MICROCHIP AN695 Interfacing Pressure Sensors to Microchip's Analog Peripherals This application note will concentrate on the signal con- Author: Bonnie Baker Microchip Technology Inc. ditioning path of the piezoresistive sensing element from sensor to microcontroller.It will show how the electrical output of this sensor can be gained,filtered INTRODUCTION and digitized in order to ready it for the microcontroller's calibration routines.This theoretical discussion will be Pressure measurement devices can be classified into followed with a specific pressure sensing design that is two groups:those where pressure is the only source of specifically designed to measure barometric pressure. power and those that require electrical excitation.The mechanical style devices that are only excited by pres- PIEZORESISTIVE PRESSURE sure,such as bellows,diaphragms,bourdons,tubes or SENSORS manometers,are usually suitable for purely mechani- cal systems.With these devices a change in pressure The piezoresistive is a solid state,monolithic sensor will initiate a mechanical reaction,such as a change in that is fabricated using silicon processing.Piezo means the position of mechanical arm or the level of liquid in a pressure,resistance means opposition to a DC current tube. flow.Since piezoresistive pressure sensors are fabri- Electrically excited pressure sensors are most syner- cated on a wafer,300 to 500 sensors can be produced gistic with the microcontroller environment.These style per wafer.Since these wafers generate a large number of sensors can be piezoresistive,Linear Variable Differ- of sensors they are available on the market at a ential Transformers (LVDT),or capacitive sensors. reduced cost as compared to mechanical sensors. Most typically,the piezoresistive sensor is used when measuring pressure. Voltage or Voltage or 个Current 不 Current Excitation Excitation Rs1 Rs2 VOUT- VOUT- 3 Rs4 Rs3 oVOUT+ Rs4 Rs3 oVOUT+ (a.)single element bridge (b.)two element bridge Dielectric Voltage or Contact Contact Current Excitation Si-P Rs2 Silicon VoUT- Substrate Si-N Rs3 VOUT+ Diaphragm (c.)four element or full bridge (d.)single side of a sandwiched piezoelectric pressure sensor Figure 1:The resistive wheatstone bridge configuration can have one variable element(a.),two elements that vary with excitation(b.)or four elements(c.).The piezoresistive pressure sensing element is usually a four element bridge and is constructed in silicon(d.). 2000 Microchip Technology Inc. Preliminary DS00695A-page 1
2000 Microchip Technology Inc. Preliminary DS00695A-page 1 INTRODUCTION Pressure measurement devices can be classified into two groups: those where pressure is the only source of power and those that require electrical excitation. The mechanical style devices that are only excited by pressure, such as bellows, diaphragms, bourdons, tubes or manometers, are usually suitable for purely mechanical systems. With these devices a change in pressure will initiate a mechanical reaction, such as a change in the position of mechanical arm or the level of liquid in a tube. Electrically excited pressure sensors are most synergistic with the microcontroller environment. These style of sensors can be piezoresistive, Linear Variable Differential Transformers (LVDT), or capacitive sensors. Most typically, the piezoresistive sensor is used when measuring pressure. This application note will concentrate on the signal conditioning path of the piezoresistive sensing element from sensor to microcontroller. It will show how the electrical output of this sensor can be gained, filtered and digitized in order to ready it for the microcontroller’s calibration routines. This theoretical discussion will be followed with a specific pressure sensing design that is specifically designed to measure barometric pressure. PIEZORESISTIVE PRESSURE SENSORS The piezoresistive is a solid state, monolithic sensor that is fabricated using silicon processing. Piezo means pressure, resistance means opposition to a DC current flow. Since piezoresistive pressure sensors are fabricated on a wafer, 300 to 500 sensors can be produced per wafer. Since these wafers generate a large number of sensors they are available on the market at a reduced cost as compared to mechanical sensors. Figure 1: The resistive wheatstone bridge configuration can have one variable element (a.), two elements that vary with excitation (b.) or four elements (c.). The piezoresistive pressure sensing element is usually a four element bridge and is constructed in silicon (d.). Author: Bonnie Baker Microchip Technology Inc. RS1 RS1 RS1 RS2 RS2 RS2 RS4 RS4 RS3 RS4 RS3 RS3 Voltage or Current Excitation Voltage or Current Excitation Voltage or Current Excitation VOUT+ VOUTVOUT+ VOUTVOUT+ VOUT- (a.) single element bridge (b.) two element bridge (c.) four element or full bridge (d.) single side of a sandwiched piezoelectric pressure sensor Silicon Substrate Contact Dielectric Contact Si-P Si-N Diaphragm AN695 Interfacing Pressure Sensors to Microchip’s Analog Peripherals
AN695 With this sensor,the resistors are arranged in a full ELECTRONICS SIGNAL PATH wheatstone bridge configuration,which has improved sensitivity as compared to a single element or two ele- There are several ways of capturing the small differen- ment sensors (see Figure 1.d).When a positive differ- tial output signal of the sensor and transforming it into ential pressure is applied to the four element bridge, a usable digital code.One approach that can be taken two of the elements respond by compressing and the is shown in the block diagram in Figure 2.a.With this other two change to a tension state.When a negative approach,the small differential output of the bridge is differential pressure is applied to the sensor,the dia- gained and converted from differential to single ended phragm is strained in the opposite direction and the with an instrumentation amplifier(IA).The signal may resistors that were compressed go into a tension state, or may not travel through a multiplexer.The signal then while the resistors that were in a tension state change passes through a low pass filter.The low pass filter into a compression state.Piezoresistive pressure sen- eliminates out-of-band noise and unwanted frequen- sors may or may not have an internal pressure refer- cies in the system before the A/D conversion is per- ence.If they do,a pressure reference cavity is formed.This is followed by a stand-alone A/D converter generally fabricated by sealing two wafers together. which transforms the analog signal into a usable digital The top side of this fabricated sensor is the resistive code.The microcontroller takes the converter code, material and the bottom is the diaphragm. further calibrates and translates if need be for display The high side of the piezoresistive bridges shown in purposes.In this signal path only one analog filter is Figure 1 can have a voltage excitation or current exci- required and it is positioned at the output of the multi- tation applied.Although the magnitude of excitation plexer. (whether it is voltage or current)effects the dynamic The second signal path shown in Figure 2.b also has range of the output of the sensor,the maximum differ- an instrumentation amplifier(IA)in the signal path.Fol- ence between VoUT+and VouT-generally ranges from lowing the instrumentation amplifier stage the signal is 10s of millivolts to several hundred millivolts.The elec- filtered in the analog domain and then digitized with an tronics that follow the sensor are used to change the on-chip microcontroller's A/D converter.When this type differential output signal to single ended as well as gain of signal path is used,every signal going into the mul- and filter it in preparation for digitization. tiplexer will require its own analog filter.Additionally, the accuracy and speed of the converter in the micro- controller is less than a stand-alone A/D converter.This may or may not be an issue in a particular application. (a.) MUX FILTER SAR Micro- A/D Controller REF FILTER MUX A/D VREF LOw MUX Pass COMP> Filter uC VREF Figure 2: Three block diagrams for the piezoresistive pressure sensor signal conditioning path are shown in this Figure.The top two block diagrams,a.and b.,are discussed in detail in this application note.The bottom block diagram(c.)is discussed in detail in AN717(Microchip Technology Inc.). DS00695A-page 2 Preliminary 2000 Microchip Technology Inc
