 100Hz – 0.5Hz = 0.81 μV, rms (EQ. 61) e n ( I1 ) = 0.1pA ⁄ Hz × 50Hz × L n ( 50Hz ⁄ 0.5Hz ) + 1.57 × 100Hz – 0.5Hz × 8k = 0.12 μV, rms (EQ. 62) e n ( I2 ) = 0.1pAnV ⁄ Hz × 50Hz × L n ( 50Hz ⁄ 0.5Hz ) + 1.57 × 100Hz – 0.5Hz × 990 = 0.0014 μV, rms e n ( Rs ) = (EQ. 63) ( 4 × k × 300°K × 8k × ( 100Hz – 0.5Hz ) ) = 0.12 μV, rms e n ( Rfg ) = (EQ. 64) ( 4 × k × 300°K × 990 × ( 100Hz – 0.5Hz ) ) = 0.04 μV, rms (EQ. 65) 18 AN1298.2 May 27, 2009 Application Note 1298 2 2 2 2 +5V 2 e n = ( 1 + R f ⁄ R g ) × ( e n ( V ) + e n ( I1 ) + e n ( I2 ) + e n ( Rs ) + e n ( Rfg ) ) 2 2 2 2 7 3 2 = ( 1 + 90.9k ⁄ 1k ) × (0.81μV +0.012μV +0.0014μV +0.12μV +0.04μV ) = 100 × 0.82μV (EQ. 66) = 82 μVrms VIN 2 To determine the total rms output noise from all the sources, the rms summation is taken multiplied by the gain. 8 Note that the total output noise is dominated by basic Input Noise Voltage and higher source resistance could be used without degrading the overall error resulting from noise. e n ( P – P ) = 6 × 82μV 5 IN+ VS+ IN- VOUT 6 VOUT FB+ Rf 100k FB- VS- Rf 12k Cx 0.1µF 4 (EQ. 67) = 492 μV P – P GAIN = 100 BELOW 14Hz GAIN = 100 BELOW 140Hz Rg 1k This represents an error of 0.02% for a 2.5V full scale output. A very unique feature of the Intersil Instrumentation Amplifiers is the ability to put a filter circuit in the feedback network to shape the frequency response of the amplifier. This ability is not possible with other monolithic Instrumentation Amplifiers because they use a single resistor at the input stage to set the gain. Adding filter circuits in the feedback network of an Instrumentation Amplifier implemented with discrete components (op amps and resistors) is very difficult because capacitor mismatch will result in very poor high frequency CMRR. GAIN = 100 GAIN = 10 14Hz Flp 140Hz Fhp FIGURE 39. A complex impedance network can be added as shown in the following for a low pass function. The low frequency gain is set by Rf and Rg using the standard equation: For this circuit, it can be shown that low frequency pole and higher frequency zero are: Gain = 1 + Rf ⁄ R g Flp = 1 ⁄ ( 2 × π × C x × ( R f + R x ) ) (EQ. 68) The Instrumentation Amplifiers are not unity gain stable; i.e., they require gains greater than 10 or 100 depending on the device. Therefore, they must never be allowed at unity gain even at high frequencies! If Rx was not included in this circuit, Cx would dominate at high frequencies, and the Instrumentation Amplifier would be unstable and oscillate. Lab tests have shown that Cx >33pF is enough to cause an oscillation. Adding Rx in series with Cx creates a zero in the transfer function so that at higher frequencies Rx parallels Rf so that the new Rf = 10.7k, and the gain at high frequency is 11.7 which is a stable condition. Lab tests have shown that any value can be used for Cx with no oscillations. 19 Pole Frequency = 1 ⁄ ( 2 × π × 0.1μF × ( 100k + 12k ) ) = 14Hz Flp = 1 ⁄ ( 2 × π × C x × R x ) (EQ. 69) Zero Frequency = 1 ⁄ ( 2 × π × 0.1μF × 12k ) = 140Hz (EQ. 70) AN1298.2 May 27, 2009 Application Note 1298 6 7 8 3 1 2 A2 A1 VIN+ SW0 CLOSED V+ 10 +5V SW1 CLOSED 4 INH SW2 CLOSED SW3 CLOSED A2 A1 NO0 NO1 NO2 NO3 Com 9 U1 ISL43640 GND Vcm MEASURE VIN+ - VINMEASURE CMRR TO ZERO MEASURE VOS TO ZERO MEASURE 25mV CAL REFERENCE 5 VIN- R2, 1M A2 A1 NO0 NO1 NO2 NO3 GND 5 6 (VREF) R3 100k R1, 10k REQ = 9900Ω V+ 10 +5V INH 4 WRITE PROTECT U2 ISL43640 VOUT VIN C2 0.01µF C1 10µF R9 309 R5 48.7k R7 66.5k R6 1k -5V U3 ISL6007DIB825 OFFSET CORRECTION (±25mV) VOUT R8 150k, 1% 5 FB- RW 5 50k RL 6 2 +5V GND 1 4 R4, 2k s VOUT 6 8 FB+ 8 VCC 1 WP-L RH 7 GND U5 4 ISL95810 s +5V VS- 4 2 IN- 2 SCL RP1 I2C BUS 3 SDA Com 9 VS+ 7 3 IN+ +5V 6 7 8 3 1 2 U4 EL8173 +5V 8 VCC 7 RH WP-L 1 5 RW 50k 6 RL SCL 2 RP2 SDA 3 GND 4 R10 1.37k, 1% s s WRITE PROTECT I2C Bus U6 ISL95810 PROGRAMMABLE GAIN, 90 TO 110 CALIBRATION REFERENCE (25mV) FIGURE 40. ANALOG FRONT-END CIRCUIT Application Circuits Instrumentation Amplifier With Auto Zero and Auto Gain Calibration The circuit shown in Figure 40 shows an analog front-end circuit with an Auto Zero and Auto Gain Calibration to eliminate the offset voltage and gain errors of the EL8173. It is intended to be part of an overall data acquisition system with an A/D Converter and microprocessor to perform an auto zero/gain software routine. Figure 40 does not include the A/D Converter or processor hardware/software. TABLE 5. A2 A1 SWITCH CLOSED MODE 0 0 SWO Measure the input voltage VIN+ - VIN- 0 1 SW1 Calibrate with external common mode voltage applied 1 0 SW2 Calibrate offset voltage to zero 1 1 SW3 Calibrate gain with 25mV reference voltage applied During the calibration mode, analog switches S1 and S2 connect the inputs of the EL8173 to calibration source voltages of zero volts, an external common mode voltage, or a 25mV reference voltage. Digital potentiometer (D-Pot) U5 applies a programmable offset voltage of ±25mV to the FB+ pin of the EL8173 to adjust the EL8173 output to zero voltages. Digital potentiometer (D-Pot) U6 programs the gain of the EL8173 from 90 to 110 for a 2.5V output with the +25mV reference voltage applied to the inputs. 