INTRODUCTION Size, output flexibility efficiency advantages have made
|
|
|
Top Searches for this datasheet
"Silence perfectest herald Williams
INTRODUCTION Size, output flexibility efficiency advantages have made switching regulators common electronic apparatus. continued emphasis these attributes resulted circuitry with efficiency that requires minimal board area. Although these advantages welcome, they necessitate compromising other parameters. Switching Regulator "Noise" Something commonly referred "noise" primary concern. switched mode power delivery that permits aforementioned advantages also creates wideband harmonic energy. This undesirable energy appears radiated conducted components commonly labeled "noise." Actually, switching regulator output "noise" really noise all, coherent, high frequency residue directly related regulator's switching.1 Figure shows typical switching regulator output noise. distinct characteristics present. slow, ramping output ripple caused finite storage capacity regulator's output filter components. quickly rising spikes associated with switching transitions. Figure shows another switching regulator output. this case ripple been eliminated adequate filtering linear postregulation, wideband spikes remain. these fast spikes that cause much difficulty systems. Their high frequency content often corrupts associated circuitry, degrading performance even disabling operation. Noise gets into adjacent circuitry three paths. conducted regulator output lead, conducted
Note Noise contains regularly occurring coherent components. such, switching regulator output "noise" misnomer. Unfortunately, undesired switching related components regulated output almost always referred "noise." such, although technically incorrect, this publication treats undesired output signals "noise." Appendix "Specifying Measuring Something Called Noise."
back driving source ("reflected" noise) radiated. multiple transmission paths combine with high frequency content make noise suppression difficult. Unconscionable amounts bypass capacitors, ferrite beads, shields, Mu-metal aspirin have been expended attempts ameliorate noise-induced effects.
10mV/DIV COUPLED)
1µs/DIV
AN70
Figure Typical Switching Regulator Output "Noise." Wideband Spikes Difficult Suppress, Causing System Interference Problems. Ripple Component Harmonic Content, Relatively Easily Filtered
20mV/DIV COUPLED)
50µs/DIV
AN70
Figure Linear Regulator Eliminates Ripple, Wideband Spikes Remain. Peak-to-Peak Amplitude Exceeds 30mV (Just Visible Near 2nd, Vertical Graticule Divisions)
registered trademarks Linear Technology Corporation.
AN70-1
Application Note
Alternate approaches involve synchronizing switching regulator operation host system turning switching during critical system operation "interrupt driven" power supply). Another approach places critical system operations between switch cycles, literally running between electronic rain drops.2 difficulty debugging noise-laden system compromises involved synchronized approaches could eliminated with noise switching regulator. inherently noise switching regulator most attractive approach because eliminates noise concerns while maintaining system flexibility. Noiseless Switching Regulator Approach inherently noise regulator minimize harmonic content switching transitions. Slowing down switching interval does this, although power dissipated during transition causes some efficiency loss. Reducing switch repetition rate largely offset losses, resulting reasonably efficient design with small magnetics desired noise. Noise reduction restricting harmonic generation been employed before, although implementations were complex narrowly applicable.3 monolithic approach, broadly usable over range magnetics applications, described here. Practical, Noise Monolithic Regulator Figure describes LT®1533, monolithic regulator designed noise switching supplies. Figure details functions. Figure functional blocks show fairly conventional push-pull architecture with major exception. push-pull approach good magnetics utilization (power transfer always occurring transformer; core does store energy) pulls current
Note References details practical examples these techniques. Note Appendix History Noise DC/DC Conversion." also References through
SHDN
PGND
REGULATOR
NEGATIVE FEEDBACK INTERNAL
CURRENT
OUTPUT DRIVERS
RVSL SLEW CONTROL RCSL
1.25V
OSCILLATOR SYNC
Figure LT1533 Simplified Block Diagram. Slew-Controlled Output Stages Provide Noise Switching
AN70-2
DUTY
AN70
ERROR
COMP
Application Note
Output transistor collectors which switch out-ofphase. DUTY: Grounding this forces outputs switch duty cycle. This must float used. SYNC: Used synchronize external clock. Float ground unused. Oscillator timing capacitor. Oscillator timing resistor. Used positive output voltage sensing. NFB: Used negative output voltage sensing. GND: Analog ground pin. PGND: High current ground return. Should returned ground 50nH trace wire, small ferrite bead). Appendix schematic (Figures notes details. some package options, this internally connected "GND" pin. Frequency compensation node. SHDN: Normally high. Grounding this shuts part down. ISHDN 20µA. RCSL: Current slew control resistor. RVSL: Voltage slew control resistor. VIN: Input supply pin. 2.7V range. Undervoltage lockout 2.55V. Figure LT1533 Short Form Function Descriptions
would cause excessive, quickly rising currents, degrading efficiency generating noise. design's most significant aspect output stage. Each power transistor operates inside broadband control loop. voltage across each transistor current through sensed loop controls slew rate each parameter. voltage current slew rates independently settable external programming resistors. This ability control switching's rate-ofchange makes noise switching regulation practical. Operating switching transistors local loop permits predictable, wide range control over variety situations.4 Figure 40kHz, converter using LT1533 push-pull, "forward" configuration. feedback resistor's ratio produces output. two-section filter provides high ripple attenuation, although single section will give good performance. particularly noteworthy that high frequency noise content opposed 40kHz fundamental related ripple) unaffected output filter characteristics. This simply because there little high frequency energy developed this circuit. there's nothing there, doesn't need filtered! provides compensation output current control loop. practice, length trace, small inductor, coiled section wire ferrite bead. Appendix "Magnetics Considerations" complete discussion.
Note Patent pending.
100µH OPTIONAL (SEE TEXT) 1N4148
continuously from source. even, continuous current drain from source eliminates fast, high peak currents required flyback other approaches. source sees benign load corrupted. switches also receive nonoverlapping drive, ensuring they conduct simultaneously. Simultaneous conduction
4.7µF 3300pF SHDN DUTY SYNC
PGND LT1533 RVSL RCSL 0.01µF
1N4148
100µH
47µF
47µF
2.49k 21.5k
AN70
COILTRONICS CTX100-3 22nH TRACE INDUCTANCE, FERRITE BEAD INDUCTOR (SEE APPENDIX COILCRAFT B-07T TYPICAL CTX02-13665-X1 (SEE APPENDIX DETAILS)
Figure 100µV Noise 5V-to-12V Converter. Output Section Deleted Frequency Ripple Acceptable
AN70-3
Application Note
Measuring Output Noise Measuring LT1533's unprecedented noise levels requires care.5 Figure shows test setup taking measurement. Good connection signal handling technique combined with judicious instrumentation choice should yield 100µV noise floor 100MHz bandwidth. This corresponds noise resistor 100MHz bandwidth. Before measuring regulator output noise, good practice verify test setup performance. This done running test setup with input. Figure shows noise base line 100µV 100MHz bandwidth, indicating instrumentation operating properly. Measuring Figure noise involves coupling circuit's output into test setup's input. Figure shows this. Coaxial connections must maintained preserve measurement integrity.6 Figure waveforms detail circuit operation. Traces switching transistor collector voltages, respective transistor currents. test setup's output, representing circuit output noise, Trace Wideband spiking ripple, just visible noise floor, inside 100µV, even 100MHz bandpass.7 This spectacularly good performance fact, actually better than photo shows. Removing probes from breadboard leaves only Trace coaxial connection. This eliminates possible ground loop-induced error.8 Figure 10's trace shows 40kHz ripple with about same amplitude Figure Switching related spikes, just faintly outline noise, reduced. Measurement bandwidth reduced 10MHz Figure attenuating test fixture amplifier noise. Switching ripple residue amplitude shape change, indicating signal activity beyond this frequency. Figure 12's
Note Equipment selection measurement techniques detailed Appendix "Specifying Measuring Something Called Noise." also Appendix "Probing Connection Techniques Level, Wideband Signal Integrity." Note Again, Appendices extended treatment these related issues. Note common industry practice specify switching regulator noise 20MHz bandpass. There only reason this, disservice users. Appendix tutorial observed noise versus measurement bandwidth. Note Appendix related discussion techniques triggering oscilloscopes without invasively probing circuit.
OSCILLOSCOPE 0.01V/DIV VERTICAL SENSITIVITY 100µV/DIV REFERRED AMPLIFIER INPUT
HP461A AMPLIFIER X40dB INPUT CABLE
TERMINATOR HP-11048C EQUIVALENT
AN70
Figure Test Setup Noise Baseline 100µVP-P 100MHz Bandwidth. Performance Resistor Noise Limited. Cable Connections Terminations Provide Coaxial Environment, Ensuring Wideband, Noise Characteristics
AN70-4
Application Note
100µV/DIV
100µV/DIV
20µs/DIV
AN70
10µs/DIV
AN70
Figure Oscilloscope Verifies Test Setup 100µV Noise Floor 100MHz Bandwidth. Indicated Noise That Resistor
Figure Removing Probes from Figure Test Eliminates Ground Loops, Slightly Reducing Observed Noise. Switching Artifacts Just Discernible Above Noise Floor
OSCILLOSCOPE 0.01V/DIV VERTICAL SENSITIVITY 100µV/DIV REFERRED AMPLIFIER INPUT
VOUT SWITCHING REGULATOR UNDER TEST CABLE CONNECTORS
COUPLING CAPACITOR HP-10240B
HP461A AMPLIFIER 40dB CABLE
LOAD DESIRED)
TERMINATOR HP-11048C EQUIVALENT
AN70
Figure Connecting Figure Circuit Test Setup. Coaxial Connections Must Maintained Preserve Measurement Integrity
10V/DIV 1A/DIV 10V/DIV 1A/DIV 100µV/DIV 10µs/DIV
AN70
100µV/DIV
20µs/DIV
AN70
Figure Waveforms Figure 100mA Loading. Traces Voltage; Current, Respectively. Switching Transistion's Noise Signature Appears Trace Circuit's Output Noise
Figure Reducing Measurement Bandwidth 10MHz Attenuates Amplifier Noise. Switching Residue Characteristics Remain Unchanged, Indicating Signal Activity Beyond This Frequency
AN70-5
Application Note
horizontal expansion Figure returns 100MHz bandpass. switching spike appears center screen region. 2µs/division sweep, there wideband activity observable. Figure 10MHz bandpass version Figure retains signal information, further suggesting signal power beyond 10MHz. Figure noise floor HP4195A spectrum analyzer 500MHz sweep. When Figure circuit coupled into analyzer, output (Figure 15a) essentially identical. analyzer unable detect switching-induced noise 500MHz bandpass. Some 40kHz fundamental-related components detectable Figure 15b's 1MHz wide plot, although rest sweep analyzer noise limited. Additional filtering linear postregulator could eliminate 40kHz ripplerelated residue desired. preamplified oscilloscope more sensitive tool these measurements because triggered operation advantage synchronous detection. This demonstratable free running preamplified oscilloscope sweep; switching-related components indistinguishable noise background.
