Module Flow 11.1 Circuit Protection Overview 11.2 Circuit Protection Device Features and Options 11.3 Common Design Errors 11.4 Common Test Errors 11.5

Module Flow 11.1 Circuit Protection Overview 11.2 Circuit Protection Device Features and Options 11.3 Common Design Errors 11.4 Common Test Errors 11.5 eFuses 11.5a eFuse Overview 11.5b eFuse vs. Fuse 11.5c eFuse vs. Polyfuse 1

Circuit Protection Fundamentals 11.1 Circuit Protection Overview

Circuit Protection What is it ? Many things with many names Inrush Control Hotswap Hotplug Current Limiting Electronic Circuit Breaker Short Circuit Protection Soft Start Over Voltage Protection (OVP) eFuse Load Power Limiting FET SOA Limiting ( Protecting the Protector ! ) Reverse Current Protection (ORing) Often Required for Agency Rating UL, CSA North America EN, IEC, (CENELEC) Europe CCC Mark (CNCA) - China ~ the same functions 3

Circuit Protection What is it ? Circuits designed specifically to. Prevent Fire ! --Keep the smoke in! Keep small problems from growing big Minimize damage by quickly isolating failures Prevent potentially disruptive power bus disturbances One small transient can take down/reset an entire system What Gets Protected ? LOAD POWER FET CONNECTORSSUPPLY 4

Circuit Protection Where Is It Used ? Telecom Equipment Datacenters / Servers Storage / HDD, SSD, Midplanes Industrial Control 24 or 48 V typically Tower Mounted Antennas Merchant Power 5

Circuit Protection The Basic Parts Most Common Elements Location.. Sometimes on the Load Side of the Connector Sometimes on the Supply Side of the Connector LOAD BOARD BACKPLANE Element for controlling the FET Element for sensing current Element for modulating current 6

Thank you! 7

Circuit Protection Fundamentals 11.2 Device Features and Options

9 End customers cant make you design in protection circuits but they can make you wish you had. Design review wisdom

Device Features/Options Some of the Choices FET Internal or External Inrush control dV/dT, or di/dT Current Limit Always, Never, or just at startup Fault response Latch off or Retry Short Circuit Response Latch Off or Retry Control I2C or Analog Control Outputs Power Good, Fault, FET Fault ILIMIT Accuracy 20% Standard, 10% Pretty Good, 5% Very Good FET SOA protection.. Or not Allows use of smaller FET and provides very high survivability Current Indicator Output (IMON) Analog or Digital Output ? Digital Output requires internal ADC and typically includes PIV Monitoring ORing Control Linear or Hysteretic 10

Device Features/Options Internal FET vs. External FET 11 Internal FETExternal FET Highly Integrated -Few External Parts -Internal sense FET -Built it Power Limiting -Extremely well protected Compromises are made -FET process vs. Analog process -FET package vs. Analog package Require careful thermal design Generally not found in app > 5 A Flexible R DSON (Designers Choice) More feature options No limit on upper current limit Generally more accurate More external parts -R SENSE, FET -R S, C S for configuration Larger foot print

Device Features/Options FET SOA Protection One of the least understood but most appreciated features Allows use of smaller, less expensive FETs Analog multiplier calculates P DIS_FET in real time and compares result to PROG pin If P LOAD > PROG then gate drive reduced to lower I LOAD and P FET TI is now the ONLY manufacturer to offer true Power Limiting ! 12

Device Features/Options FET SOA Protection Dynamically adjusts I LIMIT to be approximately proportional to 1/(V DS ) 2 Limits P DIS of FET to programmed limit TPS2420 Startup into 15 , 700 F I LOAD V OUT TPS2420 Limits FET P DIS < 5 Watts P DIS V OUT I LOAD Orange = Violet x (V IN Blue) P DIS = I LOAD x V DS 13

Device Features/Options Power Limiting Startup into overload response SOA protection keeps FET safe even when starting up into a severe overload Fault timer limits T(ime) factor of SOA Some competitive devices will reduce I LIMIT over a limited range and with limited protection. ONLY TI has true FET SOA Power Limiting built into the Hotswap Controller !!! P DIS V OUT I LOAD CTCT TPS2420 Startup into overload 14