20 Since the offset calibration voltage is operating at very close to zero voltages (±25mV), the driving point impedance is kept very low (1k, R6) to avoid variations caused by the increasing bias current. The configuration of U5, R5, and R7 is carefully selected so that the D-Pot never is operated at a negative voltage. The ±25mV offset calibration source is obtained by programming U5 with the appropriate digital code in Equation 71. 1 1 5∗ ⎛ --------------------------⎞ – ------⎝R + R ⎠ R 1 pot 2 Vcal = ----------------------------------------------------------1 1 1 ⎛ -------------------------⎞ + ------- + ------⎝R + R ⎠ R 1 pot 2 R3 (EQ. 71) The +25mV reference voltage is obtained with the ISL6007’s 2.5V output voltage divided down by a factor of 100 with R1, R2, and R3. The accuracy of the gain calibration is determined by the accuracy of the ISL6007 and the tolerance of resistors R1 and R2. Therefore, it is recommended to use very low tolerance resistors for R1 and R2, or use a precision resistor divider network. The gain of the EL8173 is programmed by D-Pot, U6 according to Equation 72: R 8 + R 9 + R 10 Gain = --------------------------------------------Code R 10 + --------------- × R g 255 150000 + 309 + 1370 Gain = -------------------------------------------------------Code 1370 + --------------- × 309 255 151.7k Gain = -------------------------------------------------1370 + 1.2 × Code (EQ. 72) AN1298.2 May 27, 2009 Application Note 1298 Another complication of SPPT application is the large temperature dependence of both total bridge resistance and peizosensitivity (the ratio of bridge output to excitation voltage times pressure). Bridge resistance increases with temperature while peizosensitivity decreases. Some SPPT designs (e.g. the Nova Sensor NPC-410 series) carefully equalize these opposite-sign tempcos. The payoff comes when such SPPTs are excited with constant current because the increase with temperature of bridge resistance (and therefore of bridge excitation voltage) then cancels the simultaneous decrease of peizosensitivity. Other nominal gains and gain adjustment range can be made by changing the values of R8, R9, and R10. Pressure Sensor Interface Circuit Programmable Pressure Transducer Circuit The silicon piezoresistive-bridge pressure transducer (SPPT) is a dominant technology in automotive, industrial, medical, and environmental pressure sensor applications. All SPPTs share a similar architecture in which a thin (5µm to 200µm) micro machined silicon diaphragm incorporates an implanted piezoresistive Wheatstone-bridge strain-gauge. Applied pressure bends the diaphragm, imbalances the strain gauge, and thereby produces a differential output signal proportional to the product of pressure times bridge excitation voltage. A 10mV/psi pressure-proportional strain gauge signal is outputted differentially on pins 2 and 4 of the sensor; this signal is superimposed on a common mode voltage of 1.2V from the bridge excitation voltage. The low level differential output voltage is amplified by the EL8173 with a nominal gain set at 50. The high common mode rejection capability of the EL8173 eliminates the common mode output voltage of the bridge. The bridge is biased from a constant current source (Q2) and two digitally controlled potentiometers provide for zero (DPOT1) and full scale (gain) adjustments (DPOT2). SPPTs must be supported by appropriate signal conditioning and calibration circuits. Finite elasticity limits the SPPT diaphragm to relatively small deflections which generate only ±1% modulation of the bridge resistance elements and low signal output levels, creating the need for high gain, low-noise, temperature-stable DC amplification. The signal conditioning circuit must also include stable, high resolution, preferably non-interactive, zero and span trims. The automation of the calibration of the sensor circuit is an enormous benefit in the production environment. In the detailed circuit shown in Figure 41, the U2b and U3 circuit provides for the precision offset adjustment, via DPOT1, of any transducer initial null offset error. To +5V R3 402 R2 402 CURRENT MIRROR FOR LOW HEADROOM FROM +5V SUPPLY Q2 DMMT3906 BRIDGE EXCITATION, 600µA CURRENT SOURCE U2a +5V 1/2 ISL28276 U1 +5V IN OUT GND ISL60002-11 (1.200V) Q1 2N3904 4k 4k U2b 1/2 ISL28276 +5V DPOT VOLTAGE ALWAYS POSITIVE R3 39.2k DPOT1 1/2 X95820 50k, 256 TAPS R6 20k U3 EL8176 R4 54.9k -5V 4k R1 2k Vbridge, MAX = 600µA*6k = 3.6V R5 20k U4 EL8173 IN+ C1 1000pF BUFFERED +Vbridge Zin = Zout = 4k, TYP (2.5k TO 6k) 3 VS+ OUT+ 2 3 OUT- 4 2 IN+ IN- IN- 1 8 5 +5V 4 s IN- 4k 8 VS- 7 VOUT VOUT 0.5V/PSI 6 FB+ FB- R8 115k GAIN = 40 TO 60 PS1 NPC-410 0PSI TO 5PSI (±25mV AT MAX Vbridge = 3.6V) (±10mV AT MAX Vbridge = 1.5V) GAIN CALIBRATION CIRCUIT R7 1k OFFSET CALIBRATION CIRCUIT R9 1k DPOT2 50k, 256 TAPS 1/2 X95820 R10 1.96k -Vbridge FIGURE 41. 