Figure Noise Floor Test Fixture HP-4195A Spectrum Analyzer 500MHz Sweep
100µV/DIV
Figure 15a. Figure Circuit Connected Spectrum Analyzer Produces Essentially Identical Results Figure Circuit's Noise Undetectable
2µs/DIV
AN70
Figure Horizontal Expansion Figure Shows Wideband Components. Switching Originated Noise Appears Center Screen Region
100µV/DIV
2µs/DIV
AN70F13
Figure 10MHz Band Limited Version Figure Before, Signal Information Retained, Although Amplifier Noise Reduced. Results Indicate Signal Power Beyond 10MHz
Figure 15b. Reducing Analyzer Sweep 1MHz Width Reveals 40kHz Related Components. Remainder Plot Analyzer Noise Floor Limited, Even Sensitive 455kHz Band
AN70-6
Application Note
Figure studies ripple first filter section output. ripple's 40kHz fundamental clearly seen, although wideband spikes visible. Figure horizontally expands Figure 14's time scale, high frequency harmonics spikes observable. frequency noise rarely concern, although Figure shows inside 50µV 10Hz 10kHz bandpass. Input current noise usually more interest. Excessive "reflected" noise corrupt regulator's driving source, causing system level interference. Figure shows Figure input current with small, 40kHz fundamentalrelated sinusoidal component. There high frequency content, sinusoidal variations easily handled driving source. System-Based Noise "Measurement" final analysis, effect switching regulator output noise system powering ultimate test. Appendix "System-Based Noise "Measurement," presents results when LT1533 used power 16-bit converter. Transition Rate Effects Noise Efficiency theory, simply setting transition rate values will achieve noise. Practically, such approach, while workable, wastes power during transitions, lowering efficiency. good compromise sets transition time fastest rate permitting desired noise performance. LT1533's slew adjustments allow easy determination this point. Figure 20's photographs dramatically demonstrate relationship between transition time output noise Figure circuit. sequence shows noise reduction switch transition time slows from 100ns (20a) (20d). Figure 20d's displayed noise actually lower, probing-induced error caused monitoring switch corrupts measurement.9 Figure graphically summarizes Figure 20's information. Significant noise reduction coincides with descending transition slew time until about 1.3µs. Little additional noise benefit occurs beyond this point. Figure shows efficiency fall-off with slew time. There penalty between 100ns 1.3µs, same region where noise performance improves factor (per previous
Note Appendix "Probing Connection Techniques Level, Wideband Signal Integrity" guidance.
5mV/DIV
10µs/DIV
AN70
Figure Ripple Figure First Output Wideband Spikes
5mV/DIV
2.5µs/DIV
AN70
Figure Time Expansion Previous Figure. High Frequency Content Visible
50µV/DIV
10ms/DIV
AN70
Figure Frequency Noise 10Hz 10kHz Bandpass
10mA/DIV COUPLED 200mA LEVEL
10µs/DIV
AN70
Figure Figure Small Sinusoidal Input Current Variations Contain High Frequency Content Easily Absorbed Input Supply
AN70-7
Application Note
5V/DIV 5V/DIV
100µV/DIV
100µV/DIV
500ns/DIV
AN70 F20a
500ns/DIV
AN70 F20b
5V/DIV 5V/DIV
100µV/DIV
100µV/DIV
500ns/DIV
AN70 F20a
500ns/DIV
AN70 F20a
Figure Output Noise (Trace Different Switch Slew Rates (Trace Highest Slew Rate (Figure Causes Largest Noise. Retarding Slew Rate (Figures Decreases Noise Until Lowest Noise Performance Achieved (Figure
PEAK-TO-PEAK WIDEBAND NOISE (µV)
EFFICIENCY
1200 SLEW TIME (ns) 1600 2000
AN70
1.3µs SLEW TIME <100µV NOISE (SEE FIGURE
EFFICIENCY
1.3µs SLEW TIME
40kHz ILOAD 100mA
40kHz ILOAD 100mA
40kHz ILOAD 100mA
1200 SLEW TIME (ns)
1600
2000
AN70
WIDEBAND PEAK-TO-PEAK NOISE (µV)
AN70
Figure Figure Noise Slew Time 40kHz Switching Frequency. Noise Reduction Beyond 1.3µs Minimal
Figure Figure Efficiency Drops Slew Time Extends 1.3µs. Operation Beyond This Point Gains Little Noise Performance (See Previous Curve) with Efficiency Penalty
Figure Efficiency Noise Figure Data Shows Significant Efficiency FallOff Noise Below 80µV
AN70-8
Application Note
figure). There additional penalty beyond 1.3µs, although significant noise reduction occurs (again, Figure 21). such, operation this region undesirable. Figure clearly shows inflection point efficiency versus noise trade-off.10 Negative Output Regulator LT1533 separate feedback input that directly accepts negative inputs.11 This permits negative outputs without usual discrete level shifting stage. Figure 24's converter similar Figure circuit, except that negative output back negative feedback input. feedback scale factor change necessitated higher effective reference voltage. other respects, circuit (and performance) similar Figure Floating Output Regulator Figure 25's isolation stage permits fully floating, regulated output. LT1431 shunt regulator compares portion output internal reference drives optoisolator with error signal. optoisolator's collector output biases LT1431's pin, closing feedback loop regulate circuit output. 0.22µF capacitor stabilizes loop 240k resistor biases optoisolator into favorable operating region. This circuit's operation characteristics similar Figure with added benefit isolated output. Floating Bipolar Output Converter Grounding LT1533's "DUTY" biasing forces device into duty cycle mode. Figure 26's output full wave rectified with respect T1's secondary center tap, producing bipolar outputs. forced duty cycle combined with feedback means outputs unregulated, proportioning T1's drive voltage. output inductor usually required, Figure "forward" converter. very highest output currents, some inductance necessary limit inrush current. this done, circuit start. Typically, linear regulators provide regulation.12 Figure 26's waveforms appear Figure Collector voltage (Traces current (Traces shown, along with indicated output noise (Trace this case linear regulators output filter use. Figure probes except coaxial output connection removed. This eliminates probing induced parasitics,13 allowing higher fidelity signal presentation. Here, switching residuals barely detectable noise floor. Removing optional output filter (Figure allows linear regulator contributed noise switching spikes rise, noise still below 300µVP-P.
Note noise efficiency characteristics appearing Figures were generated bench about minutes. modeling types there might want think about that. Note Figure Block Diagram. Note Appendix "Selection Criteria Linear Regulators." Note Appendix "Probing Connection Techniques Level, Wideband Signal Integrity," relevant discussion.
100µH OPTIONAL (SEE TEXT) -12V 47µF
4.7µF 3300pF SHDN DUTY SYNC
PGND LT1533 RVSL RCSL 0.01µF
1N4148
100µH
2.4k 9.6k
AN70
22nH INDUCTOR. COILCRAFT B-07T TYPICAL, TRACE INDUCTANE BEAD (SEE APPENDIX COILTRONICS CTX100-3 COILTRONICS CTX02-13665-X1 (SEE APPENDIX
Figure Negative Output Version Figure LT1533's Negative Feedback Input Requires Minimal Configuration Changes. Noise Performance Identical Positive Output Version
1N4148
47µF
AN70-9
Application Note
4.7µF 4.2k 3300pF PGND 22nH RVSL RCSL LT1533
4.7µF 1N4148
OPTIONAL SECOND SECTION (SEE TEXT)
100µH 1N4148
OUTPUT COMMON
47µF
OUTPUT OUTPUT COMMON
4N28 9.5k 240k 0.22µF
22nH INDUCTOR. COILCRAFT B-07T, TRACE INDUCTANCE BEAD. APPENDIX COILTRONICS CTX100-3 COILTRONICS CTX-02-13665-X1. APPENDIX
LT1431
+VIN 2.5V 2.5k
AN70
Figure Optoisolated Output Variant Figure Loop Closure Bypasses LT1533 Error Amplifier, Enhancing Loop Stability. Noise Performance Maintained.
4.7µF RVSL LT1533 3300pF 22nH INDUCTOR. COILCRAFT B-07T TYPICAL, TRACE, FERRITE BEAD. APPENDIX COILTRONICS CTX-02-13666-X1. APPENDIX 1N4148 PGND 100µF OUTPUT COMMON
AN70
4.7µF OPTIONAL LINEAR REGULATORS AND/OR FILTERS, -17V TYPICALLY 100µH/47µF RCSL DUTY
100µF
Figure Bipolar, Floating Output Converter. Grounding "DUTY" Biasing Puts Regulator into Duty Cycle Mode. Floating, Unregulated Outputs Proportion T1's Center Voltage. Linear Regulators Optional
AN70-10
Application Note
Figure case, spectrum analyzer measurements instrument limited. Figure shows analyzer's noise floor 500MHz sweep when monitoring unpowered Figure 26's breadboard. Figure breadboard powered, analyzer output noise limited essentially indistinguishable from unpowered case. Similarly, Figure 32's 1MHz wide "poweron" plot identical Figure 33's noise floor limited "power-off" sweep. Note that linear postregulation 40kHz fundamental components detectable. Figure circuit have linear postregulation 40kHz fundamental residue appeared Figure 15b.
500µV/DIV
10µs/DIV
AN72
Figure Removing Optional Filter Causes Linear Regulator-Contributed Noise Switching Spikes Rise. Peak-to-Peak Noise Still <300µV
10V/DIV 500mA/DIV 10V/DIV 500mA/DIV 100µV/DIV
20µs/DIV
AN70
Figure Waveforms Floating Output Converter 100mA Loading. Linear Postregulator Optional Filter Employed. Slew-Controlled Collector Voltage (Traces Current (Traces Produce Output (Trace with Under 100µV Noise
Figure HP4195A Analyzer's Noise Floor 500MHz Sweep When Connected Unpowered Figure
100µV/DIV
10µs/DIV
AN70
Figure Removing Probes Except Coaxial Output Connection Reveals Figure 27's True Noise Figure. Switching Residue Just Detectable Amplifier Noise
Figure Figure 26's "Power-On" Output Noise Undetectable Analyzer's Noise Floor Limited 500MHz Sweep
AN70-11
Application Note
Figure Linear Postregulation Eliminates 40kHz Fundamental-Related Components 1MHz Sweep
Figure Turning Circuit Power Verifies Figure 32's Plot Analyzer Noise Floor Limited. Sweep Results Identical Figure 32's "Power-On" Data
Battery-Powered Circuits basic configurations battery-powered portable apparatus. Figure similar Figure runs from 2.7VMIN (e.g., three NiCd batteries), producing output. This design induces noise-based error when powering fast 16-bit converter, something almost DC/DC converter Appendix contributes compelling testimony this somewhat boastful claim.
Figure also operates from three NiCd cells, producing output. This design achieves 100µV output noise, qualifying electronic equivalent battery. Performance Augmentation some cases desirable augment LT1533 performance characteristics. Usually, this involves additional circuitry, necessitate trading performance area gain desired benefit.
2.7V NiCd BATTERIES) 1N5817
4.7µF 3300pF SHDN DUTY SYNC
PGND LT1533 RVSL RCSL 0.01µF
5VOUT 100µH
OPTIONAL 100µH LOWEST RIPPLE
1N5817 4.99k
47µF
47µF
AN70
COILTRONICS CTX100-3 22nH TRACE INDUCTANCE, FERRITE BEAD INDUCTOR (SEE APPENDIX COILCRAFT B-07T TYPICAL CTX02-13665-X1 (SEE APPENDIX DETAILS)
Figure Circuit Delivers from Three NiCd Batteries, 100µV Wideband Output Noise. This Design Contributes Noise-Based Error When Powering 16-Bit Converter (See Appendix
AN70-12
Application Note
2.7V NiCd BATTERIES) 1N4148 9VOUT 100µH OPTIONAL 100µH LOWEST RIPPLE
4.7µF 3300pF SHDN DUTY SYNC
PGND LT1533 RVSL RCSL 0.01µF
1N4148 3.48k
47µF
47µF
21.5k
AN70
COILTRONICS CTX100-3 22nH TRACE INDUCTANCE, FERRITE BEAD INDUCTOR (SEE APPENDIX COILCRAFT B-07T TYPICAL CTX02-13665-X1 (SEE APPENDIX DETAILS)
Figure Electronic Equivalent Battery Operates from Three NiCd Cells. Output Noise Below 100µV
Quiescent Current Regulator LT1533 quiescent current about 6mA. Figure 36's circuit reduces this figure 100µA running on-off control loop around device. control loop replaces normal error amplifier, achieving regulation switching shutdown accordance with loop demands. Comparator compares scaled version output with internal reference biases regulators shutdown pin. Loop hysteresis obtained utilizing phase shift (e.g., time delay) output components. normal continuously closed loop this phase shift must minimized compensated. this case promotes desired hysteretic control characteristic. Local positive feedback ensures clean transitions. Figure shows loop work. When circuit output drops below regulation point, C1's output (Trace goes high. This enables regulator responds with burst drive (Trace transformer. output restored goes until next cycle. During C1's time regulator shut down, resulting extremely quiescent current noted. loop's on-off control characteristic causes frequency output noise related tank ring. Trace shows 600µV peaks, although wideband components observable.
