VGA_noise

download VGA_noise

of 3

Transcript of VGA_noise

  • 7/29/2019 VGA_noise

    1/3Analog Dialogue 47-02 Back Burner, February (2013) 1

    Understanding How Amplifer

    Noise Contributes to Total

    Noise in ADC Signal ChainsBy Umesh Jayamohan

    Introduction

    Analog-to-digital converters (ADCs) provide optimal perormancewhen the analog inputs are driven to their rated ull-scale inputvoltage, but in many applications, the maximum available signaldiers rom the specifed voltage and may need to be adjusted.A useul device or handling this requirement is a variable-gainamplifer(VGA). Understanding how the VGA aects the ADCsperormance will help in optimizing perormance o the entiresignal chain.

    This article analyzes noise in a circuit that uses the AD9268dual16-bit, 125-/105-/80-MSPS, pipelined ADC, and the AD8375ultralow distortion IF VGA. The signal chain includes a VGAusedat a gain setting o +6 dB, a 5th-order Butterworth low-pass flterwith a 3-dB roll-o at 100 MHz, and the ADC. Noise calculationswill be shown or the amplifer and flter, as they dictate the dynamic

    perormance o the ADC over the band o interest.Problem

    Many real-world applications using high-speed ADCs needsome sort o driver, amplifer, or gain block that scales the inputsignal to the ull-scale analog input range1 to ensure optimumsignal-to-noise ratio (SNR) and spurious-ree dynamic range(SFDR). In addition, a dierential amplifer could also convert asingle-ended signal to a dierential signal to drive the ADC. Beingactive, these components contribute noise at the ADC ront end.The integration o this noise over the operational bandwidth willdegrade the conversion perormance.

    The choice o an appropriate ADC or an application depends onmany actors, including:

    Analog input range

    Input requency/bandwidth

    Desired resolution/SNR

    Desired SFDR

    Some applications require both high dynamic range and highresolution. The AD9268 is very good choice or these applications,delivering 78.2-dBFS (dB relative to ull-scale) SNR and 88-dBcSFDR at a 70-MHz intermediate requency.

    At the system level, the ADC ront-end could use an amplifer, atransormer, or a balun, but implementations using an amplifer

    are most common. An amplifer could be used or one or more othe ollowing reasons:

    Provide gain to the input signal to increase the ADC resolution.

    Buer or transorm the impedance between the input sourceand the ADC.

    Convert a single-ended input signal to a dierential output signal.

    www.analog.com/analogdialogue

    The AD8375 VGA, which maintains high linearity and uniormnoise perormance over its various gain settings, can be used toconvert single-ended signals to di erential. These characteristicmake it a good candidate or driving the ADC at higher intermediate requencies. Unortunately, the presence o an activedevicethe ampliferin the signal chain can limit the ADCsperormance.

    Example

    Figure 1 shows the circuit topology used in perorming the noise

    calculations. The AD8375 has a high-impedance dierentiaoutput (16 k||0.8 pF). A 5th-order low-pass antialiasing flte(AAF) with 100-MHz bandwidth and 150- input and outpuimpedances interaces the amplifer to the ADC. The requencyresponse o the circuit shown in Figure 1 can be seen in Figure 2

    16k

    5

    300

    75

    75

    6.8pF

    180nH

    180nH

    180nH

    180nH

    20pF6.8pF 3.5pF

    ADC INPUTIMPEDANCE

    @100MHz

    3.5k

    5

    0.8pF

    +5V

    +5V

    VCM CLK

    +5V

    0.1F0.1F

    TC3-1T+

    0.1F

    0.1F

    0.1F

    0.1F

    125MHz

    1H

    1H +1.8V

    AD9268-125

    AD8375

    Figure 1. The AD8375, AAF, and AD9268 signal chain.

    0

    1001 1k

    AMPLITUDE(dBFS)

    ANALOG INPUT FREQUENCY (MHz)

    10 100

    10

    20

    30

    40

    50

    60

    70

    80

    90

    3dB @ ~105MHz

    Figure 2. Frequency response of the AD8375, AAF,

    and AD9268 signal chain.