Positive Low Voltage Protection TI Device Portfolio Sample PART Input Range Package V THRESH (mV) I LIMIT Int. FET SOAOVI2CPG Imon Acc. TPS2420 3 to 20 QFN16 (4x4mm) Internal FET R DSON = 30 m I LIMIT = 10% @ 2 A AlwaysYes No Lo17%@2A TPS2590 QFN16 (4x4mm) - N/A TPS2421-1/2 SOIC8 Lo TPS24720 2.5 to 18 QFN16 (3x3)Prog Startup Only No Yes No LoProg TPS24710/1/2/3 MSOP10 25 10% YesNol/l/h/h N/A TPS24700/1 MSOP8 No Lo LM25066/A 2.9 to 17 LLP24 (4x5mm) 25 10% 46.5 11.8% AlwaysNoYes Hi 2.40% LM25066I/AI LLP24 (4x5mm) 1.00% LM25069-1/2 MSOP1050 10% NoN/A LM25061-1/2 MSOP1050 10% No TPS2480/1 9 to 26 PW2050 10% Yes 0.5% TPS2482/3 9 to 36 0.5% 15

Typical Inrush/OCP Design Steps 1.Select R SENSE to set I LIMIT and I FASTTRIP I LIMIT = V TH /R SENSE - V TH typically 25 50 mV Simplest controllers have fixed V TH High V TH Better Accuracy but Higher I 2 R Losses Fast trip (Short Circuit) threshold usually 1.5x -3x I LIMIT Level 2. Select C FAULT to get desired T FAULT Set T FAULT long enough to allow all downstream caps to charge (T CHARGE )before time out T CHARGE ~ CV/I (C = Bulk Cap, V = V OUT, I= I LIMIT ) Set T FAULT as short as reasonable to minimize FET stress during overcurrent events Ensure that T FAULT x V IN X I LIMIT is within SOA curve 3.Select FET that can withstand T FAULT x V IN x I LIMIT x ~1.5 ..SOA !! 4.Set FET SOA Power Limit on devices so equipped Design tools available for some devices - check webpage TPS24700/10/20, TPS2490/1/2/3, TPS2480/1, LM5064/6/7/9, LM25061/6/9 16

Questions To Ask During Design 1.Will a load get plugged into a live socket? 2.Will a load get unplugged from a live socket? 3.Is it OK for supply to collapse if one load shorts? 4.Are multiple loads connected to a common supply? OK for all loads to shut off if one load shorts? 5.Do loads need ability to ride through transients? 6.Do loads needs protection from voltage surges? 7.Do loads have large capacitance on the inputs? 8.Are multiple supplies powering the load or bus? 17

Thank you! 18

Circuit Protection Fundamentals 11.3 Common Design Errors

Common Design Errors SOA of FET too Small Layout Issues Inadequate Transient Protection 20

SOA of FET Too Small I DS Drain to Source Current 10 3 10 2 10 1 10 0 10 -1 10 -2 10 -2 10 -1 10 0 10 1 10 2 V DS Drain to Source Voltage V DS_MAX I D_MAX R DSON 1 ms 10 ms 100 ms 1 s DC SOA = Safe Operating Area SOA Chart Every FET has one Defines Bounds of FET Operation V DS_MAX = Vertical Limit I D_MAX = Horizontal Limit R DSON limits I D at lower voltages Theoretical P MAX = 3000 W Fault Time Is Critical Longer Fault time means bigger FET Shorter Fault Time allows higher peak power Most Stressful FET Events Startup into short Shorted load while under full load 21

SOA of FET Too Small Example - 12 V, 50 A Server 10 3 10 2 10 1 10 0 10 -1 10 -2 10 -2 10 -1 10 0 10 1 10 2 V DS Drain to Source Voltage I DS Drain to Source Current 1 ms 10 ms 100 ms 1 s DC Without Power Limiting P MAX = I LIMIT x V SUPPLY = 600 W T SOA_MAX = ~800 us With Power Limiting P MAX_LIMITED = 50 W As V DS increases I LIMIT is reduced T SOA_MAX = 10 ms Smaller FET can be used @ 50 A will start limiting when V DS > 1V Common Error to Pick FET Too Small 12 V 50 A 22

Layout Issues - A Little Stray R Can Make a Big Error Critical Kelvin Connections Sense Runs Critical Short Run Ground Gate High Current Runs Poor Kelvin Runs Inaccurate/variable ILIMIT Poor High Current Runs Voltage droop Power loss Overheating 23

Inadequate Transient Protection All wires are inductive Inductance stores energy When the FET turns off, voltage spikes occur LOAD CURRENT LOAD VOLTAGE Positive Spikes at Input to Switch/FET Negative Spikes at Output of Switch/FET 24

Inadequate Transient Protection To squelch inductive spikes from supply / load leads Caps and/or TVS at UUT Input to clamp positive spike Schottky Clamp across output to clamp negative spike Short, Wide Leads and Runs 25

Thank you! 26

Circuit Protection Fundamentals 11.4 Common Test Errors

Common Test Error Sources Current Probes Electronic Loads Transients From Long Supply Leads Supply ILIMIT Too Low 28