High Voltage Input Regulator LT1533's process limits collector breakdown 30V. complicating factor that transformer swings supply. Thus, represents maximum allowable input supply. Many applications require higher voltage inputs Figure uses cascoded14 output stage achieve such high voltage capability. This 24V-to-50V converter reminiscent previous circuits, except that appear. These devices, interposed between transformer, constitute cascoded high voltage stage. They provide voltage gain while isolating from their large collector voltage savings. Normally, high voltage cascodes designed simply supply voltage isolation. Cascoding LT1533 presents special considerations because transformers instantaneous voltage current information must accurately transmitted, albeit lower amplitude, LT1533. this done, regulator's slew control loops will
Note term "cascode," derived from "cascade cathode," applied configuration that places active devices series. benefit higher breakdown voltage, decreased input capacitance, bandwidth improvement, etc. Cascoding been employed amps, power supplies, oscilloscopes other areas obtain performance enhancement. origin term clouded author will mail magnum champagne first reader correctly identifying original author publication.
AN70-13
Application Note
2.7V Ni-Cd BATTERIES)
4.7µF 3300pF SHDN DUTY SYNC
PGND LT1533 RVSL RCSL 0.01µF
1N4148
100µH
1N4148
OPTIONAL 100µH TEXT 47µF
47µF
LTC1440 21.5k
10pF 1.18V INTERNAL LTC1440
COILTRONICS CTX100-3 22nH TRACE INDUCTANCE, FERRITE BEAD INDUCTOR (SEE APPENDIX COILCRAFT B-07T TYPICAL CTX02-13665-X1 (SEE APPENDIX DETAILS)
Figure Hysteretic "Burst ModeTM" Loop Lowers Quiescent Current 100µA While Maintaining Output Noise
5V/DIV
10V/DIV
500µV/DIV (10kHz HIGH PASS) 1ms/DIV
AN70
Figure Operating Waveforms Quiescent Current Converter. Comparator Output (Trace Restores Output Voltage Turning LT1533 (Trace Output Noise Shows Ringing (Trace Although High Frequency Content Negligible
Burst Mode trademark Linear Technology Corporation.
AN70-14
2.32k 150k
AN70
Application Note
function, causing dramatic output noise increase. compensated resistor dividers associated with Q1-Q2 base collector biasing serve this purpose. associated components provide stable termination dividers. Figure shows waveforms Q1's operation identical, although opposing phase). Trace Q1's emitter, Trace base Trace collector. T1's ring-off obscures fact that waveform fidelity maintained through cascode, although inspection reveals this case. Additional testimony given circuit output noise (Trace which measures about 100µV peak. 24V-to-5V Noise Regulator Figure extends Figure 38's cascoding technique step-down design.15 Inputs from converted 5V/2A capacity output. protect regulators from high input voltages. cascode must accommodate 100V transformer swings. this instance MOSFETs (Q1-Q2) utilized, although divider technique necessarily retained. gate damper networks prevent transformer swings coupled gatechannel capacitance from corrupting cascode's waveform transfer fidelity. Figure shows that resultant cascode response faithful, even with 100V swings. Trace Q1's source, with Traces gate drain, respectively. Under these conditions, output, noise inside 400µV peak. Note that protect regulator from excessive input voltages. 10W, Noise Regulator Figure boosts regulator's output capability over does this with simple emitter followers (Q1Q2). Theoretically, followers preserve T1's voltage current waveform information, permitting LT1533's slew control circuitry function. practice, transistors must relatively beta types. collector current their beta sources 150mA Q1-Q2 base paths, adequate proper slew loop operation.16 follower loss limits efficiency about 68%. Higher input voltages minimize follower-induced loss, permitting efficiencies range. Figure shows noise performance. Ripple measures (Trace using single section, with high frequency content just discernible. Adding optional second section drops ripple below 100µV (Trace high frequency content seen (note vertical scale factor change) inside 180µV. 7500V Isolated Noise Supply final form performance augmentation extremely high voltage isolation. This often required situations where circuitry must withstand high common mode voltage effects. Figure similar Figure 25's isolated supply, except that 7500V (peak) breakdown capability. Transformer optoisolator changes permit this. remaining operating performance characteristics identical Figure
Note This circuit developed from design Jeff Witt Linear Technology Corporation. Note Operating slew loops from follower base current suggested Dobkin Linear Technology Corporation.
AN70-15
Application Note
(20V 30V)
0.003µF 3.3k MUR-110
1N4148 1N752 5.6V SHDN DUTY SYNC
PGND LT1533 RVSL RCSL 0.01µF 2.49k MUR-110 3.3k
100µH
OPTIONAL 100µH LOWEST RIPPLE
47µF
47µF
47µF
3300pF
97.6k 0.003µF
AN70
COILTRONICS CTX100-3 22nH TRACE INDUCTANCE, FERRITE BEAD INDUCTOR (SEE APPENDIX COILCRAFT B-07T TYPICAL ZETEX ZTX-853 2N2222A CTX-02-13665-X1 (SEE APPENDIX DETAILS)
Figure Output Noise Regulator. Cascoded Bipolar Transistors Accommodate Transformer Swings, Permitting (20VIN 30VIN) Powered Operation
5V/DIV 5V/DIV
20V/DIV
100µV/DIV 1µs/DIV
AN70
Figure Cascode Transmits Instantaneous Voltage Slew Information, Permitting LT1533 Maintain Noise Output. Trace Emitter, Trace Base Trace Collector. Transformer Ring-Off Obscures Cascode Action, Study Reveals Faithful Transmission. Output (Trace 100µV Noise
AN70-16
Application Note
24VIN (20V 50V)
10µF
MBRS140 MPSA42 2N2222 0.002µF 0.002µF MBRS140 100µH 5VOUT
OPTIONAL 100µH TEXT
220µF
100µF
4.7µF 1500pF SYNC DUTY SHDN RCSL RVSL LT1533 PGND 0.01µF 7.5k 2.49k
AN70
COILTRONICS CTX100-3 22nH TRACE INDUCTANCE, FERRITE BEAD INDUCTOR (SEE APPENDIX COILCRAFT B-07T TYPICAL MTD6N15 COILTRONICS VP4-0860
Figure Noise 24V-(20VIN 50VIN)-to-5V Converter. Cascoded MOSFETs Withstand 100V Transformer Swings, Permitting LT1533 Control 5V/2A Output
20V/DIV
5V/DIV COUPLED) 100V/DIV
10µs/DIV
AN70
Figure MOSFET-Based Cascode Permits Regulator Control 100V Transformer Swings While Maintaining Noise Output. Trace Q1's Source, Trace Q1's Gate Trace Drain. Waveform Fidelity Through Cascode Permits Proper Slew Control Operation
AN70-17
Application Note
1N4148
1N5817 0.003µF 300µH 33µH
0.05
4.7µF 1500pF SHDN DUTY SYNC
PGND LT1533 RVSL RCSL 0.01µF 2.49k 21.5k 1N4148
4.7µF 0.05
OPTIONAL LOWEST RIPPLE
100µF 1N5817
100µF
AN70
COILTRONICS CTX300-4 22nH TRACE INDUCTANCE, FERRITE BEAD INDUCTOR (SEE APPENDIX COILCRAFT B-07T TYPICAL COILTRONICS CTX33-4 MOTOROLA D45C1 COILTRONICS CTX-02-13949-X1 FERRONICS FERRITE BEAD 21-110J
Figure Noise 5V-to-12V Converter. Q1-Q2 Provide Output Capacity While Preserving LT1533's Voltage/Current Slew Control. Efficiency 68%. Higher Input Voltages Minimize Follower Loss, Boosting Efficiency Above
5mV/DIV
100µV/DIV
2µs/DIV
AN70
Figure Waveforms Figure Output. Trace Shows Fundamental Ripple with Higher Frequency Residue Just Discernible. Optional Section Produces Trace 180µVP-P Wideband Noise Performance
AN70-18
Application Note
4.2k 3300pF PGND 22nH 4.7µF RVSL RCSL LT1533
4.7µF 1N4148
OPTIONAL SECOND SECTION (SEE TEXT)
100µH 1N4148
OUTPUT COMMON
47µF
OUTPUT OUTPUT COMMON
MOC-8112 9.5k 0.22µF
22nH INDUCTOR. COILCRAFT B-07T, TRACE INDUCTANCE BEAD. APPENDIX COILTRONICS CTX100-3 COILTRONICS CTX-02-13950. APPENDIX
LT1431
+VIN 2.5V 2.5k
AN70
Figure 7500V Isolation Version Figure Transformer Optoisolator Changed Achieve Isolation Noise Immunity. Circuit Operation Before
Note: This Application Note derived from manuscript originally prepared publication magazine.
AN70-19
Application Note
REFERENCES Shakespeare, William, "Much About Nothing," 319, 1598-1600. Williams, Jim, "Design DC/DC Converters Catch Noise Source," Electronic Design, October 1981, page 229. Williams, Jim, "Conversion Techniques Adapt Voltages Your Needs," EDN, November 1982, page 155. Tektronix, Inc. "Type Operating Service Manual," Circuit, 1954. Tektronix, Inc. "Type Operating Service Manual," Circuit, 1967. Tektronix, Inc. "7904 Oscilloscope Operating Service Manual," Converter-Rectifiers, 1972. Hewlett-Packard "1725A Oscilloscope Operating Service Manual," High Voltage Power Supply, 1980. Arthur, Ken, "Power Supply Circuits," High Voltage Power Supplies, Tektronix Concept Series, 1967. Williams, Huffman, Brian, "Some Thoughts DC/DC Converters," Noise ±15V Converter Ultralow Noise ±15V Converter, pages Linear Technology Corporation Application Note 1988. Williams, Huffman, Brian, "Precise Converter Designs Enhance System Performance," EDN, October 1988, pages 185. Tektronix, Inc. "Type 1A7A Differential Amplifier Instruction Manual," Check Overall Noise Level Tangentially, pages 5-36 5-37, 1968. Williams, Jim, "High Speed Amplifier Techniques," Linear Technology Corporation Application Note 1991. Witt, Jeff, "The LT1533 Heralds Class Noise Switching Regulators," Linear Technology, Vol. VII, August 1997, Linear Technology Corporation. Morrison, Ralph, "Noise Other Interfering Signals," John Wiley Sons, 1992. Morrison, Ralph, "Grounding Shielding Techniques Instrumentation," Wiley-Interscience, 1986. Sheehan, Dan, "Determine Noise DC/DC Converters," Electronic Design, September 1973. Hewlett-Packard "HP-11941A Close Field Probe Operation Note," 1987. Terrien, Mark, "The HP-11940A Close Field Probe: Characteristics Application Troubleshooting," Microwave Symposium, available from Hewlett-Packard Pressman, A.I., "Switching Linear Power Supply, Power Converter Design," Hayden Book Co., Hasbrouck Heights, Jersey, 1977. Chryssis, "High Frequency Switching Power Supplies, Theory Design," McGraw Hill, York, 1984.