    Performance

    System designers may not expect the amplifer driving the ADCinputs to degrade the overall dynamic perormance o the systembut the act that a driver and ADC combo was chosen or oneapplication does not mean that it will provide the same greatperormance in another application. The technique described hereallows a systems engineer to estimate the expected perormancebeore choosing an amplifer.

    http://www.analog.com/en/analog-to-digital-converters/ad-converters/ad9268/products/product.htmlhttp://www.analog.com/en/rfif-components/rfif-attentuators-vga-filters/ad8375/products/product.htmlhttp://www.analog.com/en/rfif-components/rfif-attentuators-vga-filters/ad8375/products/product.htmlhttp://www.analog.com/en/analog-to-digital-converters/ad-converters/ad9268/products/product.html
  • 7/29/2019 VGA_noise

    2/32 Analog Dialogue 47-02 Back Burner, February (2013)

    Figure 3 shows two dierent setups: Figure 3(a), which usespassive coupling to the converter, is avai lable as the deault optionon the customer evaluation board. Thepassive ront-end networkuses a transormer or balun to convert the single-ended signal todierential, along with a passive low-pass flter that rolls o atabout 200 MHz. Figure 3(b) shows the optional ampliferpath.A comparison o the noise contributed by these two setups ollows.A single-toneast Fourier transorm (FFT) at a low intermediaterequency (10 MHz) is used to calculate the noise added bythe amplifer.

    AAF

    BW = 105MHz

    fCLK= 125MHz

    BPF

    SIGNALGENERATOR

    AMPLIFIER AD9268-125

    (a)

    (b)

    AAF

    BW = 200MHz

    fCLK= 125MHz

    BPF

    SIGNALGENERATOR

    AD9268-125

    Figure 3. Typical ADC front-ends: (a) passive. (b) active.

    Two techniques are typically used or analyzing the noise, buteach can be cumbersome.Noise spectral density (NSD) defnes thenoise power per unit bandwidth. It is represented in mean-squaredBm/Hz or dBFS/Hz or ADCs and rms nV/ Hz or amplifers.This incompatibility in units provides an obstacle to calculatingsystem noise when an amplifer is driving an ADC.

    Noise fgure (NF) is the log ratio o input SNR to the output SNRexpressed in decibels. This specifcation, commonly used by RFengineers, makes sense in a purely RF world, but attempting touse NF calculations in a signal chain with an ADC can lead tomisleading results.2

    An alternate, but more eective, technique is to de-normalizethe noise density, representing it as an rms noise voltage ratherthan a mean-square voltage. This method, as described here, isstraightorward and allows a clear analysis o the system noise.

    Figure 4 and Figure 5 show the low-requency single-tone FFTs othe two ront ends. Note that the passive ront end has 77.7-dBFSSNR, while the active ront end has 72.5-dBFS SNR, 5.2 dBFSlower than the ADCs expected perormance.

    0

    1400 62.5

    AMPLITUDE(dBFS)

    FREQUENCY (MHz)

    20

    40

    60

    80

    100

    120

    12.5 25.0 37.5 50.0

    SAMPLES = 32kFUNDAMENTAL POWER = 1dBFSSNR = 77.7dBFSNOISE/Hz = 155.7dBFS/Hz

    Figure 4. FFT of a 10-MHz analog input tone for the

    circuit of Figure 3a.

    0

    1400 62.5

    AMPLITUDE(dBFS)

    FREQUENCY (MHz)

    20

    40

    60

    80

    100

    120

    12.5 25.0 37.5 50.0

    SAMPLES = 32kFUNDAMENTAL POWER = 1dBFSSNR = 72.5dBFSNOISE/Hz = 150.45dBFS/Hz

    Figure 5. FFT of a 10-MHz analog input tone for the

    circuit of Figure 3b.

    Analysis

    The only dierence between the setups shown in Figure 3a and3b is the addition o the amplifer in the signal chain, so it is saeto assume that the perormance degradation is caused by theamplifers noise. The ollowing calculations help to understandthe additional noise introduced by the amplifer.

    First, use the converters ull-scale dierential input voltage, asspecifed in the data sheet. Convert the peak-to-peak voltage torms by dividing by 2 2 to get 0.707 V rms.

    22VRMS=

    Vpp

    (1)

    The contribution o the converters noise based on the ADCstypical SNR at 10 MHz is

    dBFS

    VFS,RMS

    VNOISE, RMSSNR = 20 log (2)

    VNOISE, ADC= 92.2VRMS

    2010

    SNRVNOISE, RMS=

    VFS, RMS

    (3)

    Using VNOISE, ADC = 92.2 V rms and system SNR with the ampli-fer ront end = 72.5 dBFS, the system noise, using Equation 3,

    is 168 V rms.

    2010

    72.5VNOISE, TOTAL=

    0.707= 168 VRMS

    (4)

    VNOISE, AMP= 140VRMS

    VNOISE, TOTAL= (VNOISE, AMP)2 + (VNOISE, ADC)2

    (5)

  • 7/29/2019 VGA_noise

    3/3Analog Dialogue 47-02 Back Burner, February (2013) 3

    The system noise obtained rom Equation 4 is the combined noiseo the ADC and the VGA. The amplifer noise can be calculatedrom Equation 5 to be 140V rms. This calculation shows that theamplifer noise is at least 50% greater than the ADC noise, makingit the limiting actor in determining the systems ac perormance.