AN695 DS00695A-page 2 Preliminary 2000 Microchip Technology Inc. With this sensor, the resistors are arranged in a full wheatstone bridge configuration, which has improved sensitivity as compared to a single element or two element sensors (see Figure 1.d). When a positive differential pressure is applied to the four element bridge, two of the elements respond by compressing and the other two change to a tension state. When a negative differential pressure is applied to the sensor, the diaphragm is strained in the opposite direction and the resistors that were compressed go into a tension state, while the resistors that were in a tension state change into a compression state. Piezoresistive pressure sensors may or may not have an internal pressure reference. If they do, a pressure reference cavity is generally fabricated by sealing two wafers together. The top side of this fabricated sensor is the resistive material and the bottom is the diaphragm. The high side of the piezoresistive bridges shown in Figure 1 can have a voltage excitation or current excitation applied. Although the magnitude of excitation (whether it is voltage or current) effects the dynamic range of the output of the sensor, the maximum difference between VOUT+ and VOUT- generally ranges from 10s of millivolts to several hundred millivolts. The electronics that follow the sensor are used to change the differential output signal to single ended as well as gain and filter it in preparation for digitization. ELECTRONICS SIGNAL PATH There are several ways of capturing the small differential output signal of the sensor and transforming it into a usable digital code. One approach that can be taken is shown in the block diagram in Figure 2.a. With this approach, the small differential output of the bridge is gained and converted from differential to single ended with an instrumentation amplifier (IA). The signal may or may not travel through a multiplexer. The signal then passes through a low pass filter. The low pass filter eliminates out-of-band noise and unwanted frequencies in the system before the A/D conversion is performed. This is followed by a stand-alone A/D converter which transforms the analog signal into a usable digital code. The microcontroller takes the converter code, further calibrates and translates if need be for display purposes. In this signal path only one analog filter is required and it is positioned at the output of the multiplexer. The second signal path shown in Figure 2.b also has an instrumentation amplifier (IA) in the signal path. Following the instrumentation amplifier stage the signal is filtered in the analog domain and then digitized with an on-chip microcontroller’s A/D converter. When this type of signal path is used, every signal going into the multiplexer will require its own analog filter. Additionally, the accuracy and speed of the converter in the microcontroller is less than a stand-alone A/D converter. This may or may not be an issue in a particular application. Figure 2: Three block diagrams for the piezoresistive pressure sensor signal conditioning path are shown in this Figure. The top two block diagrams, a. and b., are discussed in detail in this application note. The bottom block diagram (c.) is discussed in detail in AN717 (Microchip Technology Inc.). (a.) (c.) (b.) MUX Low Pass Filter COMP µC VREF A/D µC VREF MUX IA FILTER IA REF MUX FILTER A/D SAR MicroController
AN695 INSTRUMENTATION AMPLIFIER ational amplifiers,thereby significantly reducing source OPTIONS AND DESIGN impedance mismatch problems at DC.The transfer function of this circuit is equal to: With this application,the two low voltage signals from the bridge need to be subtracted in order to produce a single ended output signal.The results of this subtrac- tion also need to be gained so that it matches the input range of the A/D converter.The implementation of the subtraction and gain functions are done so that the sensor signal is not contaminated with additional It should be noted from this transfer function that the errors.The instrumentation amplifier circuits shown in input signals are gained along with the common-mode Figure 3 and Figure 4 achieve all of these goals.Both voltage of the two signals.The common-mode voltage of these configurations take two opposing input signals, can be rejected when R1=R4 and R2=R3.Given this subtract them and apply gain.The subtraction process change the transfer function becomes: inherently rejects common-mode voltages.Combined with these functions the signal is level shifted,making R1 2R1) it synergistic with the signal supply environment. 6ur=(V4-V1+瓦2+可 VREF The Two Op Amp Instrumentation Amplifier A solution to the circuit problem discussed above is The common-mode rejection error that is caused by shown in Figure 3.The circuit in Figure 3 uses two resistor mismatch is equal to: operational amplifiers and five resistors to solve this R2 gain and subtraction problem. 1* Dual amplifiers are usually used in this discrete design CMR =100 (of mismatch error) because of their good matching of bandwidth and over temperature performance.This instrumentation ampli- fier design uses the high impedance inputs of the oper- RG W R2 VpD R4 VREF O- R1 A1 R3 MCP602 A2 VIN-O MCP602 VIN+O Where R1=30k2 and R2=10k Figure 3:The two op amp instrumentation amplifier takes the difference of two input signals,gains that difference, while rejecting any voltage that is common to both of the input signals. 2000 Microchip Technology Inc. Preliminary DS00695A-page 3