AN70-20
Application Note
APPENDIX HISTORY NOISE DC/DC CONVERSION batteries noise power sources? 60Hz power line derived linear regulators have output noise? with most innocent questions, thoughtful answers provide surprising insights. These sources have output noise because they have harmonic energy content. 60Hz fundamental driven supply produces some harmonic activity, power becomes very small well inside 1kHz. battery even better. These conclusions direction towards designing noise DC/DC converters. goal noise, reduction harmonic energy, particular, wideband harmonics. This simple guideline central LT1533 operation, although refinements necessary generally applicable History notion minimizing harmonics DC/DC conversion output noise new. Oscilloscopes have used this technique generate high voltage accelerating potentials without degrading instrument operation.1 Designing 10,000V output DC/DC converter that does disrupt 500MHz, high sensitivity vertical amplifier challenging. Figure shows DC/DC converter from Tektronix oscilloscope. Q1430, configured modified Hartley power oscillator, drives T1430. T1430's output multiplied diode-capacitor tripler, producing 12,000V. Feedback Q1414 summed against derived reference, closing regulation loop around power oscillator. sine wave transformer drive (see waveforms figure) harmonic content, resulting desired conducted radiated noise. This approach very efficient-Q1430 operates linear region-but power loss acceptable 125W instrument. Tektronix 7000 series oscilloscopes used resonant, offline converter power entire instrument. before, high voltage generated separately (see Footnote Figure partial schematic Tektronix 7904 power converter, shows series resonant network, L1237C1237 Q1234-Q1241 drive path. This results sine wave drive output transformer T1310, despite Q1234Q1241's rectagular waveshape. Feedback (not shown) closes loop around this stage, stabilizing operating point. resonant, sine wave transformer drive provides desired noise characteristics with good efficiency. less specific example appears Application Note Figure partial schematic Application Note 29's Figure shows sine wave oscillator based) driving power amplifier Q6). output transformer, provides voltage boosted secondary drive linear regulators (not shown). This brute force approach provides converter with extraordinarily noise, complex inefficient. operating their linear regions, dissipate considerable power, efficiency 30%. Figure A4's approach, also from AN29's Figure achieves better efficiency. partial schematic shows source followers driven from 100-0.003µF edge slow-down networks. This slows down transistor's transitions, resulting harmonic reduction noise. Unfortunately, drive scheme complex somewhat inflexible, requiring bootstrapped voltages fully switch transistors off. Additionally, transformer change would require drive rework maintain efficiency noise characteristics. Finally, dynamic voltage current control transistors passively determined very well controlled. LT1533 uses closed-loop control2 around output stages tightly control voltage current slewing. This allows variety circuits magnetics easily accommodated, resulting true general purpose solution. Text Figure associated discussion provide more LT1533 operating details.
Note Ancillary benefits include eliminating complex expensive high voltage winding main power transformer, avoidance long, high voltage wire runs space weight savings. Note Patent pending.
AN70-21
AN70-22
Copyright 1967 Tektronix, Inc. rights reserved. Reproduced permission
Application Note
Figure Tektronix Circuit Uses Sine Wave Drive Noise DC/DC Conversion. Efficiency Poor, Because Q1430 Remains Linear Region
Application Note
Copyright 1972 Tektronix, Inc. rights reserved. Reproduced permission
Figure Tektronix 7904 Main Inverter Obtains Noise Converting Q1234-Q1241 Rectangular Drive Sine Wave L1237-C1237 Resonating Network. Output Transformer Produces Noise Power with Good Efficiency. Approach Application Specific Inflexible
AN70-23
0.22µF 0.22µF 200k 1N4001 0.22µF LT1006
LT1013 THERMALLY MATED 3.1k LT1009-2.5
OSCILLATOR STABILITY LOOP
Figure Sine Wave-Based DC/DC Converter Appeared Application Note Output Noise Low, Circuit Complex Inefficient
AN70-24
5VIN 4.5V 5.5V
22µF 100µH POWER
0.01µF
16kHz OSCILLATOR
MJE2955 330µF 0.01µF 22µF
1N4001
Application Note
47µF
LT1006
2N2905 MJE3055 0.1µF
2N2219
0.1µF
OUTPUT RECTIFIER/FILTER LINEAR REGULATION
(SELECTED VALUE)
LT1013 0.33µF
CONTROL LOOP
AN70 FA03
Application Note
1N5817
100µF 74C74 74C14 0.001µF NONOVERLAP GENERATOR 74C02 POINT BOOST OUTPUT 17VDC
10µF LT1054 BOOST 1N5817 1N5817
47µF
0.001µF
1N5817 TURBO BOOST
5VIN 4.5V 5.5V
74C02 74C14 74C14 15kHz, NONOVERLAP DRIVERS EDGE SHAPING ±15V COMMON GROUND 150k 0.003µF 0.001µF 74C14 74C14 150k
OUTPUT RECTIFIER/FILTER LINEAR REGULATION
0.003µF 1N5817
OUTPUT
LEVEL SHIFTS
4VDC
Figure Application Note Circuit Slopes Edge Drive Noise Better Efficiency. Gate Drive Circuitry Complex Poorly Controlled, Making Circuit Inflexible
MTP3055E 2N3906 2N3904 PULSE ENGINEERING, INC. #PE-61592 FERRITE BEAD, FERRONICS #21-110J
22µF
AN70
AN70-25
Application Note
APPENDIX SPECIFYING MEASURING SOMETHING CALLED NOISE Undesired output components switching regulators commonly referred "noise." rapid, switched mode power delivery that permits high efficiency conversion also creates wideband harmonic energy. This undesirable energy appears radiated conducted components, "noise." Actually switching regulator output "noise" isn't really noise all, coherent, high frequency residue directly related regulator's switching. Unfortunately, almost universal practice refer these parasitics "noise," this publication maintains this common, albeit inaccurate, terminology.1 Measuring Noise There almost uncountable number ways specify noise switching regulator's output. common industrial practice specify peak-to-peak noise 20MHz bandpass.2 Realistically, electronic systems readily upset spectral energy beyond 20MHz, this specification restriction benefits one.3 Considering this, seems appropriate specify peak-to-peak noise verified 100MHz bandwidth. Reliable level measurements this bandpass require careful instrumentation choice connection practices. study begins selecting test instrumentation verifying bandwidth noise. This necessitates arrangement shown Figure Figure diagrams signal flow. pulse generator supplies subnanosecond rise time step attenuator, which produces <1mV version step. amplifier takes 40dB gain 100) oscilloscope displays result. "front-to-back" cascaded bandwidth this system should about 100MHz (trise 3.5ns) Figure reveals this Figure B3's trace shows 3.5ns rise time about 100µV noise. noise limited amplifier's noise floor.4 Figure B4's presentation text Figure output noise shows barely visible switching artifacts vertical graticule lines 100MHz bandpass. Fundamental ripple seen more clearly, although similarly noise floor dominated. Restricting measurement bandwidth 10MHz (Figure reduces noise floor amplitude, although switching noise ripple amplitudes preserved. This indicates that there signal power beyond 10MHz. Further measurements bandwidth successively reduced determine highest frequency content present. importance measurement bandwidth further illustrated Figures Figure measures commercially available DC/DC converter 1MHz bandpass. unit appears meet claimed 5mVP-P noise specification. Figure bandwidth increased 10MHz. Spike amplitude enlarges 6mVP-P, about outside specification limit. Figure B8's 50MHz viewpoint brings unpleasant surprise. Spikes measure 30mVP-P -six times specified limit!5
Note Less genteelly, can't beat 'em, join 'em." Note DC/DC converter manufacturer specifies noise 20MHz bandwidth. This beyond deviousness unworthy comment. Note Except, course, eager purveyors power sources specify them this manner. Note Observed peak-to-peak noise somewhat affected oscilloscope's "intensity" setting. Reference describes method normalizing measurement. Note Caveat Emptor.
AN70-26
Application Note
AN70-27
Figure 100MHz Bandwidth Verification Test Setup. Note Coaxial Connections Wideband Signal Integrity
Application Note
PULSE GENERATOR HP-215A RISE TIME 350MHz AMPLIFIER X40dB HP-461A 150MHZ 2.4ns)
ATTENUATOR HP-355D 1000MHZ
OSCILLOSCOPE TEKTRONIX 454A 150MHZ 2.4ns)
<1mV RISETIME (350MHZ)
CASCADED BANDWIDTH 100MHz 3.5ns RISE TIME)
AN70
Figure Subnanosecond Pulse Generator Wideband Attenuator Provide Fast Step Verify Test Setup Bandwidth
100µV/DIV
100µV/DIV
2ns/DIV
AN70B03
10µs/DIV
AN70
Figure Oscilloscope Display Verifies Test Setup's 100MHz (3.5ns Rise Time) Bandwidth. Baseline Noise Derives from Amplifier's Input Noise Floor
Figure Text Figure Output Switching Noise Just Discernible 100MHz Bandpass
100µV/DIV 10mV/DIV
10µs/DIV
AN70
Figure 10MHz Band Limited Version Preceding Photo. Switching Noise Information Preserved, Indicating Adequate Bandwidth
50µs/DIV
AN70
Figure Commercially Available Switching Regulator's Output Noise 1MHz Bandpass. Unit Appears Meet 5mVP-P Noise Specification
AN70-28
Application Note
10mV/DIV
20mV/DIV
50µs/DIV
AN70
50µs/DIV
AN70
Figure Figure A6's Regulator Noise 10MHz Bandpass. 6mVP-P Noise Exceeds Regulator's Claimed Specification
Figure Wideband Observation Figure Shows 30mVP-P Noise Times Regulator's Specification!
Frequency Noise frequency noise rarely concern, because almost never affects system operation. Text Figure frequency noise shown Figure possible reduce frequency noise rolling control loop bandwidth (e.g., 0.68µF feedback capacitor across value 2000pF text Figure Figure shows about five times improvement when this done, even with greater measurement bandwidth. possible disadvantage loss loop bandwidth slower transient response. Preamplifier Oscilloscope Selection level measurements described require some form preamplification oscilloscope. Current generation oscilloscopes rarely have greater than 2mV/DIV sensitivity, although older instruments offer more capability. Figure lists representative preamplifiers oscilloscope plug-ins suitable noise measurement. These
units feature wideband, noise performance. particularly significant that majority these instruments longer produced. This keeping with current instrumentation trends, which emphasize digital signal acquisition opposed analog measurement capability. monitoring oscilloscope should have adequate bandwidth exceptional trace clarity. latter regard high quality analog oscilloscopes unmatched. exceptionally small spot size these instruments well-suited level noise measurement.6 digitizing uncertainties raster scan limitations DSOs impose display resolution penalties. Many displays will even register small levels switching-based noise.
Note work have found Tektronix types 454, 454A, excellent choices. Their pristine trace presentation ideal discerning small signals interest against noise floor limited background.