21 AN1298.2 May 27, 2009 Application Note 1298 accomplish this, the bridge excitation voltage is programmably attenuated by DPOT1 and applied to the FB+ pin of the EL8173. The range for the zero adjustment voltage is from +25mV to -25mV. The resolution is 200µV and is proportional to the bridge excitation voltage, thus improving the temperature stability of the zero adjustment. The 10mV/psi bridge output signal is amplified by 50x to a convenient 0.5V/psi output level with the EL8173 its feedback and calibration network consisting of R8, R9, R10, and DPOT2. The gain of U3 can be varied from 40 to 60 with a resolution of 0.10. Bridge bias is provided by the constant current circuit (U1, U2a, and Q1) which sets a current in Q1 of 1.2 V/2k = 600µA. A current mirror (Q2, R2, R3) reflects the output current so as to source the 600µA into the top of a grounded bridge (PS1). The net result of the combination of transducer and the EL8173 circuitry is a signal conditioned precision pressure sensor that is compatible (thanks to DPOT1 and DPOT2) with full automation of the calibration process, is very low in total power draw (<2mA), most of which goes to transducer excitation and current mirror circuit. Thermocouple Input with A/D Converter Output Thermocouples are the industry standard temperature sensor for measuring a wide range of temperatures from -250°C to + 2300°C. The four most popular thermocouple types are shown in Table 6; however, any time two dissimilar metals are placed in contact, a thermocouple is created via the Seebeck Effect. Thermocouples present several unique challenges when interfacing them to a real world measurement system. 1. Thermocouples generate a very low output voltage that must be amplified with a high gain amplifier. Each thermocouple type requires a different gain when interfacing to an A/D Converter with a fixed full scale voltage, VFS/VoMAX. 2. Thermocouples do not generate an absolute voltage that is proportional to temperature. Instead, they generate a voltage that is a relative voltage that is the proportional to the temperature difference between the “hot” end and the “cold” end. All thermocouple tables showing output voltage vs temperature are for the “cold” end placed in an ice bath at 0°C. Since it is very impractical to place an ice bath on a PCB, electronic cold junction compensation is used. Each thermocouple type requires a cold junction compensation rate, dVO/dT. 3. The output voltage of a thermocouple is non-linear, and is dependant on the type of thermocouple. Linearization is most often done with diode break-point techniques or via microprocessor software, and is not covered in this Application Note. The circuit shown in Figure 42 uses the unique features of the Intersil EL8173 Instrumentation Amplifier to simplify the Thermocouple interface to a high resolution A/D Converter (U5). A programmable gain digital pot (U3) and programmable temperature sensor (U2) allows digital selection of the four most popular thermocouple types: E, J, K, and T. TABLE 6. POPULAR THERMOCOUPLE TYPES TEMPERATURE RANGE VO at TMIN VO at TMAX dVO/dT 0°C to +50°C TYPE MINIMUM MAXIMUM (mV) (mV) (µV/°C) E -200°C -328°F +900°C +1652°F -8.83 68.79 61.00 J 0°C +32°F +750°C +1382°F 0.00 42.30 51.70 K -200°C -328°F +1250°C +2282°F -5.89 50.64 40.50 T -250°C -328°F +350°C +662°F -5.60 17.82 40.70 22 AN1298.2 May 27, 2009 Application Note 1298 U1 EL8173 E, J, K, T TYPE THERMOCOUPLE 8Hz INPUT FILTER VS+ 7 R1 1k (VTC) R3 (VCJC) 1M R2 1k +5V I2C BUS SLAVE ADDRESS BUS (0101000x) 5 6 3 2 1 R7 4.53k 7 RH 4 s 1.6Hz LPF s ISL21400 OSC1 OSC2 16 13 AVDD DVDD 15 5 RW R5 10.7k 6 RL COLD JUNCTION COMPENSATION R8 1.05k s SCLK 1 CLOCK (VOUT) 12 VINHI SDIO 3 DATA I/O R6 150.0k 11 VINLO SDO 2 DATA OUT +5V 8 VCC WP-L 1 SCL 2 50k 3 SDA SYNC 19 10 VCM 8 VRLO 6 DGND CS 4 DRDY 5 RST 18 SYNC CS DATA READY RESET MODE 20 9 VRHI WRITE PROTECT GND 4 U3 s ISL95810 I2C BUS +5V 2 VIN VOUT GND U4 1 4 ISL21009-25 PROGRAMMABLE GAIN GAIN = 30 TO 150 s +5V 7 AVSS -5V 14 AGND SCL R4 SDA 7 357k VOUT A0 A1 C2 10µF A2 VSS U2 VOUT 8 FB+ 6 5 FB- s 8 VCC +5V s 2 IN- 10Mhz 17 +5V 3 IN+ VS- 4 C1 10µF Y? U5 HI7190 6 C3 0.01µF R9 2k C4 10µF s FIGURE 42. The programmable gain amplifier (U1, U3) provide a gain from 30 to 150 that is programmed via the I2C bus with the digital pot for each of the thermocouple types as shown in Table 7. TABLE 7. THERMOCOUPLE TYPES TC TYPE MAX VOUT GAIN D-POT CODE10 E 68.97mV 36.34 195 J 42.30mV 59.10 094 K 50.64mV 49.37 126 T 17.82mV 140.3 000 Cold junction compensation is provided by a programmable reference/temperature sensor (U2) and resistor divider network R4 and R5 according to the following table with AV = 1 and N register = 0. TABLE 8. COLD JUNCTION COMPENSATION TC TYPE VCJC (µV) M REGISTER E 61.0 0 J 51.7 20 K 40.5 43 T 40.7 43 23 Low pass filters (R1, R2, C1) provide noise filtering with a 8Hz cut-off frequency. R3 is used for a return current path for the EL8173 input bias current. An additional low pass filter (R4, R5, C2) attenuates the ISL21400’s output noise voltage with a 1.6 Hz cut-off frequency. A high resolution (24-bit) Sigma-Delta A/D Converter, HI7190, converts the output of the instrumentation amplifier, EL8173, with a full scale input voltage of 2.5V set by the ISL21009-2.5 voltage reference. Thermocouple Input with 4mA to 20mA Output Current Another output option for a thermocouple input circuit is an industry standard 4mA to 20mA current transmitter. The theory of operation for a 4mA to 20mA current transmitter circuit is described in Intersil Application Note AN177 with Figures 33, 34, and 35; this theory of operation applies to the thermocouple circuit shown in Figure 43, and therefore, will not be repeated. AN1298.2 May 27, 2009 Application Note 1298 +5V J-TYPE THERMOCOUPLE (51.7µV/C) U1 EL8173 8Hz INPUT FILTER R1, 1k R2, 1k D2 BAT54C +5V PROG. 8 VCC I2C BUS U2 ISL21400 (VTC) 2 IN- R3 1M VOUT 6 (VCJC) 8 FB+ +5V R14 R15 SCL SDA C1 10µF 7 VS+ 3 IN+ VS- 4 R4 357k 5 FB- VOUT 7 5 SCL (51.7µV/C) 6 SDA 4 VSS A0 A1 A2 R5 3 2 1 10.7k, 1% SLAVE ADDRESS (010100x) RTN PROG. COLD JUNCTION COMPENSATION GAIN = 58.6 U5 ISL60002BIH325Z-TK LM2936M-5.0 U4 1 VOUT VIN 8 VOUT VIN GND GND C3 C4 0.001µF 10µF 2 3 6 7 C5 4.7µF +5V R8 U3 Q1 499k EL8176 IRLL014N 7 (0V TO 2.5V) 3 6 R9 2 R13 127k R10 10 80.6k 4 R12 R6 100k 57.6k, 1% D1 B140 +Vloop 7VDC TO 30VDC LOOP RESISTOR (INTERNAL "GROUND") R11 100 R7 1k, 1% 1.6Hz LPF CURRENT TRANSMITTER C2 10µF ISL21400 REGISTER VALUES AV = 1, N = 0 J-TYPE M = 20 VCJC = 51.7µV/C FIGURE 43. The circuit uses the unique features of the Intersil EL8173 Instrumentation Amplifier to simplify the Thermocouple interface to a 4mA to 20mA Current Transmitter circuit. Since this is circuit is shown for a single J-type thermocouple, a fixed gain of 58.6 is used so that the output voltage of the EL8173 is +2.5V at the maximum thermocouple temperature. The ISL21400 programmable voltage reference/temperature sensor is used for cold junction compensation. Since the ISL21400 has non-volatile storage of the register values, it can be programmed either prior to PCB assembly or programmed via the I2C bus as shown in this schematic. It must be cautioned that the I2C programming “ground” is not at the same potential as the “Internal Ground” or loop supply ground; therefore, when programming U2, the loop supply power supply or associated grounds must not be connected, or the I2C programming system must be floating off ground. Low pass filters (R1, R2, C1) provide noise filtering with a 8Hz cut-off frequency. R3 is used for a return current path for the EL8173 input bias current. An additional low pass filter (R4, R5, C2) attenuates the ISL21400’s output noise voltage with a 1.6Hz cut-off frequency. Since the 4mA to 20mA loop voltage can be as high as 24VDC, a high voltage linear voltage regulator (U5) is used to generate an internal +5V supply. 24 RTD Input with A/D Converter Output Another popular industry standard temperature sensor is the RTD whose resistance varies with temperature, and is typically specified with a nominal resistance at +25°C. For example, a PT100 RTD has a resistance of 100Ω at 0°C. The shape of the resistance vs temperature curve is described by Equation 73, a second order equation, with a unique alpha value as defined by DIN EN 60751. For a PT100 RTD with alpha = 0.385%/ °C: 2 3 RTD = R 0 ( 1 + A∗ T + B∗ T + C∗ ( T – 100 )∗ T ) (EQ. 73) Where: A = 3.9083 E-3, B = -5.775 E-7, C = -4.183 E-12 below 0°C and zero above 0°C. RTDs are typically biased with 1mA to 5mA to minimize selfheating effects; this low operating current generates very low voltage levels shown in Table 9. TABLE 9. TYPICALLY BIAS