2000 Microchip Technology Inc. Preliminary DS00695A-page 3 AN695 INSTRUMENTATION AMPLIFIER OPTIONS AND DESIGN With this application, the two low voltage signals from the bridge need to be subtracted in order to produce a single ended output signal. The results of this subtraction also need to be gained so that it matches the input range of the A/D converter. The implementation of the subtraction and gain functions are done so that the sensor signal is not contaminated with additional errors. The instrumentation amplifier circuits shown in Figure 3 and Figure 4 achieve all of these goals. Both of these configurations take two opposing input signals, subtract them and apply gain. The subtraction process inherently rejects common-mode voltages. Combined with these functions the signal is level shifted, making it synergistic with the signal supply environment. The Two Op Amp Instrumentation Amplifier A solution to the circuit problem discussed above is shown in Figure 3. The circuit in Figure 3 uses two operational amplifiers and five resistors to solve this gain and subtraction problem. Dual amplifiers are usually used in this discrete design because of their good matching of bandwidth and over temperature performance. This instrumentation amplifier design uses the high impedance inputs of the operational amplifiers, thereby significantly reducing source impedance mismatch problems at DC. The transfer function of this circuit is equal to: It should be noted from this transfer function that the input signals are gained along with the common-mode voltage of the two signals. The common-mode voltage can be rejected when R1 = R4 and R2 = R3. Given this change the transfer function becomes: The common-mode rejection error that is caused by resistor mismatch is equal to: Figure 3: The two op amp instrumentation amplifier takes the difference of two input signals, gains that difference, while rejecting any voltage that is common to both of the input signals. VOUT VIN+ VIN- ( ) – R4 R3 ------ 1 1 2 -- R2 R1 ------ R3 R4 ------ + + = + R2 R3 + RG -------------------- VCM + R4 R3 ------ R3 R4 ------ R2 R1 – ------ VREF + VOUT VIN+ VIN- ( ) – 1 R1 R2 + ------ 2R1 RG ---------- VREF = + + CMR 100* 1 R2 R1 + ------ ( ) % of mismatch error ---------------------------------------------------------- = + - + - A2 MCP602 A1 MCP602 VOUT 1 R1 R2 + ------ 2R1 RG ---------- VIN+ VIN- ( ) – VREF = 4 60 + kΩ RG ------------------------ VIN+ VIN- – ) VREF ( + = + + R4 VREF VINVIN+ R1 RG R2 R3 VDD Where R1 = 30kΩ and R2 = 10kΩ
AN695 The ac common mode rejection for this configuration is The second factor that limits the common-mode input poor.This is due to the fact that the common mode sig- range of this circuit comes from the input swing restric- nal at VIN.is inverted once with A1 and then it travels tions of the amplifiers themselves. through A2 causing a second propagation delay.The common mode signal at VIN+only travels through one If this circuit is in a single supply environment,it will typ- operational amplifier(A2).Additionally,the two opera- ically require a reference that is centered at the com- mon-mode voltage of the input signals.In Figure 4, tional amplifiers have different closed loop gains,and consequently different closed loop bandwidths. VREF serves that function.This voltage can be imple- mented discretely with a precision reference chip as In terms of common-mode input range,there are two shown in Figure 4.a or with two equal resistors in series factors that limit the range of this instrumentation ampli- between the power supply as shown in Figure 4.b. fier.The first factor involves the operation of A1 as it Another added benefit to matching R2/R=Ra/R4 is responds to the VIN-and VIN input signals and the volt- that the gain of the circuit can be changed with one age reference,VREF The signal at the non-inverting resistor,RG. input to A1 and A2 gained by the output of A1 by: This instrumentation amplifier circuit has high imped- VOUT-A1 VIN- (RGR2+RR2+RRG) ance inputs and programmable gain capability.The (R1R2) features that could be improved in this circuit solution is to have the common-mode rejection independent of VIN+RG gain and better over frequency.These performance characteristics can only be obtained by an instrumen- tation amplifier configuration that has three operational amplifiers R1=RIAlI R1B VpD Vop个 个 VREF M- RiA R2 RG R1B VoD R1 VREF R2 OR ViN- MCP602> ◆ MCP602> 0- VoUT VIN+ (a) (b) Figure 4:The reference voltage for a two op amp instrumentation amplifier in a single supply environment can be implemented with a stand-alone voltage reference(a)or a resistor divider across a voltage reference or the supply voltage(b). DS00695A-page 4 Preliminary @2000 Microchip Technology Inc
AN695 DS00695A-page 4 Preliminary 2000 Microchip Technology Inc. The ac common mode rejection for this configuration is poor. This is due to the fact that the common mode signal at VIN- is inverted once with A1 and then it travels through A2 causing a second propagation delay. The common mode signal at VIN+ only travels through one operational amplifier (A2). Additionally, the two operational amplifiers have different closed loop gains, and consequently different closed loop bandwidths. In terms of common-mode input range, there are two factors that limit the range of this instrumentation amplifier. The first factor involves the operation of A1 as it responds to the VIN- and VIN+ input signals and the voltage reference, VREF. The signal at the non-inverting input to A1 and A2 gained by the output of A1 by: The second factor that limits the common-mode input range of this circuit comes from the input swing restrictions of the amplifiers themselves. If this circuit is in a single supply environment, it will typically require a reference that is centered at the common-mode voltage of the input signals. In Figure 4, VREF serves that function. This voltage can be implemented discretely with a precision reference chip as shown in Figure 4.a or with two equal resistors in series between the power supply as shown in Figure 4.b. Another added benefit to matching R2/R1 = R3/R4 is that the gain of the circuit can be changed with one resistor, RG. This instrumentation amplifier circuit has high impedance inputs and programmable gain capability. The features that could be improved in this circuit solution is to have the common-mode rejection independent of gain and better over frequency. These performance characteristics can only be obtained by an instrumentation amplifier configuration that has three operational amplifiers. Figure 4: The reference voltage for a two op amp instrumentation amplifier in a single supply environment can be implemented with a stand-alone voltage reference (a) or a resistor divider across a voltage reference or the supply voltage (b). VOUT A– 1 VINRGR2 R1R2 R1RG ( ) + + (R1R2 ) ---------------------------------------------------------------- = –VIN+ R2 RG ------- VREF R2 R1 ------ – OR R2 + - + - MCP602 Precision Voltage MCP602 Reference (a) (b) R1 = R1A || R1B R1A R1B R2 RG VDD VINVIN+ R1 VOUT VDD VREF VREF VDD
AN695 The Three Op Amp Instrumentation Amplifier all gains as long as the signals stay within A1 and A2 An example of a more versatile instrumentation ampli- input and output head room limitations.If the common fier configuration is shown in Figure 5. mode errors of the input amplifiers track they will be cancelled by the output stage. With this circuit configuration.two of the three amplifi- ers (A1 and A2)gain the two input signals.The third If the assumption that R/R2 equals Rg/R4 is not cor- amplifier,A3,is used to subtract the two gained input rect,there could be a noticeable common mode volt- signals,thereby providing a single ended output.The age error.The calculated common-mode rejection transfer function of this circuit is equal to: (CMR)error that is attributed to resistor mismatches in this circuit is equal to: R1+R2 Vour=V:(1+R后)H,+R)i R2 1+ R1+R2 CMR =100 (of mismatch error) for R =Ra and R2 R4 If RE2=RF1,R=R3,and R2=R4,this formula can be simplified to equal: 2RF An example of the impact of this error is demonstrated VoUT (VIN-ViN.1+G VREF with a 12-bit,5V system,where the gain of the circuit is 100V/V,the common-mode voltage ranges 0 to 5V and Quad amplifiers are typically used in the three op-amp the matching error can be as large as +1%.Using the