500µV/DIV
50µV/DIV
10ms/DIV
AN70
10ms/DIV
AN70
Figure 3kHz Noise Using Standard Frequency Compensation. Almost Noise Power Below 1kHz
Figure B10. Feedback Lead Network Decreases Frequency Noise, Even Measurement Bandwidth Expands 100kHz
AN70-29
Application Note
INSTRUMENT TYPE Amplifier MANUFACTURER Hewlett-Packard MODEL MAXIMUM NUMBER BANDWIDTH SENSITIVITY/GAIN 461A 7A13 11A33 P6046 1855 1A7/1A7A 7A22 5A22 ADA-400A 1822 SR-560 150MHz 50MHz 100MHz 150MHz 100MHz 100MHz 1MHz 1MHz 1MHz 1MHz 10MHz 1MHz Gain 1mV/DIV 1mV/DIV 1mV/DIV 1mV/DIV Gain 10µV/DIV 10µV/DIV 10µV/DIV 10µV/DIV Gain 1000 Gain 50000 AVAILABILITY Secondary Market Secondary Market Secondary Market Secondary Market Secondary Market Secondary Market Secondary Market Secondary Market COMMENTS Input, Stand-Alone Requires Series Mainframe Requires 7000 Series Mainframe Requires 11000 Series Mainframe Stand-Alone Requires Series Mainframe, Settable Bandstops Requires 7000 Series Mainframe, Settable Bandstops Requires 5000 Series Mainframe, Settable Bandstops
Differential Amplifier Tektronix Differential Amplifier Tektronix Differential Amplifier Tektronix Differential Amplifier Tektronix Differential Amplifier Preamble Differential Amplifier Tektronix Differential Amplifier Tektronix Differential Amplifier Tektronix Differential Amplifier Tektronix Differential Amplifier Preamble Differential Amplifier Stanford Research Systems
Current Production Stand-Alone, Settable Bandstops
Current Production Stand-Alone with Optional Power Supply, Settable Bandstops Current Production Stand-Alone, Settable Bandstops Current Production Stand-Alone, Settable Bandstops, Battery Line Operation
Figure B11. Some Applicable High Sensitivity, Noise Amplifiers. Trade-Offs Include Bandwidth, Sensitivity Availability
APPENDIX PROBING CONNECTION TECHNIQUES LEVEL, WIDEBAND SIGNAL INTEGRITY most carefully prepared breadboard cannot fulfill mission signal connections introduce distortion. Connections circuit crucial accurate information extraction. level, wideband measurements demand care routing signals test instrumentation. Ground Loops Figure shows effects ground loop between pieces line-powered test equipment. Small current flow between test equipment's nominally grounded chassis creates 60Hz modulation measured circuit output.
This problem avoided grounding line powered test equipment same outlet strip otherwise ensuring that chassis same ground potential. Similarly, test arrangement that permits circuit current flow chassis interconnects must avoided.
Pickup Figure also shows 60Hz modulation noise measurement. this case, 4-inch voltmeter probe feedback input culprit. Minimize number test connections circuit keep leads short.
100µV/DIV
500µV/DIV
2ms/DIV
AN70
5ms/DIV
AN70
Figure Ground Loop Between Pieces Test Equipment Induces 60Hz Display Modulation
Figure 60Hz Pickup Excessive Probe Length Feedback Node
AN70-30
Application Note
AN70-31
Figure Poor Probing Technique. Trigger Probe Ground Lead Cause Ground Loop-Induced Artifacts Appear Display
Application Note
Poor Probing Technique Figure C3's photograph shows short ground strap affixed scope probe. probe connects point which provides trigger signal oscilloscope. Circuit output noise monitored oscilloscope coaxial cable shown photo. Figure shows results. ground loop board between probe ground strap ground referred cable shield causes apparent excessive ripple display. Minimize number test connections circuit avoid ground loops. strap eliminated, replaced grounding attachment. Figure shows better results over preceding case, although signal corruption still evident. Maintain coaxial connections noise signal monitoring path. Proper Coaxial Connection Path Figure coaxial cable transmits noise signal amplifier-oscilloscope combination. theory, this affords highest integrity cable signal transmission. Figure C10's trace shows this true. former examples aberrations excessive noise have disappeared. switching residuals faintly outlined amplifier noise floor. Maintain coaxial connections noise signal monitoring path. Direct Connection Path
100µV/DIV
5µs/DIV
AN70
Figure Apparent Excessive Ripple Results from Figure C3's Probe Misuse. Ground Loop Board Introduces Serious Measurement Error
good verify there cable-based errors eliminate cable. Figure C11's approach eliminates cable between breadboard, amplifier oscilloscope. Figure C12's presentation indistinguishable from Figure C10, indicating cable-introduced infidelity. When results seem optimal, design experiment test them. When results seem poor, design experiment test them. When results expected, design experiment test them. When results unexpected, design experiment test them. Test Lead Connections theory, attaching voltmeter lead regulator's output should introduce noise. Figure C13's increased noise reading contradicts theory. regulator's output impedance, albeit low, zero, especially frequency scales noise injected test lead works against finite output impedance, producing 200µV noise indicated figure. voltmeter lead must connected output during testing, should done through 10k-10µF filter. Such network eliminates Figure C13's problem while introducing minimal error monitoring DVM. Minimize number test lead connections circuit while checking noise. Prevent test leads from injecting into test circuit.
Violating Coaxial Signal Transmission-Felony Case Figure coaxial cable used transmit circuit output noise amplifier-oscilloscope been replaced with probe. short ground strap employed probe's return. error inducing trigger channel probe previous case been eliminated; 'scope triggered noninvasive, isolated probe.1 Figure shows excessive display noise breakup coaxial signal environment. probe's ground strap violates coaxial transmission signal corrupted Maintain coaxial connections noise signal monitoring path. Violating Coaxial Signal Transmission- Misdemeanor Case Figure C7's probe connection also violates coaxial signal flow, less offensive extent. probe's ground
Note discussed. Read
AN70-32
Application Note
Figure Floating Trigger Probe Eliminates Ground Loop, Output Probe Ground Lead (Photo Upper Right) Violates Coaxial Signal Transmission
500µV/DIV
5µs/DIV
AN70
Figure Signal Corruption Figure C5's Noncoaxial Probe Connection
AN70-33
Application Note
Figure Probe with Grounding Attachment Approximates Coaxial Connection
100µV/DIV
5µs/DIV
AN70
Figure Probe with Grounding Attachment Improves Results. Some Corruption Still Evident
AN70-34
Application Note
Figure Coaxial Connection Theoretically Affords Highest Fidelity Signal Transmission
100µV/DIV
5µs/DIV
AN70
Figure C10. Life Agrees with Theory. Coaxial Signal Transmission Maintains Signal Integrity. Switching Residuals Faintly Outlined Amplifier Noise
AN70-35
Application Note
Figure C11. Direct Connection Equipment Eliminates Possible Cable-Termination Parasitics, Providing Best Possible Signal Transmission
100µV/DIV
5µs/DIV
AN70
Figure C12. Direct Connection Equipment Provides Identical Results Cable-Termination Approach. Cable Termination Therefore Acceptable
AN70-36
Application Note
200µV/DIV
5µs/DIV
AN70
Figure C13. Voltmeter Lead Attached Regulator Output Introduces Pickup, Multiplying Apparent Noise Floor
Isolated Trigger Probe text associated with Figure somewhat cryptically alluded "isolated trigger probe." Figure reveals this simply choke terminated against ringing. choke picks residual radiated field, generating isolated trigger signal. This arrangement furnishes 'scope trigger signal with essentially measurement corruption. probe's physical form appears Figure C15. good results termination should adjusted minimum ringing while preserving highest possible amplitude output. Light compensatory damping produces Figure C16's output, which will cause poor 'scope triggering. Proper adjustment results more favorable output (Figure C17), characterized minimal ringing welldefined edges. Trigger Probe Amplifier field around switching magnetics small adequate reliably trigger some oscilloscopes. such cases, Figure C18's trigger probe amplifier useful. uses adaptive triggering scheme compensate variations probe output amplitude. stable trigger output maintained over 50:1 probe output range. operating gain 100, provides wideband gain. output this stage biases 2-way peak detector through Q4). maximum peak stored Q2's emitter capacitor, while minimum excursion retained Q4's emitter capacitor. value midpoint A1's
output signal appears junction 500pF capacitor units. This point always sits midway between signal's excursions, regardless absolute amplitude. This signal-adaptive voltage buffered trigger voltage LT1116's positive input. LT1116's negative input biased directly from A1's output. LT1116's output, circuit's trigger output, unaffected >50:1 signal amplitude variations. X100 analog output available Figure shows circuit's digital output (Trace responding amplified probe signal (Trace Figure typical noise testing setup. includes breadboard, trigger probe, amplifier, oscilloscope coaxial components.
PROBE SHIELDED CABLE TERMINATION OUTPUT
CONNECTION TERMINATION J.W. MILLER #100267
DAMPING ADJUST 4700pF
AN70 FC14
Figure C14. Simple Trigger Probe Eliminates Board Level Ground Loops. Termination Components Damp L1's Ringing Response
AN70-37
AN70-38
Figure C15. Trigger Probe Termination Box. Clip Lead Facilitates Mounting Probe, Electrically Neutral
Application Note
Application Note
10mV/DIV
10mV/DIV
10µs/DIV
AN70
10µs/DIV
AN70
Figure C16. Misadjusted Termination Causes Inadequate Damping. Unstable Oscilloscope Triggering Result
Figure C17. Properly Adjusted Termination Minimizes Ringing with Small Amplitude Penalty
ANALOG OUTPUT 'SCOPE TRIGGER INPUT
0.005µF 0.005µF 500pF
LT1227
10µF
0.1µF
100µF
0.1µF
0.1µF
CA3096 ARRAY: SUBSTRATE (PIN GROUND 1N4148 TRIGGER PROBE TERMINATION (SEE FIGURE DETAILS)
Figure C18. Trigger Probe Amplifier Analog Digital Outputs. Adaptive Threshold Maintains Digital Output over 50:1 Probe Signal Variations
1V/DIV COUPLED
5V/DIV
10µs/DIV (UNCALIB)
AN70
Figure C19. Trigger Probe Amplifier Analog (Trace Digital (Trace Outputs
LT1006
DIGITAL TRIGGER 'SCOPE
AN70
LT1116
AN70-39
AN70-40
Figure C20. Typical Noise Test Setup Includes Trigger Probe, Amplifier, Oscilloscope Coaxial Components
Application Note
Application Note
APPENDIX BREADBOARDING LAYOUT CONSIDERATIONS LT1533-based circuit's harmonic content allows their noise performance less layout sensitive than other switching regulators. However, some degree prudence order. things, cavalierness direct route disappointment. Obtaining absolute lowest noise figure requires care, performance below 500µV readily achieved. general, lowest noise obtained preventing mixing ground currents return path. Indiscriminate disposition ground currents into ground plane will cause such mixing, raising observed output noise. LT1533's restricted edge rates mitigate against corrupted ground path-induced problems, best noise performance occurs "single-point" ground scheme. Single-point return schemes impractical production boards. such cases, provide lowest possible impedance path power entry point from inductor associated with LT1533's power ground pin. (Pin 16). Locate output component ground returns close circuit load point possible. Minimize return current mixing between input output sections restricting such mixing smallest possible common conductive area. Breadboard Figure shows text Figure breadboard. keeping with breadboard's purpose, constructed fast easy modify. Single-point returns arrive separately from output area (right side photo) LT1533 (center left photo). ground plane carries current. dummy load resistors terminated plane, returned transformer's center tap. center plane separately tied into ground system power input common jack. ±15V Breadboard Text Figure 24's breadboard appears Figure Layout considerations similar Figure although design's floating output mandates changes. output load (photo's right, above connector) returns directly transformer secondary, which floats from input (and plane) ground potential. main ground plane tied input common power entry port (left banana jack). floating output potentials referred separate, smaller planed area (photo lower right) which tied transformer secondary center tap. Demonstration Board Figure enticingly portrays LT1533 demonstration board. board's practical layout readily adaptable production versions. This board useful observing LT1533 performance example practical layout. Noise performance similar text's breadboards.