RTD’s TEMPERATURE (°C) RTD (Ω) VRTD @ 1mA (mV) -40 84.3 84.3 0 100.0 100.0 +100 138.5 138.5 +200 175.8 175.8 AN1298.2 May 27, 2009 Application Note 1298 In the circuit shown in Figure 44, RTD excitation current is supplied by R1 and R2 operating from +5V. It would appear that this current is not accurate enough for a high precision temperature measurement. And, that is true except for the trick that is played by utilizing the ratiometric mode of operation with the A/D Converter. Since the RTD is often operated a great distance from receiving electronics, the use of differential voltage sensing is used to reduce the errors generated by high mode voltage induced noise. The circuit in Figure 44 shows an RTD interface to a high resolution A/D Converter using an EL8173 Instrumentation Amplifier to differentially sense the RTD output and provide the proper gain for the input to the A/D Converter. Ratiometric mode operation of the A/D Converter eliminates the error introduced by variations in the RTD excitation current. TABLE 10. TEMP. (°C) RTD (Ω) IEXT (mA) VRTD @ 1mA (mV) CODE OUT10 -40 84.3 1.22 84.3 7 778 756 0 100.0 1.22 100.0 9 227 469 +100 138.5 1.21 138.5 12 780 044 +200 175.8 1.20 175.8 16 240 345 IEXT Rw + VOUT RTD Rw - Rw FIGURE 44. R1 2k +5V Y1 R2 2k Rw1 RTD PT100 3-WIRE U1 EL8173 VS+ 7 3 IN+ +5V VS- 4 s 2 IN- (VOUT) VOUT 6 Rw2 +5V 8 FB+ R3 10k 5 FB- Rw3 GAIN = 11 16 OSC2 13 AVDD DVDD 15 +5V 9 VRHI SCLK 1 CLOCK 8 VRLO SDIO 3 DATA I/O 12 VINHI SDO 2 DATA OUT 11 VINLO SYNC 19 14 AGND R4 1k +5V 17 OSC1 10 VCM GAIN = 11 Rw1, Rw2, Rw3 - LEAD RESISTANCE #22 AWG WIRE - 0.0168Ω/Ft -5V 10MHz CS 4 DRDY 5 6 DGND RST 18 7 AVSS MODE 20 SYNC CS DATA READY RESET U5 HI7190 24 BIT SIGMA DELTA A/D CONVERTER R5 10k R6 1k FIGURE 45. 25 AN1298.2 May 27, 2009 Application Note 1298 A/D CONVERTER VCC Code = 2N*(VOUT - 0)/(IEXT*R1) Code = 2N*Gain*IEXT*RTD/(IEXT*R1) REFHI Code = 2N*Gain*RTD/R1 R1 Now, the output code is only dependant on the gain of the EL8173 and value of R1, and the variations of IEXT are cancelled out by the ratiometric operation of the A/D Converter. REFLO Iext Also, there is an error created by the wire resistance from the RTD leads from the RTD to the voltage sensing point. Therefore, RTDs are often connected with 3-wire and 4-wire lead configurations to reduce the effect of wire resistance. By far, the most common configuration is the 3-wire connection, and many general purpose 3-wire RTDs are available. Three different configurations for RTD wiring are summarized in the following. IN+ VOUT Rtd IN- IN+ IN- EL8173 GAIN FIGURE 46. In the simplified circuit shown in Figure 46, IEXT = VCC/(R1 + RTD) and VRTD = IEXT*RTD, VOUT = Gain*IEXT*RTD For the A/D Converter, the digital output code, N 2 × ( IN+ - IN- ) CODE = -------------------------------------------------REF HI – REF LOW (EQ. 74) Where N = Resolution However, even with a 3-wire configuration, there is still an error associated with the voltage drop caused by the wire resistance. The RTD circuit incorporates a technique which provides 4-wire accuracy with a 3-wire RTD, and the effect of wire resistance is eliminated completely. The voltage drop created by the wire resistance, Rw, is multiplied by the same gain as the EL8173, and then the differential input of the A/D Converter (U5) subtracts off the effect of the wire resistance, Rw. REFHi - REFLo = IEXT*R1 Iext Iext Iext Rw Rw Rw RTD PT100 2-WIRE VOUT Rw RTD PT100 3-WIRE VOUT Rw Rw VOUT = Iext*(RTD + 2*Rw) ERROR = Iext*2*Rw VOUT = Iext*(RTD + Rw) ERROR = Iext*Rw FIGURE 47A. 2-WIRE CONNECTION FIGURE 47B. 3-WIRE CONNECTION RTD PT100 4-WIRE Rw VOUT Rw Rw VOUT = Iext*RTD ERROR = 0 FIGURE 47C. 4-WIRE CONNECTION FIGURE 47. 26 AN1298.2 May 27, 2009 Application Note 1298 Rs 0.005Ω 1.2V OUTPUT 10A OUTPUT VOLTAGE REMOTE SENSE AFTER Rs AT POINT OF LOAD Ra FB OR SENSE POWER SUPPLY CIRCUIT Rb 3.3V R1 10k 7 R2 10k EL8173 VS+ 2 IN- Ra AND Rb SET THE OUTPUT VOLTAGE 3 IN+ C1 0.1µF VOUT 6 8 FB+ C2 0.1µF 5 FBVS4 GAIN = 50 VOUT = 0V to +2.5V 0.25V/A Rf 48.7k Rg 1k FIGURE 48. Low Voltage High Side Current Sense Due to the rail-to-rail input stage of the EL8173, high side current sensing is very easy to implement, as shown in Figure 48. This circuit is appropriate for any power supply circuit with or without remote sense capability. The output current is measured by a current sense resistor, Rs that is scaled for the desired output voltage and resistor power rating. R1, R2, C1, and C2 are a simple low pass filter to attenuate the power supply output ripple and noise. Resistors Rf and Rg set the gain of the EL8173 for the desired full scale output voltage. Rf V OUT = I OUT × R S × 1 + ------R (EQ. 75) g In this