instrumentation amplifier discrete designs because of formula above,the contributed error of this type of com- the matching qualities of amplifiers with the same sili- mon-mode excursion is equal to 1mV.This voltage is con.In contrast to the two op-amp instrumentation slightly less than 1LSB. amplifier,the input signal paths(at VIN+and VIN-)are In a single supply environment,the voltage reference completely balanced.This is achieved by sending VIN+ should be equal to the center of the input signals.This and VIN-signals through the same number of amplifiers voltage is represented in the circuit in Figure 5 as VREF to the output and using a common gain resistor,RG. The purpose and effects of this reference voltage is to Since this input stage is balanced,common mode cur- simply shift the output signal into the linear region of the rents will not flow through RG.The common-mode amplifier. rejection of this circuit is primarily dependent on the resistor matching around A3.When R1=R2=R3=R4, common mode signals will be gained by a factor of one regardless of gain of the front end of the circuit.Conse- quently,large common mode signals can be handled at VIN- A1 MCP604 R2 w RF1 R1 W A3 RG MCP604 R3 + VoUT A2 MCP604 R4 VIN+o VREF Where RF1=RF2 and R1=R2=R3=R4 2RE)(ViN-ViN)+VAEF our=(1+ Figure 5: This is a three op amp implementation of an instrumentation amplifier. 2000 Microchip Technology Inc. Preliminary DS00695A-page 5
2000 Microchip Technology Inc. Preliminary DS00695A-page 5 AN695 The Three Op Amp Instrumentation Amplifier An example of a more versatile instrumentation amplifier configuration is shown in Figure 5. With this circuit configuration, two of the three amplifiers (A1 and A2) gain the two input signals. The third amplifier, A3, is used to subtract the two gained input signals, thereby providing a single ended output. The transfer function of this circuit is equal to: If RF2 = RF1, R1 = R3, and R2 = R4, this formula can be simplified to equal: Quad amplifiers are typically used in the three op-amp instrumentation amplifier discrete designs because of the matching qualities of amplifiers with the same silicon. In contrast to the two op-amp instrumentation amplifier, the input signal paths (at VIN+ and VIN-) are completely balanced. This is achieved by sending VIN+ and VIN- signals through the same number of amplifiers to the output and using a common gain resistor, RG. Since this input stage is balanced, common mode currents will not flow through RG. The common-mode rejection of this circuit is primarily dependent on the resistor matching around A3. When R1 = R2 = R3 = R4, common mode signals will be gained by a factor of one regardless of gain of the front end of the circuit. Consequently, large common mode signals can be handled at all gains as long as the signals stay within A1 and A2 input and output head room limitations. If the common mode errors of the input amplifiers track they will be cancelled by the output stage. If the assumption that R1/R2 equals R3/R4 is not correct, there could be a noticeable common mode voltage error. The calculated common-mode rejection (CMR) error that is attributed to resistor mismatches in this circuit is equal to: An example of the impact of this error is demonstrated with a 12-bit, 5V system, where the gain of the circuit is 100V/V, the common-mode voltage ranges 0 to 5V and the matching error can be as large as ±1%. Using the formula above, the contributed error of this type of common-mode excursion is equal to 1mV. This voltage is slightly less than 1LSB. In a single supply environment, the voltage reference should be equal to the center of the input signals. This voltage is represented in the circuit in Figure 5 as VREF. The purpose and effects of this reference voltage is to simply shift the output signal into the linear region of the amplifier. Figure 5: This is a three op amp implementation of an instrumentation amplifier. VOUT VIN+ 1 2RF2 RG + ------------- R4 R1 R2 + (R3 R4 )R1 + --------------------------------- – = VIN- 1 2RF1 RG + ------------- R2 R1 ------ VREF R3 R1 R2 + (R3 R4 )R1 + --------------------------------- + VOUT VIN+ VIN- ( ) – 1 2RF RG ---------- VREF + + = CMR 100* 1 R2 R1 + ------ ( ) % of mismatch error ---------------------------------------------------------- for R1 R3 and R2 = R4 = = RG VINVIN+ RF1 RF2 R1 R3 R2 R4 VOUT VREF VOUT 1 2RF RG + ---------- VIN+ VIN- ( ) – VREF = + Where RF1 = RF2 and R1 = R2 = R3 = R4 + - MCP604 A2 A3 A1 MCP604 MCP604 - - + +
AN695 R From the output of O- A3 MCP604 From the R1 output of-w◆ VoUT A VD木 O VSHIFT OR VSHIFT R2A R2B R2=R2A ll R2B (a) (b) Figure 6: The reference voltage in Figure 5 can be implemented by using a precision reference circuit(a.)or a resistive voltage divider circuit(b.). The VREF circuit function can be implemented with a These signals are usually unintentional,but almost precision voltage reference or with the resistive net- always destructive if not controlled.On the down side, work shown in Figure 6. analog filters can add to the noise floor particularly if a noisy amplifier is used with a large gain. ANALOG FILTERING Where analog filters eamn their worth by rejecting noise A big topic for debate in digital design circles is whether in the out of band region,digital filters can be utilized to or not an analog filter is needed and more importantly, reduce the in-band noise floor.This is implemented can a digital filter replace the analog filter. with oversampling algorithms.These types of filters are much easier to change on the fly because it is a matter A common assumption with designers that are trying to tackle analog challenges of this type is that they claim of programming instead of a matter of changing resis- tors and capacitors as it is with analog filters.With all of that they are only measuring DC so they don't have to worry about filtering.Unfortunately,the noise genera- these benefits there is a price to pay in terms of response time.Digital filters must collect a certain tors in the electronics and the environment do not have amount of conversion data before calculations can be the "intelligence"to accommodate the designer's desires.Consequently,if a filter is not included,the cir- performed.The digital filter algorithms tend to slow down the response time as well as delay the output.If cuit will be surprisingly noisier than anticipated. real time responses are not critical,the digital filter dis- Once it is accepted that a filter is required.the next advantages are not detrimental to the operation of cir- debate that ensues is whether the filtering strategy cuit. should be analog,digital or both. A common assumption that is made by programmers is that they can eliminate all ills with digital filtering.To some extent this is true,however,it is at a high price of time and memory and truthfully,it may not be possible to succeed. Analog filtering removes a considerable amount of headaches for the programmer from the start.Analog filters have their place in circuit designs as do digital fil- ters.For instance,analog filters will eliminate aliasing errors that will occur through the A/D conversion pro- cess if they are allowed to go through.Once these errors are allowed in the conversion it is impossible to discriminate good signal from aliased signal in the dig- ital domain.The analog filter also removes large signal noise that is generated by spikes or spurs in the signal DS00695A-page 6 Preliminary @2000 Microchip Technology Inc