AN70-41
AN70-42
Figure Text Figure 5V-to-12V Converter Breadboard. Construction Easy Change, Facilitating Experiments. Note Single-Point Ground Returns. Ground Plane Carries Current, Tied Input Common Board Entry Point (Middle Banana Jack)
Application Note
Application Note
Figure Text Figure 26's Breadboard with Linear Output Regulators. Construction Encourages Changes Measurement. Layout Similar Figure Although Floating Output Necessitates Changes. Separate Planed Area (Photo Center Right) Maintains Impedance Between Output-Related Returns
AN70-43
AN70-44
Figure Very Civilized LT1533 Demonstration Board Comely Splendor
Application Note
Application Note
APPENDIX SELECTION CRITERIA LINEAR REGULATORS Some applications, particularly floating output circuits, require linear postregulators. Selection criteria include regulator output accuracy, dropout, ripple rejection line regulation. Often, short-circuit protection needed because drive circuit output impedance current limiting prevent destructive overload. such cases, relatively poor output load regulation accuracy acceptable, simple Zener diode-emitter followerbased regulation suffice. LM78L/79L type devices offer output accuracy improved line regulation, although dropout about 2V-significantly higher than simple Zener-emitter follower regulator. Ripple rejection LM78L/79L types degrades they approach dropout, which desirable operating region best efficiency. High performance regulators such LT1575 (negative) LT1521 (positive) offer dropout voltages below 0.5V, tight line regulation, accuracy fully specified ripple rejection close dropout. usually desirable operate close dropout maintain good overall efficiency. Because this, regulator ripple rejection should tested this intended operating region. Additionally, cost, size performance trade-offs various filter components regulators should evaluated determine best solution particular application. Testing Ripple Rejection Ripple rejection tested with Figure E1's arrangement. generator should operate over frequency range interest capable supplying required output drive. practice, generator supply regulator input operating voltage expected LT1533 switching frequency. Comparison different regulators filter components under varying operating conditions easily carried out.
HIGH VOLTAGE, FLOATING OUTPUT SINE WAVE GENERATOR HP200A TYPICAL 100µH
LINEAR REGULATOR UNDER TEST 20µF
AMPLIFIER
OUTPUT 'SCOPE
0VRMS 20VRMS OUTPUT CAPABILITY MUR110 COILTRONICS CTX100-4 20µF
PREAMBLE 1822 TYPICAL LOAD DESIRED)
AN70
Figure Ripple Rejection Test Setup Linear Regulators. Combinations Regulators Evaluated
AN70-45
AN70-46
Figure Ripple Rejection Test Setup Includes Sine Wave Generator, Breadboard, Amplifier Oscilloscope
Application Note
Application Note
APPENDIX MAGNETICS CONSIDERATIONS Transformers LT1533's symmetrical "push-pull" drive makes transformer behavior quite predictable. such, transformers usually specified indicating operating frequency, power desired input/output voltages. Figure lists transformers used text circuits along with some their characteristics. These components, variations them, available from Coiltronics, telephone #561-241-7876. Inductors Inductors LT1533 circuits have special characteristics. Text Figure circuit, "forward" type converter,1 requires inductor ahead filter capacitor, although additional filtering optional. Figure 26's "50%" mode circuit output inductor requirement unless heavily loaded (see text), although sections used best possible ripple attenuation. either case, inductor characteristics particularly critical. circuits shown text Coiltronics "Octa-Pak" type toroidal core-based inductors. 22nH inductor used LT1533's power ground return (Pin mandatory. take several forms, including trace inductance, small coil wire, ferrite bead packaged inductor specified schematics. coiled wire employed, five turns sufficient. equivalent length trace gives similar results. ferrite bead (e.g., Ferronics #21-110J equivalent) with turns wire also works well. example packaged 22nH inductor Coilcraft B-07T which specified test circuits.
Note References basic forward converter theory.
NOMINAL INPUT VOLTAGE
NOMINAL OUTPUT VOLTAGE AFTER LINEAR REGULATOR ±15V ±15V
OUTPUT POWER 1.5W 3.0W 1.5W 3.0W 1.5W
COILTRONICS PART NUMBER CTX-02-13716-X1 CTX-02-13665-X1 CTX-02-13713-X1 CTX-02-13664-X1 CTX-02-13834-X3* CTX-02-13949-X1
CONNECTION DIAGRAM
PRIMARY PRIMARY
SECTION SECTION
PRIMARY PRIMARY
SECTION SECTION OUTPUT COMMON THIS POINT
TIED TOGETHER HIGH TURNS RATIO VERSION CTX-02-13716-X1. ACCOMMODATES SUPPLY VOLTAGES HIGH DROPOUT REGULATORS
AN70 FF01
Figure Transformer Types Used Text Circuits. Variations Specific Requirements Available from Coiltronics, 561-241-7876
AN70-47
Application Note
APPENDIX VOLTAGE CURRENT SLEW CONTROL? Carl Nelson LT1533 gives dramatic reduction high frequency noise controlling both voltage current slew rates switch. This technique also advantage controlling noise other switching regulator components, namely, catch diode input output capacitors. Figure G1's block diagram shows basic concepts slew control. switch driven with currents switched These currents large enough drive switch very high slew rates. Actual slew rates voltage slew current slew. During switch turn-on, collector initially high current zero. Inductor current holds switch high until switch current equals inductor current. first limiting action occurs current builds Current sensed fixed gain amplifier increasing current generates current through proportional switch current slew rate. This current compared difference amplified which shunts away excess current control switch current slew rate. When switch current exceeds inductor current, Q1's collector would normally fall speed limited only diode switch parasitic capacitance. control voltage slew, current through compared difference amplified clamp base current This stops further rise switch current forces switch voltage fall controlled rate. switch turn-off, current voltage must controlled reverse order. Switch flipped provide reverse base drive, polarity reversed. Almost immediately, switch current falls slightly below inductor current. This would normally cause switch voltage slew limited only diode switch capacitance. Here, senses voltage slew controls switch base drive limit switch rise time. Switch current remains essentially constant during voltage slew.
VOLTAGE SLEW CONTROL CIRCUIT
VSLEW PROGRAMMING CURRENT CURRENT SLEW CONTROL CIRCUIT VSLEW PROGRAMMING CURRENT SWITCH DRIVE CURRENT
SWITCH CONTROL
Figure Slew Control Conceptual Block Diagram
AN70-48
AN70
Application Note
When switch voltage reaches level where catch diode turns switch current would normally drop rapidly, creating fast field transients around switch, diode output capacitor lines. come into play here, sensing decreasing switch current controlling base drive force controlled decrease switch, diode capacitor current. Figure shows switch, diode output capacitor waveforms with controlled switch drive operation. Note that current voltage slew limiting occur simultaneously. must take over when first complete. This requires very fast control circuitry avoid crossover glitches that would create noise spikes.
VOLTAGE CURRENT
TIME
AN70
Figure Switch Voltage Current During Turn-On Turn-Off
APPENDIX HINTS LOWEST NOISE PERFORMANCE LT1533's controlled switching times allow extraordinarily noise DC/DC conversion with surprisingly little design effort. Wideband output noise well below 500µV easily achieved. most situations this level performance entirely adequate. Applications requiring lowest possible output noise will benefit from special attention several areas. Noise Tweaking slew time versus efficiency trace-off discussed text should weighted towards lowest noise extent tolerable. Typically, slew times beyond 1.3µs result "expensive" noise reduction terms lost efficiency, benefit available. issue much power expendable obtain incremental decreases output noise. Similarly, layout techniques discussed Appendix should reviewed. Rigid adherence these guidelines will result correspondingly lower noise performance. text's breadboards were originally constructed provide lowest possible noise levels, then systematically degraded test layout sensitivity. This approach allows experimentation determine best layout without expending fanatical attention details that provide essentially benefit. slow edge times greatly minimize radiated EMI, experimentation with component's physical orientation sometimes improve things. Look components (yes, literally!) imagine just what their residual radiated field impinges particular, optional output inductor pick field radiated other magnetics, resulting increased output noise. Appropriate physical layout will eliminate this effect, experimentation useful. probe described Appendix useful tool this pursuit highly recommended. Appendix contributes hints magnetics-based noise similarly recommended. Capacitors filter capacitors used should have parasitic impedance. Sanyo OS-CON types excellent this regard contributed performance levels quoted text. Tantalum types nearly good. input supply bypass capacitor, which should located directly transformer center tap, needs similarly good characteristics. Aluminum electrolytics suitable service LT1533 circuits. Damper Network Some circuits benefit from small (e.g., 3301000pF) damper network across transformer secondary absolutely lowest noise needed. Extremely small (20µV 30µV) excursions briefly appear during
AN70-49
Application Note
switching interval when energy coming through transformer. These events minuscule that they barely measurable noise floor, damper will eliminate them. Measurement Technique Strictly speaking, measurement technique obtain lowest noise performance. Realistically, essenNote pedantic here. guilt this offense runs deep.
tial that measurement technique trustworthy. Uncountable hours have been lost chasing "circuit problems" that reality manifestations poor measurement technique. Please read Appendices before pursuing solutions circuit noise that isn't really there.1
APPENDIX PROTECTION AGAINST MAGNETICS NOISE KNOWLEDGE GOOD COMMON SENSE Roman-Coiltronics, Inc. Noise Test Data this test chose four most common magnetics geometries that currently production today. They follows: Core, Core, Core Toroid. following test data taken using methods described following paragraphs. test circuit used determine amount noise radiation shown Figure push-pull configuration, power ratings turns ratios were chosen align with Williams' noise designs presently under study. distance chosen Sniffer Noise Probe1 0.250 inches from surface core structure. This distance chosen result preliminary testing allow measurable reading smallest amount flux lines coming from quietest core structure. measured worst-case full load noise shown Figure each four geometries chosen this test. Note
INPUT DRIVE OUTPUT 0.255A,
that noise shown millivolts rather than gauss, conversion gauss VPROBE(mVP-P) 2.88mGauss/µs After taking noise reading each UUT's under their full load conditions, load resistor removed allow observation magnetizing flux noise. Then, using same measurement techniques, noise measured second time determine difference between load noise magnetizing noise. measured worst-case magnetizing noise shown Figure
Note Appendix Note Appendix GEOMETRY Core Core Core Toroid FULL LOAD MAGNETIZING 20mV 63mV 488mV 860mV
Figure Worst-Case Full Load Noise
GEOMETRY Core Core Core Toroid
FULL LOAD MAGNETIZING 16mV 49mV 95mV 91mV
AN72
Figure Test Circuit
Figure Worst-Case Magnetizing Load Noise
AN70-50
Application Note
Core core tested predicted quietest geometry ones tested. Just expected "Hot Spot" noise located window where leads exit core. Reference waveform shown Figure Note waveform shows voltage input, middle waveform shows noise recorded using amplifier bottom waveform current through UUT. Core core tested surprise group with much lower noise reading than would have originally thought possible. Reference waveform shown Figure Note waveform shows voltage input, middle waveform shows noise recorded using amplifier bottom waveform current through UUT.