circuit, Equation 76 shows a full scale voltage of 2.5V. 1 + 48.7K V OUT = I OUT × 0.005 × ⎛ -------------------------⎞ ⎝ ⎠ 1K V OUT = 0.25 × I OUT (EQ. 76) An accurate output voltage is obtained since remote sense is used by connecting Ra after the sense resistor, Rs. If remote sense is not possible, care should be exercised to minimize the voltage drop across Rs. The previous circuit uses an external sense resistor to monitor the output current. If the DC/DC is a buck converter which uses an internal controller with current mode control, 27 there is often a current sense resistor used to monitor the inductor current. If this is the case, the output current can be measured with that current sense resistor since the average value of the inductor current is equal to the load current in a buck regulator. The circuit shown in Figure 49 is an example of using the current sense resistor, Rs, that is already a part of the current mode control loop to sense load current. There is a ripple voltage across Rs that is the inductor ripple current * Rs. The inductor ripple current is usually 30% to 50% of the load current and is set by the switching frequency and inductor value. This ripple voltage is essential for the current mode control loop, but must be filtered to obtain the output load current; this filter is performed by R1, R2, C1, and C2. Notice that even though the input voltage of the DC/DC converter is +12V, the common mode voltage that is applied to the EL8173 is the output voltage; in this case, 1.2V. As long as the output voltage is <5V, the input voltage can be much higher. If the inductor current is sensed by using the switching FET’s ON resistance as a current sense element, this method is not possible with the EL8173 circuit. The output current can also be sensed by using the inductor’s DCR (DC Resistance) as a current sense element as shown in the following circuit. This example in Figure 50 shows using the EL7566, but this method applies to any buck mode switching regulator. AN1298.2 May 27, 2009 Application Note 1298 VIN = 12V Q1 TG CURRENT MODE PWM CONTROLLER Rs 0.005Ω L1 1.2V OUTPUT 10A Cout Q2 BG +IS -IS 3.3V R1 10k 7 R2 10k EL8173 VS+ 2 IN3 IN+ C2 0.1µF C1 0.1µF VOUT 6 8 FB+ VOUT = 0V TO +2.5V 0.25V/A Rf 48.7k 5 FBVS4 GAIN = 50 Rg 1k FIGURE 49. P/O EL7566 LX +5V L1 COILCRAFT, DO3316P-272HC 2.7µH DCR = 12mΩ 2.5V OUTPUT 6A C5 150µF EL7566 DEMO BOARD +5V R1 10k R2 10k 7 VS+ EL8173 2 IN3 IN+ C1 0.1µF C2 0.1µF VOUT 6 VOUT = 0V to +3.0V 0.5V/A 8 FB+ 5 FBVS- Rf 40.2k 4 GAIN = 41.7 Rg 1k FIGURE 50. 28 AN1298.2 May 27, 2009 Application Note 1298 sensing the output current. Due to the high currents, the use of current sense resistors become very impractical due to their low values to minimize their power dissipation. For example, with a 50A output current, the current sense resistor must be <400µΩ to keep the power dissipation <1W. In addition, it is very difficult from a PCB layout viewpoint to break a high current power plane to insert a current sense resistor, and that forces the plane to neck-down to a very narrow current flow path. In the circuit shown in Figure 50, the R1 and C1 filter is extremely important because it remove the voltage square wave (swinging between the input voltage and ground) that is applied to the inductor. The values for R1 and C1 should be selected to attenuate the signal to a level that is appropriate for the VOUT noise that is acceptable. Since the DCR of the inductor is being used as a current sense resistor, there are several factors which degrade the accuracy of this approach. First, most inductors are specified for only maximum DCR; for example, the Coilcraft DO3316P-27HC shown above is specified for a DCR of 12mΩ, maximum. Actual lab measurements should the DCR to be 9mΩ. Vishay offers a product line of inductors with a specified tolerance on the inductor DCR; for example, IHLP2525CZ-07 product family guarantees a DCR with ±5% tolerance. High current DC/DC converter outputs now take advantage of advances in multi-phase DC/DC converters where a multiple lower current DC/DC conversion stages are connected in parallel to obtain the necessary high output current. Rule of thumb operates each phase at 15A to 20A so that for a 40A output current either 2- or 3-phases can be operated in parallel. Second, the inductor’s internal winding’s have a temperature coefficient of +0.393%/°C (copper wire) which can be a large error source if the inductor is allowed to get hot from ambient temperature or self-heating due to core losses and DCR power loss. If the error from this source is critical to the application, a thermistor could be mounted in close proximity to the