AN695 DS00695A-page 6 Preliminary 2000 Microchip Technology Inc. Figure 6: The reference voltage in Figure 5 can be implemented by using a precision reference circuit (a.) or a resistive voltage divider circuit (b.). The VREF circuit function can be implemented with a precision voltage reference or with the resistive network shown in Figure 6. ANALOG FILTERING A big topic for debate in digital design circles is whether or not an analog filter is needed and more importantly, can a digital filter replace the analog filter. A common assumption with designers that are trying to tackle analog challenges of this type is that they claim that they are only measuring DC so they don’t have to worry about filtering. Unfortunately, the noise generators in the electronics and the environment do not have the “intelligence” to accommodate the designer’s desires. Consequently, if a filter is not included, the circuit will be surprisingly noisier than anticipated. Once it is accepted that a filter is required, the next debate that ensues is whether the filtering strategy should be analog, digital or both. A common assumption that is made by programmers is that they can eliminate all ills with digital filtering. To some extent this is true, however, it is at a high price of time and memory and truthfully, it may not be possible to succeed. Analog filtering removes a considerable amount of headaches for the programmer from the start. Analog filters have their place in circuit designs as do digital filters. For instance, analog filters will eliminate aliasing errors that will occur through the A/D conversion process if they are allowed to go through. Once these errors are allowed in the conversion it is impossible to discriminate good signal from aliased signal in the digital domain. The analog filter also removes large signal noise that is generated by spikes or spurs in the signal. These signals are usually unintentional, but almost always destructive if not controlled. On the down side, analog filters can add to the noise floor particularly if a noisy amplifier is used with a large gain. Where analog filters earn their worth by rejecting noise in the out of band region, digital filters can be utilized to reduce the in-band noise floor. This is implemented with oversampling algorithms. These types of filters are much easier to change on the fly because it is a matter of programming instead of a matter of changing resistors and capacitors as it is with analog filters. With all of these benefits there is a price to pay in terms of response time. Digital filters must collect a certain amount of conversion data before calculations can be performed. The digital filter algorithms tend to slow down the response time as well as delay the output. If real time responses are not critical, the digital filter disadvantages are not detrimental to the operation of circuit. R2 = R2A || R2B VDD VSHIFT R1 R2 R1 VDD VOUT VSHIFT R2A R2B Prec si ion Voltage Reference From the output of A1 From the output of A2 - + MCP604 OR (a) (b) A3
AN695 R2 OM●M R1 MCP VoUT 606 R3 R4 Figure 7:By using FilterLabTM software,this 2nd order low pass filter that has a non-inverting gain in the pass band can be configured as a Butterworth,Bessel,or Chebyshev filter. As discussed previously,the hardware implementation Close inspection of this filter shows that the circuit can of a low pass filter at its most fundamental level be configured in a gain of +1V/V by shorting R4 and requires a capacitor and resistor for each pole.Active opening Ra.In this configuration it is likely that the input filters,which have one amplifier for every two poles, of the amplifier will be exercised across a full rail-to-rail have the added benefit of preventing conflicting imped- input range. ances and degrading the signal path. The second order Multiple Feedback circuit implemen- The 2nd order lower pass filter shown in Figure 7 is one tation of a low pass filter uses an amplifier,three resis- of a class of circuits that were described in 1995 by R.P tors and two capacitors,as is shown in Figure 8.The Sallen and E.L.Key.With this filter the DC gain is pos- DC gain of this filter is negative and easily adjusted with itive.In a single supply environment this eases the the ratio of R3 and R1.When used in a single supply implementation considerably,because a mid-supply environment,this circuit usually needs a voltage refer- reference is not required.This circuit not only filters ence on the non-inverting input of the amplifier. high frequencies,but it can be used to gain the incom- This is the filter circuit that will be used in a barometric ing signal. pressure application.An adjustable voltage reference will be Included in this filter design. M R3 C R1 VIN O R2 wN MCP 606 oVOUT VREF Figure 8:By using FilterLabTM software,this 2nd order low pass filter that has an inverting gain in the pass band can be configured as a Butterworth,Bessel,or Chebyshev filter. 2000 Microchip Technology Inc. Preliminary DS00695A-page 7
2000 Microchip Technology Inc. Preliminary DS00695A-page 7 AN695 Figure 7: By using FilterLabTM software, this 2nd order low pass filter that has a non-inverting gain in the pass band can be configured as a Butterworth, Bessel, or Chebyshev filter. As discussed previously, the hardware implementation of a low pass filter at its most fundamental level requires a capacitor and resistor for each pole. Active filters, which have one amplifier for every two poles, have the added benefit of preventing conflicting impedances and degrading the signal path. The 2nd order lower pass filter shown in Figure 7 is one of a class of circuits that were described in 1995 by R.P Sallen and E.L. Key. With this filter the DC gain is positive. In a single supply environment this eases the implementation considerably, because a mid-supply reference is not required. This circuit not only filters high frequencies, but it can be used to gain the incoming signal. Close inspection of this filter shows that the circuit can be configured in a gain of +1V/V by shorting R4 and opening R3. In this configuration it is likely that the input of the amplifier will be exercised across a full rail-to-rail input range. The second order Multiple Feedback circuit implementation of a low pass filter uses an amplifier, three resistors and two capacitors, as is shown in Figure 8. The DC gain of this filter is negative and easily adjusted with the ratio of R3 and R1. When used in a single supply environment, this circuit usually needs a voltage reference on the non-inverting input of the amplifier. This is the filter circuit that will be used in a barometric pressure application. An adjustable voltage reference will be Included in this filter design. Figure 8: By using FilterLabTM software, this 2nd order low pass filter that has an inverting gain in the pass band can be configured as a Butterworth, Bessel, or Chebyshev filter. VIN C1 C2 R2 R1 R3 R4 - + MCP 606 VOUT VIN R1 R2 C1 R3 VOUT - + MCP C 606 2 VREF