Figure
Figure
AN70-51
Application Note
Toroid Toroid core (Figure tested showed much higher electronic noise than originally expected from closed-path geometry. worst-case noise came from core, with winding placed evenly core possible. Note waveform shows voltage input, middle waveform shows noise recorded using amplifier bottom waveform current through UUT. Core E-core (Figure showed highest concentration noise just above winding center device. noise surrounding sides measurable below field that found directly above below part. Note waveform shows voltage input, middle waveform shows noise recorded using amplifier bottom waveform current through UUT.
Figure
Figure
AN70-52
Application Note
Summary Figure graph showing relative difference comparing full load noise against magnetizing load noise. recognized that closed core structures such toroids inductors produce less stray (leakage) flux than open structures like cores bobbin cores. Some recent products offer "magnetic shields" tubes magnetic material around bobbin-type core attempt provide "magnetic shunt" flux follow. These structures offer very little reduction noise because high reluctance between "shield" inner bobbin core. reluctance this much higher than that magnetic material. resultant leakage flux from defeat purpose shield almost entirely! best approach reducing noise inductors true closed-field geometries, such toroid. When designing lowest possible noise transformer applications, important observe effects
CORE CORE LOAD NOISE CORE NOISE TOROID CORE
load current, opposed magnetizing current. preceding test demonstrates that traditional noise structures (toroid) radiate relatively high amounts leakage flux coupling characteristics between windings. reflected load currents both primary secondary affect magnetizing flux, create magnetic leakage field around wire, coupling less than perfect. This definition, leakage flux. size shape window area have effect coupling between windings, well shape flux field emanating from transformer. Winding technique also effect coupling noise. Multifilar winding, opposed layer winding, offer better coupling characteristics, which turn, lowers noise lessening leakage flux. Conclusion Every millivolt counts.
Figure
APPENDIX MEASURING RADIATION (Electromagnetic Interference) form switching regulator noise. radiated, opposed conducted, phenomenon. LT1533-based circuits produce amounts same reason they minimize conducted noise-controlled switching times. This appendix, guest written Bruce Carsten, describes excellent tool relative measurement it.1 Carsten's methods only show measure relative EMI, identify silence source.
Note Calibrated measurements discussed References
AN70-53
Application Note
APPLICATION NOTE E101: "SNIFFER" PROBE Bruce Carsten Associates, Inc. 6410 Sisters Place, Corvallis, Oregon 97330 541-745-3935 Sniffer Probe2 used with oscilloscope locate identify magnetic field sources electromagnetic interference (EMI) electronic equipment. probe consists miniature turn pickup coil located small shielded tube, with connector provided connection coaxial cable (Figure J1). Sniffer Probe output voltage essentially proportional rate change ambient magnetic field, thus rate change nearby currents. principal advantages Sniffer Probe over simple pickup loops are: Spatial resolution about millimeter. Relatively high sensitivity small coil. source termination minimize cable reflections with unterminated scope inputs. Faraday shielding minimize sensitivity electric fields. Sniffer Probe developed diagnose sources switch mode power converters, also used high speed logic systems other electronic equipment. SOURCES Rapidly changing voltages currents electrical electronic equipment easily result radiated conducted noise. Most switch mode power converters thus generated during switching transients when power transistors turned off. Conventional scope probes readily used dynamic voltages, which principal sources common mode conducted EMI. (High dV/dt also feed through poorly designed filters normal mode voltage spikes radiate fields from circuit without conductive enclosure.) Dynamic currents produce rapidly changing magnetic fields which radiate more easily than electric fields they more difficult shield. These changing magnetic fields also induce impedance voltage transients other circuits, resulting unexpected normal common mode conducted EMI. These high dI/dt currents resultant fields directly sensed voltage probes, readily detected located with Sniffer Probe. While current probes sense currents discrete conductors wires, they little with printed circuit traces detecting dynamic magnetic fields. PROBE RESPONSE CHARACTERISTICS Sniffer Probe sensitive magnetic fields only along probe axis. This directionality useful locating paths sources high dI/dt currents. resolution usually sufficient locate which trace printed circuit board, which lead component package, conducting generating current. "isolated" single conductors traces, Probe response greatest just either side conductor where magnetic flux along with probe axis. (Probe response little greater with axis tilted towards center conductor.) shown Figure there sharp response null middle conductor, with 180° phase shift either side decreasing response with distance. response will increase inside bend where flux lines crowded together, reduced outside bend where flux lines spread apart. When return current adjacent parallel conductor, Probe response greatest between conductors shown Figure There will sharp null phase shift over each conductor, with lower peak response outside conductor pair, again decreasing with distance.
Note Sniffer Probe available from Bruce Carsten Associates address noted title this appendix.
AN70-54
Application Note
response trace with return current opposite side board similar that single isolated trace, except that probe response greater with Probe axis tilted away from trace. "ground plane" below trace will have similar effect, there will counter-flowing "image" current ground plane.
1997, Bruce Carsten Associates, Inc.
Approx. 160µ Wire, 1.5mm Coil Dia.
Figure Construction "Sniffer Probe" Locating Identifying Magnetic Field Sources
AN70-55
Application Note
PROBE TRACE PROBE VOLTAGE PROBE VOLTAGE
TRACES PROBE
Figure Sniffer Probe Response Current Physically "Isolated" Conductor
Figure Sniffer Probe Response with Return Current Parallel Conductor
Probe frequency response uniform magnetic field shown Figure large variations field strength around conductor, Probe should considered qualitative indicator only, with attempt made "calibrate" response fall-off near 300MHz pickup coil inductance driving coax cable impedance, mild resonant peaks (with scope termination) multiples 80MHz transmission line reflections. PRINCIPLES PROBE Sniffer Probe used with least 2-channel scope. channel used view noise whose source located (which also provide scope trigger) other channel used Sniffer Probe. probe response nulls make inadvisable this scope channel triggering. third scope trigger channel very useful, particularly difficult trigger noise. Transistor drive waveforms their predecessors upstream logic) ideal triggering; they usually stable, allow immediate precursors noise viewed.
Start with Probe some distance from circuit with Probe channel maximum sensitivity. Move probe around circuit, looking "something happening" circuit's magnetic fields same time noise problem. precise "time domain" correlation between noise transients internal circuit fields fundamental diagnostic approach. candidate noise source located, Probe moved closer while scope sensitivity decreased keep probe waveform on-screen. should possible quickly bring probe down board trace wiring) where probe signal seems maximum. This near point generation, should near trace other conductor carrying current from source. This verified moving probe back forth several directions; when appropriate trace crossed roughly right angles, probe output will through sharp null over trace, with evident phase reversal probe voltage each side trace noted above). This "hot" trace followed (like bloodhound scent trail) find much generating current loop. trace hidden back side
AN70-56
Application Note
©1996, Bruce Carsten Associates, Inc.
Figure Typical "Sniffer" Probe Frequency Response Measured with 1.3m (51") Coax Scope Upper Traces: 1Meg Scope Input Impedance Lower Traces: Scope Input Impedance
inside) board, mark path with felt locate trace disassembly, another board artwork. From current path timing noise transient, source problem usually becomes almost self-evident. Several not-uncommon problems (all which have been diagnosed with various versions Sniffer Probe) discussed here with suggested solutions fixes. TYPICAL DI/DT PROBLEMS Rectifier Reverse Recovery Reverse recovery rectifiers most common source dI/dt-related power converters; charge stored junction diodes during conduction causes momentary reverse current flow when voltage reverses. This reverse current stop very quickly (<1ns) diodes with "snap" recovery (more likely devices with rating less than 200V), reverse current
decay more gradually with "soft" recovery. Typical Sniffer Probe waveforms each type recovery shown Figure sudden change current creates rapidly changing magnetic field, which will both radiate external fields induce impedance voltage spikes other circuits. This reverse recovery "shock" parasitic circuits into ringing, which will result oscillatory waveforms with varying degrees damping when diode recovers. series damper circuit parallel with diode usual solution. Output rectifiers generally carry highest currents thus most prone this problem, this often recognized they well-snubbed. uncommon unsnubbed catch clamp diodes more problem. (The fact that diode R-C-D snubber need snubber always selfevident, example).
AN70-57
Application Note
TYP. PROBE WAVEFORMS:
"SOFT" REVOVERY PROBE POINTS "SNAP" REVOVERY
Figure Rectifier Reverse Recovery Typical Fix: Tightly Coupled Snubber
problem usually identified placing Sniffer Probe near rectifier lead. signal will strongest inside lead bend axial package, between anode cathode leads TO-220, TO-247 similar type package, shown Figure Using "softer" recovery diodes possible solution Schottky diodes ideal voltage applications. However, must recognized that diode with soft recovery also inherently lossy (while "snap" recovery not), diode simultaneously develops reverse voltage while still conducting current: fastest possible diode (lowest recovered charge) with moderately soft recovery usually best choice. Sometimes faster, slightly "snappy" diode with tightly coupled snubber works well better than soft excessively slow recovery diode. significant ringing occurs, "quick-and-dirty" snubber design approach works fairly well: increasingly large damper capacitors placed across diode until ringing frequency halved. know that total ringing capacity quadrupled that original ringing capacity added capacity. damper resistance required about equal capacitive reactance
original ringing capacity original ringing frequency. "frequency halving" capacity then connected series with damping resistance placed across diode, tightly coupled possible. Snubber capacitors must have high pulse current capability dielectric loss. Temperature stable (disc multilayer) ceramic, silvered mica some plastic filmfoil capacitors suitable. Snubber resistors should noninductive; metal film, carbon film carbon composition resistors good, wirewound resistors must avoided. maximum snubber resistor dissipation estimated from product damper capacity, switching frequency square peak snubber capacitor voltage. Snubbers passive switches (diodes) active switches (transistors) should always coupled closely physically possible, with minimal loop inductance. This minimizes radiated field from change current path from switch snubber. also minimizes turn-off voltage overshoot "required" force current change path through switch-snubber loop inductance.
AN70-58
Application Note
Ringing Clamp Zeners capacitor-to-capacitor ringing problem occur when voltage clamping Zener TransZorb® placed across output converter overvoltage protection (OVP). Power Zeners have large junction capacity, this ring series with lead output capacitors, with some ringing voltage showing output. This ringing current most easily detected near Zener leads, particularly inside bend shown Figure snubbers have been found work well this case ringing loop inductance often lower than obtainable parasitic inductance snubber. Increasing external loop inductance allow damping advisable this would limit dynamic clamping capability. this case, found that small ferrite bead both Zener leads dampened oscillations with minimal adverse side effects high permeability ferrite bead quickly saturates soon Zener begins conduct significant current).
TransZorb registered trademark General Instruments, GSI.
Paralleled Rectifiers less evident problem occur when dual rectifier diodes package paralleled increased current capability, even with tightly coupled snubber. diodes seldom recover exactly same time, which cause very high frequency oscillation (hundreds MHz) occur between capacities diodes series with anode lead inductances, shown Figure This effect really only observed placing probe between anode leads, ringing current exists almost nowhere else (the ringing nearly "invisible" conventional voltage probe, like many other effects that easily found with magnetic field Sniffer Probe). This "teeter-totter" oscillation voltage "null" about where snubber connected, provides little damping (see Figure J7a). actually very difficult insert suitable damping resistance into this circuit. easiest dampen oscillation "slit" anode trace inch place damping resistor anode leads shown Figure J7b. This
PROBE POINTS TYP. PROBE WAVEFORM:
100-500MHz RINGING
Figure Ringing Between Clamp Zener Capacitor Typical Fix: Small Ferrite Bead Zener Lead(s)
AN70-59
Application Note
Probe Point
IRING
Package
Lead
Connections
Figure Ringing Paralleled Dual Rectifiers
increases inductance series with diode-diode loop external package leads, while having minimal effect effective series inductance. Even better damping obtained placing resistor across anode leads entry point case, shown Figure J7c, this violates mindset many production engineers. also preferable split original damper into (2R) (C/2) dampers, each side dual rectifier (also shown Figure 7c). practice, always preferable dual dampers, each side diode; loop inductance about half, external dI/dt field reduced even further oppositely "handed" currents snubber networks.