    Note that we must determine whether the value o V NOISE, AMPcalculated above matches the amplifers data sheet. The specifednoise spectral density is about 20 nV/ Hz with a 150- dierentialoutput impedance.

    Although the data sheet specifes that the VGAs noise is airlyconstant with gain, this noise will change with the load, so the noisespectral density should be scaled to the total impedance driven bythe amplifer outputs. Because the di erential output impedanceo the amplifer is large (16 k||0.8 pF), the impedance seen bythe amplifer (see Figure 1) can be calculated as

    [10 + (300 ||150 ||3.5 k)] = 107 .

    Using this number, the derated noise spectral density or theAD8375 in this application can be ound rom Equation 6:

    HzHz

    nVnVAmplifierNSD

    150

    10720 == 14.3

    ShapeFactorBWHz

    nV

    VNOISE, AMP(Calculated) =NSD

    (6)

    Note that when calculating system noise using a real flter, thenoise bandwidth takes on a dierent shape than that o an idealflter. This deviation in requency response is characterized bythe term, shape actor, and accounts or the noise in the roll-oregion. The shape actor, which depends on the order o the flter,is the ratio o the noise bandwidth to the 3-dB bandwidth.3 Themore poles in the flter, the closer the shape actor is to unity. Thisrelationship can be seen in Table 1.

    Table 1. Relationship Between

    System Order and Shape Factor

    System Order Shape Factor

    1 1.572 1.11

    3 1.05

    4 1.03

    5 1.02

    In the example o Figure 1, the shape actor is 1.02. UsingEquation 6, the noise injected by the amplifer will be;

    105 MHz 1.02 = 149VRMHz

    nVVNOISE, AMP(Calculated) = 14.3

    VNOISE, AMP(Calculated) = 149VRMS

    This estimated value o noise injected into the system by the VGAmatches very well with the measured value using Equation 5

    proving that the perormance o the signal chain comprising theAD8375 and AD9268 is dominated by the amplifer.

    Conclusion

    In many cases, an ampliferVGA or gain blockis requiredto drive a ull-scale signal to an ADC in the system signal chainThe systems designer must be aware o the degradation in optimaperormance o the ADC caused by the choice o the ampliferBeore designing with the chosen amplifer and ADC, the designecan use the method shown here to calculate the noise contributiono the amplifer to estimate the expected dynamic perormance, ascharacterized by the SNR, or the intended system implementation

    References1MT-006, ADC Noi se Figure An O ten Misunderstood andMisinterpreted Specifcation.

    2The Data Conversion Handbook.

    3 Reeder, Rob and Jim Caserta, Ask The Application Engineer 36Wideband A/D Converter Front-End Design Considerations IIAmplifer-or Transormer Drive or the ADC?Analog Dialogue41-02, 2007.

    Author

    Umesh Jayamohan [[email protected]] is anapplications engineer in the High Speed Converter Group atAnalog Devices in Greensboro, NC. Umeshreceived a BSEE rom University o Kerala, Indiain 1998 and a MSEE rom Arizona State Universityin 2002, and has over seven years o experience asa design and applications engineer.

    http://www.analog.com/static/imported-files/tutorials/MT-006.pdfhttp://www.analog.com/static/imported-files/tutorials/MT-006.pdfhttp://www.analog.com/static/imported-files/tutorials/MT-006.pdfhttp://www.analog.com/library/analogDialogue/archives/39-06/data_conversion_handbook.htmlhttp://www.analog.com/library/analogDialogue/archives/41-02/transformer_vs_amp.htmlhttp://www.analog.com/library/analogDialogue/archives/41-02/transformer_vs_amp.htmlhttp://www.analog.com/library/analogDialogue/archives/41-02/transformer_vs_amp.htmlmailto:umesh.jayamohan%40analog.com?subject=mailto:umesh.jayamohan%40analog.com?subject=http://www.analog.com/library/analogDialogue/archives/41-02/transformer_vs_amp.htmlhttp://www.analog.com/library/analogDialogue/archives/41-02/transformer_vs_amp.htmlhttp://www.analog.com/library/analogDialogue/archives/41-02/transformer_vs_amp.htmlhttp://www.analog.com/library/analogDialogue/archives/39-06/data_conversion_handbook.htmlhttp://www.analog.com/static/imported-files/tutorials/MT-006.pdfhttp://www.analog.com/static/imported-files/tutorials/MT-006.pdf