AN695 BAROMETRIC PRESSURE SENSING Parameter (w/5V excitation) Specification The considerations for the design of a barometric sens- Operating Pressure Range 0 to 15 psi ing system encompasses altitude and resolution.The Sensitivity 6.0mV/psi(typ) expected altitude that our sensor will be placed in is Full-Scale Span 90mV(typ) approximately from sea level to 20,000 ft.The nominal pressure at sea level is 14.7 psi and the nominal pres- Zero Pressure Offset ±300uV sure at 20,000 ft.is 6.75 psi.The difference in pressure Temperature Effect on Span ±0.5%FS0 between these two altitudes is 7.95 psi.With this range, (0to70C) the appropriate pressure sensor should be an absolute Temperature Effect on Offset ±500uV version that is referenced to an on-chip vacuum and (0to70℃) have a range up to 15psi.Since the change in pressure for major weather changes is approximately 0.18 psi,a Table 1:The specifications of the SCX015 from resolution of 0.015 psi is over ten times more accurate SenSym indicates that this is a good pressure than the measured value.The circuit that will be used sensor that can be used to measure barometric for this design discussion is shown in Figure 9. pressure. The critical pressure sensor specifications for this The specifications of the SCX015 from SenSym indi- application include the operating pressure range,sen- cates that this is a good pressure sensor that can be sitivity,room temperature(25C)span and offset errors used to measure barometric pressure. as well as over temperature (see Table 1).Although, Note that temperature issues are beyond the scope of the range of this sensor extends from 0 psi to 15 psi, this application note.Detailed information about tem- this application will not be using that lower range.The perature sensing circuits can be found in Microchip's minimum differential output voltage from the sensor will AN679,AN684,AN685,and AN687. be 40.5mV(6.75psi or 20,000 ft.)and the maximum sensor voltage will be 88.2mV(14.7psi or sea level). The voltage at the output of the sensor is gained before it is digitized using an instrumentation amplifier. Instrumentation Amplifier RG M R2 2nd Order Butterworth W Low Pass Filter VDD=5V M 1/2 MCP602 12 A1 SCLK MCP602 MCP MCP3201 DOUT 606 A2 cS a SCX A3 015 A6 R1=30k2 35.7k R2=10k2 SCK Rg=1.15k2 SI A5 =957k 10k2 R4=172kQ cs A4 and A6 can be R5=304 ko 68.1k2 replaced with a PICmicro that has an on-chip A/D Level Shift Voltage converter A1 =A2=A3=Single Supply,CMOS op amp and Offset Adjust A4=12-bit,A/D SAR Converter A5=10k Digital Potentiometer Figure 9: The voltage at the output of the SCX015 pressure sensor is gained by the instrumentation amplifier(A1 and A2)then filtered,gained and level shifted(A4)with a 2nd order low pass filter(A3)and digitized with a 12-bit A/D converter(A5). DS00695A-page 8 Preliminary @2000 Microchip Technology Inc
AN695 DS00695A-page 8 Preliminary 2000 Microchip Technology Inc. BAROMETRIC PRESSURE SENSING The considerations for the design of a barometric sensing system encompasses altitude and resolution. The expected altitude that our sensor will be placed in is approximately from sea level to 20,000 ft. The nominal pressure at sea level is 14.7 psi and the nominal pressure at 20,000 ft. is 6.75 psi. The difference in pressure between these two altitudes is 7.95 psi. With this range, the appropriate pressure sensor should be an absolute version that is referenced to an on-chip vacuum and have a range up to 15psi. Since the change in pressure for major weather changes is approximately 0.18 psi, a resolution of 0.015 psi is over ten times more accurate than the measured value. The circuit that will be used for this design discussion is shown in Figure 9. The critical pressure sensor specifications for this application include the operating pressure range, sensitivity, room temperature (25°C) span and offset errors as well as over temperature (see Table 1). Although, the range of this sensor extends from 0 psi to 15 psi, this application will not be using that lower range. The minimum differential output voltage from the sensor will be 40.5mV (6.75psi or 20,000 ft.) and the maximum sensor voltage will be 88.2mV (14.7psi or sea level). The voltage at the output of the sensor is gained before it is digitized using an instrumentation amplifier. Table 1: The specifications of the SCX015 from SenSym indicates that this is a good pressure sensor that can be used to measure barometric pressure. The specifications of the SCX015 from SenSym indicates that this is a good pressure sensor that can be used to measure barometric pressure. Note that temperature issues are beyond the scope of this application note. Detailed information about temperature sensing circuits can be found in Microchip’s AN679, AN684, AN685, and AN687. Figure 9: The voltage at the output of the SCX015 pressure sensor is gained by the instrumentation amplifier (A1 and A2) then filtered, gained and level shifted (A4) with a 2nd order low pass filter (A3) and digitized with a 12-bit A/D converter (A5). Parameter (w/ 5V excitation) Specification Operating Pressure Range 0 to 15 psi Sensitivity 6.0mV/psi (typ) Full-Scale Span 90mV (typ) Zero Pressure Offset ± 300µV Temperature Effect on Span (0 to 70°C) ± 0.5% FSO Temperature Effect on Offset (0 to 70°C) ± 500µV A1 RG R2 R1 R1 R2 R1 R2 VDD SCX 015 SCK SI CS 68.1kΩ 10kΩ A5 35.7kΩ R1 = 30kΩ R2 = 10kΩ RG = 1.15 kΩ R3 = 95.7 kΩ R4 = 172 kΩ R5 = 304 kΩ C1 = 0.22 µF C2 = 0.22 µF A1 = A2 = A3 = Single Supply, CMOS op amp A4 = 12-bit, A/D SAR Converter A5 = 10kΩ Digital Potentiometer Level Shift Voltage and Offset Adjust A4 and A6 can be replaced with a PICmicro that has an on-chip A/D converter A3 A4 PIC16C6xx - MCP 606 + - 1/2 MCP602 + A2 - 1/2 MCP602 + Instrumentation Amplifier 2nd Order Butterworth Low Pass Filter R1 C2 R3 R5 R4 C1 2 3 4 1 8 VDD = 5V MCP3201 SCLK DOUT CS A6 R2
AN695 Instrumentation Amplifier Design Any cut-off frequency lower than 10Hz,requires capac- This sensor requires voltage excitation.In order to itors that are too large,making the board implementa- determine the required gain of the circuit in Figure 9 the tion awkward.The total attenuation between 10Hz and relationship between the maximum sensor output and 60Hz is approximately-30dB.In other words,a 60Hz allowable instrumentation amplifier output is used in signal that is part of the output signal of the instrumen- the calculation.As stated previously,the maximum dif- tation amplifier is attenuated by 0.031 times.Keeping in ferential output of the sensor is 88.2mV.The allowable mind that the instrumentation amplifier has already output range of the instrumentation amplifier is equal to rejected a major portion of any 60Hz common-mode VDD-100mV.In a five volt system where VDD=5V,the signal,this level of attenuation is enough to remove any amplifier output maximum is equal to 4.9V.The mini- remaining 60Hz noise that exists in the signal path. mum output of the sensor is 40.5mV.Since this is a The gain and offset adjust features of this filter are also positive voltage and the instrumentation amplifier is in used in this segment of the application circuit.Given a single supply environment,this minimum sensor out- that the output from the instrumentation amplifier is put voltage will not drive the output of the instrumenta- 2.3V to 4.9V,the peak-to-peak voltage of this signal is tion amplifier below ground.Consequently,the 2.6V.A gain of 1.8V/V will produce an output swing of reference voltage called out in Figure 3 and Figure 4 is approximately 4.7V peak-to-peak.The adjustable off- made to be equal to ground. set voltage of this circuit which is gained by 2.8V/will Gain is calculated by dividing the maximum output volt- be configured to insure that the signal will fall at the out- age with the maximum input voltage.Using this calcu- put of the amplifier between the supplies.This adjust- lation,the appropriate gain for our system is 55.6V/V. ment circuit can also be used to remove system offset By using the gain formula in Figure 5: errors that originate in the sensor or instrumentation amplifier. 