Paralleled Snubber Damper Caps problem similar that with paralleled diodes occurs when more loss capacitors paralleled driven with sudden current change. There tendency current ring between capacitors series with their lead inductances ESL), shown Figure J8a. This type oscillation usually detected placing Sniffer Probe between leads paralleled capacitors. ringing frequency much lower than with paralleled diodes (due larger capacity), effect benign capacitors sufficiently close together. resultant ringing picked externally, damped similar with parallel diodes shown Figure J8b. either case, dissipation damping resistor tends relatively small.
AN70-60
Application Note
Probe Point
IRING
Figure Ringing Paralleled "Snubber" Capacitors
Ringing Transformer Shield Leads capacity transformer shield other shields windings Figure forms series resonant circuit with "drain wire" inductance (LS) bypass point. This resonant circuit readily excited typical square wave voltages windings, poorly damped oscillatory current flow drain wire. shield current radiate noise into other circuits, shield voltage will often show common mode conducted noise. shield voltage very difficult detect with voltage probe most transformers, ringing shield current observed holding Sniffer Probe near shield drain wire (Figure J10), shield current's return path circuit. This ringing dampened placing resistor series with shield drain wire, whose value approximately equal surge impedance resonant circuit, which calculated from formula Figure shield capacitance (CS) readily measured with bridge capacity from shield facing shields and/or windings), usually best calculated from ringing frequency sensed Sniffer Probe). This resistance typically order tens ohms. more small ferrite beads also placed drain wire instead provide damping. This option preferable late "fix" when board already been laid out.
either case, damper losses typically quite small. damper resistor moderately adverse impact shield effectiveness below shield drain wire resonant frequency; damper beads superior this respect their impedance less lower frequencies. drain wire connection should also short possible circuit bypass point, both minimize raise shield's maximum effective (i.e., resonant) frequency. Leakage Inductance Fields Transformer leakage inductance fields emanate from between primary secondary windings. With single primary secondary, significant dipole field created, which seen placing Sniffer Probe near winding ends shown Figure J11a. this field generating EMI, there principal fixes: Split Primary Secondary two, "sandwich" other winding, and/or: Place shorted copper strap "electromagnetic shield" around complete-core winding assembly shown Figure J12. Eddy currents shorted strap largely cancel external magnetic field. first approach creates "quadrupole" instead dipole leakage field, which significantly reduces distant field intensity. also reduces eddy current losses shorted strap electromagnetic shield used, which important consideration.
AN70-61
Application Note
SHIELD PARASITICS
SHIELD RESONANCE DAMPING
SHIELD VOLTAGE FEEDTHROUGH
DAMPED WITH RESISTOR SMALL FERRITE BEAD: SHIELD RESONANCE
0.01
NORMALIZED FREQUENCY
Figure Shield Effectiveness High Frequencies Limited Shield Capacity Lead Inductance
AN70-62
Application Note
PROBE POINT (NEAR SHIELD DRAIN
TYP. PROBE WAVEFORM:
10-100MHz RINGING
Figure J10. Transformer Shield Ringing Typical Fix: Resistor Ferrite Bead Drain Wire)
TRANSFORMER LEAKAGE INDUCTANCE FIELD
INDUCTOR EXTERNAL FIELD
PROBE POINTS
TYPICAL FIXES: SANDWICHED WINDINGS: SHORTED STRAP SHIELD
TYPICAL FIX: EXTERNAL GAPS
Figure J11. Probe Voltages Resemble Transformer Inductor Winding Waveforms
AN70-63
Application Note
ELECTROMAGNETIC SHIELD FORMED SHORTED COPPER STRAP AROUND CORE WINDING
SHIELD INDUCTOR WITH LARGE EXTERNAL CORE GAPS WILL HAVE HIGH LOCALIZED EDDY CURRENT LOSSES NEAR GAPS
Figure J12. "Sandwiched" PRI-SEC Transformer Winding Construction Reduces Electromagnetic Shield Eddy Current Losses
AN70-64
Application Note
External Fields External gaps inductor, such those open "bobbin core" inductors with cores spaced apart (Figure J11b), major source external magnetic fields when significant ripple currents present. These fields also easily located with Sniffer Probe; response will maximum near near open inductor winding. "Open" inductor fields readily shielded they present problem inductor must usually redesigned reduce external fields. external field around spaced cores virtually eliminated placing center leg. Fields (possibly intentional) residual minor outside minimized with shorted strap electromagnetic shield Figure J12, eddy current losses prove high. less obvious problem occur when inductors with "open" cores used second stage filter chokes. minimal ripple current create significant field, such inductor "pick external magnetic fields convert them noise voltages susceptibility problem.3 Poorly Bypassed High Speed Logic Ideally, high speed logic should have tightly coupled bypass capacitor each and/or have power ground distribution planes multilayer PCB. other extreme, have seen bypass capacitor used power entrance logic board, with power ground from opposite sides board. This created large spikes logic supply voltage produced significant electromagnetic fields around board. With Sniffer Probe, able show which pins which larger current transients synchronism with supply voltage transients. (The logic design engineers were accusing power supply vendor creating noise. found that supplies were fairly quiet; poorly designed logic power distribution system that problem.) Probe with "LISN" test setup using Sniffer Probe with Line Impedance Stabilization Network (LISN) shown Figure J13. optional "LISN LINE FILTER" reduces line voltage feedthrough from 100mV microvolt levels, simplifying diagnosis when suitable voltage source available cannot used. TESTING SNIFFER PROBE Sniffer Probe functionally tested with similar that shown Figure J14, which used test probes production.
Note Note. Appendix additional commentary.
AN70-65
Application Note
LINE INPUT
LINE IMPEDANCE STABILIZATION NETWORK
LINE OUTPUT
(LISN)
EQUIPMENT UNDER INVESTIGATION
NOISE OUTPUT (BNC)
INPUT
OUTPUT (OPTIONAL) SCOPE
CABLE LENGTH EXT. TRIG. (OPTIONAL TRIGGER INPUT)
Figure J13. Using Probe with "LISN"
AN70-66
LISN LINE FILTER
"SNIFFER" PROBE
Application Note
SCALE 12.4, 1/4W M.F. RESISTOR
SCOPE (50TERM.)
SIGNAL GENERATOR
0.5"
3/16" 1/8" PLASTIC TUBE, 3/4" LONG (AVAILABLE MANY MODEL HOBBY SHOPS) TURNS WIRE WRAP WIRE SUB. MAGNET WIRE)
Sniffer Probe centered inside test coil where Probe voltage greatest. approximate flux density middle coil calculated from formula:
1.257
(CGS Units)
1.27cm long, 20-turn test coil, flux density about Gauss amp. 1MHz, Sniffer Probe voltage 19mV (±10%) 100mA load impedance, half that load.
Figure J14. "Sniffer" Probe Test Coil
AN70-67
Application Note
CONCLUSION Sniffer Probe simple, very fast effective means locate dI/dt sources EMI. These sources very difficult locate with conventional voltage current probes. SUMMARY summarized procedure using "Sniffer" Probe appears Figure J15.
2-channel scope, preferably with external trigger. scope channel used Sniffer Probe, which used triggering. second channel used view noise transient whose source located, which also used triggering practical. More stable reliable triggering achieved with "external trigger" channel) transistor drive waveform preceding logic transition), allowing immediate precursors transient viewed. (Nearly noise transients occur during, just after, power transistor turn-on turn-off. Start with Probe some distance from circuit with maximum sensitivity "sniff around" something happening precise sync with noise transient. Probe waveform will identical noise transient, will usually have strong resemblance. Move Probe closer suspected source while decreasing sensitivity. conductor carrying responsible current located sharp response null conductor with inverted polarity each side. Trace noise current path much possible. Identify current path schematic. source noise transient usually evident from current path timing information.
1997, BRUCE CARSTEN ASSOCIATES, Inc.
Figure J15. "Sniffer" Probe Procedure Outline
AN70-68
Application Note
SNIFFER PROBE AMPLIFIER Figure shows 40MHz amplifier Sniffer Probe. gain allows oscilloscope display probe output over wide range sensed inputs. amplifier built into small aluminum box. probe should connect amplifier cable, although termination does have high quality coaxial type. probe's uncalibrated, relative output means high frequency termination aberrations irrelevant. simple film resistor, contained amplifier box, adequate. Figure shows Sniffer Probe amplifier.
"SNIFFER" PROBE (SEE APPENDIX DETAILS)
LT1223
LT1223
LT1223
LT1223
OUTPUT SCOPE
-15V LAYOUT TECHNIQUES. SUPPLIES ±15V, BYPASS EACH AMPLIFIER WITH 0.1µF CERAMIC CAPACITORS. DIODE CLAMP SUPPLIES REVERSE VOLTAGE
AN72
Figure J16. 40MHz Amplifier Probe
AN70-69
AN70-70
Figure J17. Sniffer Probe Amplifier. Note BNC-Based Signal Transmission. ±15V Power Enters Separate Cable
Application Note
Application Note
APPENDIX SYSTEM-BASED NOISE "MEASUREMENT" ultimate test switching regulator noise effect system being powered. data below taken using LT1533 powering LT1605 16-bit converter. Crossplots integral differential nonlinearity shown bench supply LT1533 supply powered operation. difference within test systems limitof-error.
Figure Differential Nonlinearity Using Bench Supply
Figure Integral Nonlinearity Using Bench Supply
Figure Differential Nonlinearity Using LT1533 Supply
Figure Integral Nonlinearity Using LT1533 Supply
Figure Subtraction Above Plots. Residual Error Test System Limited
Figure Subtraction Above Plots. Residual Error Test System Limited
AN70-71
Application Note
AN70-72
Linear Technology Corporation
1630 McCarthy Blvd., Milpitas, 95035-7417 (408) 432-1900 FAX: (408) 434-0507q TELEX: 499-3977 www.linear-tech.com
an70 LT/TP 1097 PRINTED
LINEAR TECHNOLOGY CORPORATION 1997
Other recent searches
VTA52Z - VTA52Z VTA52Z Datasheet
VTA56Z - VTA56Z VTA56Z Datasheet
SN74CBT - SN74CBT SN74CBT Datasheet
SN74CBT6800 - SN74CBT6800 SN74CBT6800 Datasheet
SN74CBT6800A - SN74CBT6800A SN74CBT6800A Datasheet
Si4423DY - Si4423DY Si4423DY Datasheet
PSA0210A - PSA0210A PSA0210A Datasheet
IN7407 - IN7407 IN7407 Datasheet
IDT7210L - IDT7210L IDT7210L Datasheet
CY3207ISSP - CY3207ISSP CY3207ISSP Datasheet
BT136X-600E - BT136X-600E BT136X-600E Datasheet
BL55076 - BL55076 BL55076 Datasheet
Privacy Policy | Disclaimer
© 2012 Datasheet Archive