inductor and be used to compensate the temperature coefficient of the copper windings. With multi-phase DC/DC converters the output current of each phase can be measured with a current sense resistor or DCR current sensing. The output from each current sense circuit is summed together to get the total load current. An additional feature of this scheme is that the current balance of each phase can be monitored. The circuit using a ISL28273 (dual EL8173) current sense circuit is shown in Figure 51 for a 30A, two phase circuit. High current (>30A) DC/DC converter outputs for microprocessor cores present very unique challenges for VIN = 5V Q1 L1 0.33µH VOUT = 1.2V AT 30A Cout Q2 U1 ISL28273 +5V 16 R1 10k VS+ 6 IN+A OUTA 2 R2 10k 5 IN-A VIN = 5V Q3 C1 0.1µF R5 61.9k FB-A 4 C2 0.1µF L2 0.33µH U2 ISL6568 F = 600kHz L1, L2: IHLP-2525CZ-07 ±5% DCR TOLERANCE DCR = 3.2mΩ R9 4.99k TOTAL Iout Vout = 0V to +3.0V 100mV/A R6 1k FB+A 3 Q4 PHASE 1 Iout Vout1 200mV/A R3 10k 11 R4 10k 12 IN-B C3 0.1µF R10 4.99k OUTB 15 IN+B FB-B 13 C4 0.1µF FB+B 14 R7 61.9k PHASE 2 Iout Vout2 200mV/A GAIN = 62.5 R8 1k VS8 FIGURE 51. 29 AN1298.2 May 27, 2009 Application Note 1298 The output current of each phase is measured by the DCR of L1 and L2 as explained in the previous example. VOUT1 and VOUT2 are proportional to the output current of each phase with a scale factor of 200mV/A. The two outputs, VOUT1 and VOUT2, are summed together with R9 and R10 to give a total output current, Total IOUT, with a scale factor of 100mV/A. This basic scheme can be extended to any number of phases for extremely high output currents exceeding 100A. Multiplexed Low Voltage Current Sense In a multiple voltage computer power supply, it is often necessary to monitor the output current from each DC/DC converter with an A/D Converter. When the EN pin of the ISL28271 and ISL28272 are enabled (i.e., device shutdown), the output stage goes into a high impedance state. This allows multiple VOUT pins to be connected together for multiplexed output applications. Since the output stage is in a high impedance state, only one set of feedback resistors (Rf, Rg) are required if all the amplifiers are operating with the same gain. Likewise, if different gains are required for each amplifier, separate feedback resistors can be used to set a unique gain on each amplifier. Figure 52 demonstrates the multiplexing scheme for a power system with four output voltages; 1.2V at 20A, 1.8V at 10A, 3.3V at 4A, and 5.0V at 7.5A. The dual instrumentation amplifier, ISL28271, is used to minimize parts count and circuit size. Each power supply load current is monitored with a low value current sense resistor (RS1, RS2, RS3, and RS4) to minimize voltage drop across the resistor. Input protection resistors (R1 to R8) limit the input fault current to <5mA in case of a short circuit on the OUT connection. 30 For this application, it is assumed VOUT would be measured with a microprocessor A/D Converter with a full scale voltage of 2.5V. Each channel is scaled for an output voltage, VOUT, equal to 2.0V at maximum load current to provide an overload measurement capability of 25%. For each amplifier, Sensed voltage, Vs = IOUT*Rs VOUT = Gain*Vs VOUT = Gain*Rs*IOUT TABLE 11. SENSED VOLTAGE VOUT AT MAX LOAD CURRENT (V) GAIN EN IOUT(AMPS) VOUT SENSITIVITY (V/A) 1 20 0.10 2.0 100 2 10 0.20 2.0 100 3 4.0 0.50 2.0 100 4 7.5 0.27 2.0 134 The gains of the two amplifiers of U1 are both set to 100 by a single feedback resistor divider network, R9 and R10. The gains of U2 are different in order to get the same VOUT sensitivity using standard current sense resistor values. Gain A is set to 100 by R11 and R12, and Gain B is set to 134 by R13 and R14. The EN lines (EN1, EN2, EN3, EN4) select the desired measurement channel. AN1298.2 May 27, 2009 Application Note 1298 Rs1 0.001Ω 1.2V IN 1.2V OUT, 20A Rs2 0.002Ω 1.8V IN R1 1k EN1 +5V 16 8 R3 1k R2 1k R4 1k 1.8V OUT, 10A U1 ISL28271 7 6 5 11 12 VS+ ENA IN+A IN-A IN+B IN-B VOUTA VS- ENB FB+A FB-A FB+B FB-B VOUTB 3 4 10 14 2 VOUT 15 R9 97.6k 13 EN2 GAIN A = GAIN B = 100 R10 1.0k Rs3 0.005Ω 3.3V IN 3.3V OUT, 4A Rs4 0.002Ω 5.0V IN 5.0V OUT, 7.5A R5 1k EN3 +5V 16 8 R6 1k R7 1k R8 1k U1 ISL28271 7 6 5 11 12 VS+ ENA IN+A IN-A IN+B IN-B VOUTA VS- ENB FB+A FB-A FB+B FB-B VOUTB 3 4 10 14 EN4 13 2 15 R11 97.6k R13 133k R12 1.0k R14 1.0k GAIN A = 100 GAIN B = 133.5 FIGURE 52. MEASUREMENT OF POSITIVE AND NEGATIVE CURRENT FLOW 31 AN1298.2 May 27, 2009 Application Note 1298 LOAD CURRENT = 2A, MAX X1 AND X2 MUST BE 0V TO +5V Rs 0.01Ω X1 X2 R2 4.7k R1 4.7k +5V U1 EL8170 +5V 7 VS+ 4 VS- D1 BAT54S VOUT 6 2 IN- VOUT = IOUT + 2 R3 250k 3 IN+ 8 R5 100k (20mv) FB+ 5 FBR4 1.0k R6 1.0k