2R1 Ra= The filter circuit in Figure 9 can be designed with the Gain- FilterLabTM software from Microchip Technology.The two capacitors are adjusted using the FilterLab pro- gram to be equal to 0.22uF.This adjustment is made in If R 30kQ and R2 =10k2, order to keep the capacitor packages small enough so surface mount capacitors can be used. 60k2 The offset adjust of the filter circuit is implemented with Rg=55.6-4) a 10ko digital potentiometer in series with a 68.1k and 35.7k resistors.The range of the offset adjust portion of this circuit at the wiper of the digital potenti- 1.15k (closest 1%value) ometer is from 3.0V to 3.4V.This offset circuitry is gained by the filter/amplifier circuit so that the nominal value of the offset circuitry in combination with the sen- With this gain,the maximum output of the lA will be 88.2mV*55.6V/V or 4.9V and the minimum output will sor signal is equal to: be40.5mV*55.6V/Vor2.3V. VoUT-FILTER=-1.8V/V(Nominal Input Signal) 2.8V/V(Nominal Reference voltage) Since the gain of this instrumentation amplifier stage is relatively large,it is desirable to use an amplifier that VoUT-FILTER =-1.8V/V(3.6V)+2.8V/V (3.2V) has a low offset voltage.The MCP607,dual CMOS VOUT-FILTER =2.48V amplifier has a guaranteed input offset voltage of 250uV(max).This amplifier's low quiescent current of A key amplifier specification for this filtering circuit is input voltage noise.The MCP601,single CMOS ampli- 25uA(max)make this device attractive for battery pow- ered applications fier has a typical noise density of 29 nV/Hz 1kHz. Filter Design A/D Converter Design Now that the signal from the pressure sensor has been The final design step for this analog signal path is to properly differentiated and gained,noise is removed to insert the analog-to digital converter.The converter making the results from the 12-bit A/D conversion quantizes a continuous analog signal into discrete repeatable and reliable.Remember that the output of buckets.The appropriate converter can be selected the instrumentation amplifier circuit does not swing a once it is determined how many bits the application requires. full OV to 5V.Consequently,the filter stage will also be used to implement a second gain cell as well as offset The range of the analog signal has been closely adjust. matched to the input range of a zero to 5V in A/D con- The stop frequency of this filter is 60Hz.This will verter. removes any mains frequencies that may be aliased The barometric pressure range is 14.7 psi to 6.75 psi. back into the signal path during conversion.This being The expected increase from good weather to a strong the case,the cut-off frequency is selected to be 10Hz. storm system would be approximately 0.18 psi.Given 2000 Microchip Technology Inc. Preliminary DS00695A-page 9
2000 Microchip Technology Inc. Preliminary DS00695A-page 9 AN695 Instrumentation Amplifier Design This sensor requires voltage excitation. In order to determine the required gain of the circuit in Figure 9 the relationship between the maximum sensor output and allowable instrumentation amplifier output is used in the calculation. As stated previously, the maximum differential output of the sensor is 88.2mV. The allowable output range of the instrumentation amplifier is equal to VDD - 100mV. In a five volt system where VDD = 5V, the amplifier output maximum is equal to 4.9V. The minimum output of the sensor is 40.5mV. Since this is a positive voltage and the instrumentation amplifier is in a single supply environment, this minimum sensor output voltage will not drive the output of the instrumentation amplifier below ground. Consequently, the reference voltage called out in Figure 3 and Figure 4 is made to be equal to ground. Gain is calculated by dividing the maximum output voltage with the maximum input voltage. Using this calculation, the appropriate gain for our system is 55.6V/V. By using the gain formula in Figure 5: With this gain, the maximum output of the IA will be 88.2mV*55.6V/V or 4.9V and the minimum output will be 40.5mV*55.6V/V or 2.3V. Since the gain of this instrumentation amplifier stage is relatively large, it is desirable to use an amplifier that has a low offset voltage. The MCP607, dual CMOS amplifier has a guaranteed input offset voltage of 250µV (max). This amplifier’s low quiescent current of 25µA (max) make this device attractive for battery powered applications. Filter Design Now that the signal from the pressure sensor has been properly differentiated and gained, noise is removed to making the results from the 12-bit A/D conversion repeatable and reliable. Remember that the output of the instrumentation amplifier circuit does not swing a full 0V to 5V. Consequently, the filter stage will also be used to implement a second gain cell as well as offset adjust. The stop frequency of this filter is 60Hz. This will removes any mains frequencies that may be aliased back into the signal path during conversion. This being the case, the cut-off frequency is selected to be 10Hz. Any cut-off frequency lower than 10Hz, requires capacitors that are too large, making the board implementation awkward. The total attenuation between 10Hz and 60Hz is approximately -30dB. In other words, a 60Hz signal that is part of the output signal of the instrumentation amplifier is attenuated by 0.031 times. Keeping in mind that the instrumentation amplifier has already rejected a major portion of any 60Hz common-mode signal, this level of attenuation is enough to remove any remaining 60Hz noise that exists in the signal path. The gain and offset adjust features of this filter are also used in this segment of the application circuit. Given that the output from the instrumentation amplifier is 2.3V to 4.9V, the peak-to-peak voltage of this signal is 2.6V. A gain of 1.8V/V will produce an output swing of approximately 4.7V peak-to-peak. The adjustable offset voltage of this circuit which is gained by 2.8V/V will be configured to insure that the signal will fall at the output of the amplifier between the supplies. This adjustment circuit can also be used to remove system offset errors that originate in the sensor or instrumentation amplifier. The filter circuit in Figure 9 can be designed with the FilterLabTM software from Microchip Technology. The two capacitors are adjusted using the FilterLab program to be equal to 0.22uF. This adjustment is made in order to keep the capacitor packages small enough so surface mount capacitors can be used. The offset adjust of the filter circuit is implemented with a 10kΩ digital potentiometer in series with a 68.1kΩ and 35.7kΩ resistors. The range of the offset adjust portion of this circuit at the wiper of the digital potentiometer is from 3.0V to 3.4V. This offset circuitry is gained by the filter/amplifier circuit so that the nominal value of the offset circuitry in combination with the sensor signal is equal to: VOUT-FILTER = -1.8V/V (Nominal Input Signal) + 2.8V/V (Nominal Reference voltage) VOUT-FILTER = -1.8V/V (3.6V) + 2.8V/V (3.2V) VOUT-FILTER = 2.48V A key amplifier specification for this filtering circuit is input voltage noise. The MCP601, single CMOS amplifier has a typical noise density of 29 nV/√Hz @ 1kHz. A/D Converter Design The final design step for this analog signal path is to insert the analog-to digital converter. The converter quantizes a continuous analog signal into discrete buckets. The appropriate converter can be selected once it is determined how many bits the application requires. The range of the analog signal has been closely matched to the input range of a zero to 5V in A/D converter. The barometric pressure range is 14.7 psi to 6.75 psi. The expected increase from good weather to a strong storm system would be approximately 0.18 psi. Given RG 2R1 Gain 1 R1 R2 – – ------ ---------------------------------------- If R1 30kΩ and R2 10kΩ, RG 60kΩ (55.6 – 4) -------------------------- 1.15kΩ (closest 1% value) = = = = =
AN695 this,the equipment should resolve to at least 0.015psi. CONCLUSION This is easily achieved with a 10-bit converter.If reso- lution to 0.002 is needed a 12-bit a/d converter would The design challenge that has been tackled in this be more suitable. application note is gaining,filtering,and digitizing the Microchip has a large variety of analog to digital con- small differential signal of a pressure sensor bridge.In verters that can be used for this application.If the order to achieve this goal,we used a two-op amp stand-alone solution is appealing,the MPC320X family instrumentation amplifier which gained the differential signal from the pressure sensor and converted it to a of 12-bit and the MCP300X 10-bit family of converters are available signal ended output.After this gain stage,a 2nd order, Butterworth,anti-aliasing filter was used to reduce Generally speaking,stand-alone A/D converters have noise so that the A/D converter could achieve a full 10- better accuracy than those compared to on-board con- bit accuracy. verters.They also have features such a pseudo differ- ential inputs and faster conversion speeds.The pseudo The suggested A/D converter strategy could be on differential capability of these devices allow for config- board or off board and the trade-offs were presented. urations that reject small common mode signals.Addi- Digital filtering was not needed in this application. tionally,the single channel devices can be used in REFERENCES simultaneous sampling applications,such as motor control.The application circuits using the singe con- Tandeske,Duane,Pressure Sensors,Marcel Dekker, verter also require fewer analog filters because the lnc.,1991 multiplexer is typically placed before the anti-aliasing filter. "Anti-Aliasing Analog Filters for Data Acquisition Sys- tems",Baker,Bonnie C.,AN699,Microchip Technology If an on-board a/d converter fits the application better, Inc. the PICmicro line has a large array of converters com- bined with other peripherals on a variety of micros that "Making a Delta-Sigma converter with a Microcontrol- ler",Baker,Peter,Darmawaskita,Butler,AN700,Micro- can be used. chip Technology,Inc. The integrated solution offers a degree of flexibility that the stand-alone solution does not have.This flexibility "Using Operational Amplifiers for Analog Gain in comes in the form of operational flexibility where the Embedded System Design",Baker,Bonnie C.,AN682, device's voltage reference and sampling speed can be Microchip Technology,Inc. reconfigured on the fly.The l/O configuration is also "Building a 10-bit Bridge Sensing Circuit using the very flexible allowing for easy implementation of the PIC16C6XX and MCP601 Operational Amplifier", board layout. Baker,Bonnie C.,AN717,Microchip Technology,Inc. The stand-alone and integrated A/D converters from "Precision Temperature Sensing with RTD Circuits", Microchip are both suitable for the pressure sensor cir- Baker,Bonnie C.,AN687,Microchip Technology,Inc. cuit that is shown in Figure 9. "Temperature Sensing Technologies",Baker,Bonnie C.,AN679,Microchip Technology,Inc. "Thermistors in Single Supply Temperature Sensing Circuits",Baker,Bonnie C..AN685,Microchip Technol- ogy,Inc. "Single Supply Temperature Sensing with Thermocou- ples",Baker,Bonnie C.,AN684,Microchip Technology, Inc. DS00695A-page 10 Preliminary @2000 Microchip Technology Inc
AN695 DS00695A-page 10 Preliminary 2000 Microchip Technology Inc. this, the equipment should resolve to at least 0.015psi. This is easily achieved with a 10-bit converter. If resolution to 0.002 is needed a 12-bit a/d converter would be more suitable. Microchip has a large variety of analog to digital converters that can be used for this application. If the stand-alone solution is appealing, the MPC320X family of 12-bit and the MCP300X 10-bit family of converters are available. Generally speaking, stand-alone A/D converters have better accuracy than those compared to on-board converters. They also have features such a pseudo differential inputs and faster conversion speeds. The pseudo differential capability of these devices allow for configurations that reject small common mode signals. Additionally, the single channel devices can be used in simultaneous sampling applications, such as motor control. The application circuits using the singe converter also require fewer analog filters because the multiplexer is typically placed before the anti-aliasing filter. If an on-board a/d converter fits the application better, the PICmicro line has a large array of converters combined with other peripherals on a variety of micros that can be used. The integrated solution offers a degree of flexibility that the stand-alone solution does not have. This flexibility comes in the form of operational flexibility where the device’s voltage reference and sampling speed can be reconfigured on the fly. The I/O configuration is also very flexible allowing for easy implementation of the board layout. The stand-alone and integrated A/D converters from Microchip are both suitable for the pressure sensor circuit that is shown in Figure 9. CONCLUSION The design challenge that has been tackled in this application note is gaining, filtering, and digitizing the small differential signal of a pressure sensor bridge. In order to achieve this goal, we used a two-op amp instrumentation amplifier which gained the differential signal from the pressure sensor and converted it to a signal ended output. After this gain stage, a 2nd order, Butterworth, anti-aliasing filter was used to reduce noise so that the A/D converter could achieve a full 10- bit accuracy. The suggested A/D converter strategy could be on board or off board and the trade-offs were presented. Digital filtering was not needed in this application. REFERENCES Tandeske, Duane, Pressure Sensors, Marcel Dekker, Inc., 1991 “Anti-Aliasing Analog Filters for Data Acquisition Systems”, Baker, Bonnie C., AN699, Microchip Technology Inc. “Making a Delta-Sigma converter with a Microcontroller”, Baker, Peter, Darmawaskita, Butler, AN700, Microchip Technology, Inc. “Using Operational Amplifiers for Analog Gain in Embedded System Design”, Baker, Bonnie C., AN682, Microchip Technology, Inc. “Building a 10-bit Bridge Sensing Circuit using the PIC16C6XX and MCP601 Operational Amplifier”, Baker, Bonnie C., AN717, Microchip Technology, Inc. “Precision Temperature Sensing with RTD Circuits”, Baker, Bonnie C., AN687, Microchip Technology, Inc. “Temperature Sensing Technologies”, Baker, Bonnie C., AN679, Microchip Technology, Inc. “Thermistors in Single Supply Temperature Sensing Circuits”, Baker, Bonnie C., AN685, Microchip Technology, Inc. “Single Supply Temperature Sensing with Thermocouples”, Baker, Bonnie C., AN684, Microchip Technology, Inc