FIGURE 53. CURRENT MONITORED WITH LOW VALUE RESISTOR Bi-Directional Current Sense TABLE 12. The use of the FB pins of the EL8170 make it an ideal choice for a bi-directional current sense circuit for battery gas gauging or current monitor in a H-bridge configuration as shown in the following circuits. In Figure 53, current is monitored with the use of a low value resistor Rs. R1, R2, and D1 protect the EL8170 from overvoltage which would be applied with excessive load current or a short circuit on the output, X1 or X2. The amplifier is set for a gain of 100 with R5 and R6. R4 and R5 offset the FB+ pin at 20mV to center 0A at mid-range of the output voltage VOUT. LOAD CURRENT (ILOAD) VOUT -2A 0.0V 0A +2.0V +2A +4.0V The range of the measured current can easily be changed by proper selection of Rs, Gain and FB+ voltage. The circuit in Figure 54 shows the EL8170 set-up as a battery gas gauge to monitor both charging current and discharging current. In this circuit, when the battery is charging, the current in Rs will be negative (i.e., flowing from X2 to X1). The EL8170 output voltage will be between 0V and +2V. When the battery is being discharged, the current flow will be from X1 to X2, and the EL8170 output voltage will be between +2V and +4V. Sensed voltage is shown in Equation 77. V S = I LOAD∗ R S V OUT = Gain∗ ( V S + V FB+ ) V OUT = Gain∗ ( I LOAD × R S + V FB+ ) +5 × R 4 V FB+ = --------------------R4 + R5 V OUT = 100∗ ( I LOAD × 0.01 + 0.02 ) V OUT = ( I LOAD + 2 ) (EQ. 77) 32 AN1298.2 May 27, 2009 Application Note 1298 BATTERY CHARGER, POWER SUPPLY D3 Rs 0.01Ω SYSTEM LOAD X2 X1 D2 +5V BT1 R1 4.7k LITHIUM-ION (4.2V) R2 4.7k +5V U1 EL8170 7 VS+ 4 VS- D1 BAT54S VOUT 6 2 IN- VOUT = IOUT + 2 R3 250k 3 IN+ FB+ FB- R5 100k 8 (20 mv) 5 R4 1.0k R6 1.0k FIGURE 54. EL8170 SETUP AS A BATTERY CHARGER If a more direct measurement for the current polarity and increased output voltage sensitivity is required, the circuit shown in Figure 55 can be used. In Figure 55, U1 is used to measure positive current flow (X1 to X2) and U2 is used to measure negative current flow (X2 to X1). The polarity of the current is detected by U3 which is being used a zero crossing detect comparator. The EN pins of the EL8170 (U1, U2) are used to turn on the proper amplifier depending if the current flow is positive or negative. In the above circuit, current is monitored with the use of a low value resistor Rs. R1, R2, and D1 protect the EL8170’s from over-voltage which would be applied with excessive load current or a short circuit on the output, X1 or X2. When this bi-directional current sense circuit is used in a PWM application such as a H-bridge, C1 and C2 can be added to filter the PWM signal for an average current value. Figure 56 shows the bi-directional current source circuit configured to monitor the motor current in a H-bridge circuit. The direction of the motor (CW, CCW) is monitored by the polarity bit depending on the direction of current flow in the motor. The rail-to- rail input capability of the EL8170 allows current sensing at ground level (Q1 and Q4 ON) or at +5V (Q3 and Q2 ON). If pulse width modulation is used to control the speed of the motor, filter capacitors C1 and C2 should be used to obtain the average value of the motor current. The value of the capacitors should be selected based on the PWM frequency and desired overall accuracy. The amplifiers are set for a gain of 100 with R5 and R6. The minimum sensed current is set by the EL8170 offset voltage. V OS I MIN = -----------RS 0.25mV I MIN = --------------------0.01Ω (EQ. 78) I MIN = 25mA 33 AN1298.2 May 27, 2009 Application Note 1298 Rs 0.01Ω X2 X1 R1 4.7k D1 BAT54S C1 SEE TEXT R2 4.7k +5V C2 SEE TEXT +5V U1 ISL28271 7 VS+ ENA 7 4 VS- VOUTA 2 3 IN+ FB+A 3 2 IN- FB-A 4 10k +Iout AMP +5V U1 ISL28271 7 VS+ ENB 10 4 VS- VOUTB 15 3 IN+ FB+B 14 2 IN- FB-B 13 -Iout AMP +5V VOUT 1V/A R3 100k R4 1.0k AV = 100 U2 ISL28271 7 VS+ ENA 7 4 VS- VOUTA 2 3 IN+ FB+A 3 2 IN- FBA- 4 +Iout -Iout Q1 2N7002 POLARITY DETECT FIGURE 55. MEASUREMENT OF POSITIVE AND NEGATIVE CURRENT FLOW 34 AN1298.2 May 27, 2009 Application Note 1298 +5V Q1 Q3 + - Rs 0.01Ω MOTOR X2 VOUT X1 POLARITY VOUT = Imotor CLOCKWISE = +5V COUNTER CLOCKWISE = 0V EL8170 CIRCUIT Q2 Q4 (GAIN = 100) FIGURE 56. BI-DIRECTIONAL CURRENT SOURCE CIRCUIT Intersil Corporation reserves the right to make changes in circuit design, software and/or specifications at any time without notice. Accordingly, the reader is cautioned to verify that the Application Note or Technical Brief is current before proceeding. For information regarding Intersil Corporation and its products, see www.intersil.com 35 AN1298.2 May 27, 2009