Current Probes Current Probe Behavior Great For Observing Waveform Shapes Dont have to be In The LoopNice !! Simply Clamps around feed or RTN wire Need Frequent Degaussing/Cal Not So Great for Precise Measurements Limited Bandwidth 1% Accuracy at Best 29

Current Probes For precise DC current measurements If I LOAD < 10 Amps use Multimeter on Current Scale If I LOAD > 10 Amps Use Shunt and Multimeter Pick R SENSE so V RSENSE @ I LIMIT = 50-100 mV Note.Now V OUT_SUPPLY VIN_LOAD...so measure V LOAD at The Load! 30

Electronic Loads Good for DC Loading and Automated Tests Proper Setup Very Important Ex. - Constant Current, Constant Power, Constant Resistance Butoften Have Switch Transients When Stepping Load Transients Can Cause Premature Trip When Measuring ILIMIT So What Do We DO ?? 31

Electronic Loads For Minimal Transients While Adjusting Load 32 Method 1: Use Power Resistors as Loads Method 2: Use Power FET as Load A bit tedious and Old School but accurate A collection of fixed and variable resistors is best Apply Last Half Amp With Small Wire Rheostat Can be effective with eLoads also Connect FET and Series Resistor as Load Adjust Potentiometer to vary Current Make Sure the FET can Handle the power !!!!

Long Supply Leads All wires are inductive Long Supply Leads can have significant L Lab Test Environment Usually Worse Than Final Application! Reason for TVS and diodes on most TI EVMs When the FET turns off, voltage spikes occur To counter inductive spikes from supply / load leads Caps and/or TVS at UUT Input to clamp positive spike Schottky Clamp across output to clamp negative spike Twisted Supply leads 33

Supply I LIMIT Too Low Lab Supply Limit Sometimes Set Slightly Above I LIMIT_LOAD V SUPPLY can sag due to I limiting during overload / short circuit testing Sagging V SUPPLY can cause UV shutdown before I LOAD reaches I FASTTRIP UV Shutdown is typically much slower than Fast trip (SC) Shutdown Slow shut down can violate FET SOA, resulting in dead FETs Fix 1: Ensure PS set to supply currents ABOVE fast trip level Fix 2: Attach bulk caps at input of UUT before test is run 34

Trends in Circuit Protection Accuracy Current limit, power limit, monitoring Efficiency Low R DSON, low I Q High levels of integration i.e. bring FET, R SENSE into the package I 2 C, PMBus for control and monitoring Especially PMBus with Intel Grantley processors eFuses replacing/augmenting fuses & polyfuses High Power POE Systems (25-100 Watts) 35

Thank you! 36

Circuit Protection Fundamentals 11.5a eFuse Overview

Integrated Circuit Protection Types Power controlling element contained therein Initial $ Level of Protection Wishful Thinking Fuse eFusePolyfuse (PTC) 38

What is an eFuse? An active circuit protection device that Will: Limit current at inrush Prevent load or source damage due to OC events Have an internal FET to control the load current Might: Provide OVP (none, fixed, adjustable) Have adjustable fault time and/or current limit Have indicator outputs (Fault, PG, etc.) Be able to control turn on slew rate Have a load current indicator output Be on source side of a connector or load side or.. Be nowhere near a connector 39

Typical Applications for eFuse Enterprise Class SSD SAS HDD Storage Server Chassis Set-Top Box DVD Player Internet TV m-SATA SSD Appliances 40

Brand Damage A Hidden Cost Its not fair and that wont change Most end users dont know or care how a product works Even fewer know or care about circuit protection A good fuse design in a bad system can still get the blame Dirty power, poor transient control, can cause a fuse to blow A load with a blown fuse is viewed as the problemnot the faulty source Blame should go to the source of bad power...but rarely does END CUSTOMER DOESNT CARE about power specs !!! Your board died.now fix it!!! Replacement board likely to blow a fuse, too Customer not happy switches to competitive brand Control your products destiny !! Dont rely on other systems to do the right thing Protect the product, the brand, the profit, your career ! Someone will pay.dont let it be you! 41

Backend costs of fuse only designs Tangible costs Replacement of nonfunctional product RMA admin costs / time Shipping broken/new devices from/to customer Truck rolls, service personnel Intangible backend costs Unhappy retailers Brand damage Loss of customer(s) In the end, we all want happy customers Its that simple, its that complicated 42

Thank you! 43

Circuit Protection Fundamentals 11.5b eFuse vs. Fuse

Why Not Use a Fuse? Slow Inaccurate Lossy Leave a load unpowered after event 45

Fast Blo Fuse Trip Time vs. Current eFuse vs. Fuse Time and trip limit inaccuracies mean bigger power supplies eFuse Limit ! Fusible fuse trip range eFuse trip range Time (sec.) 46

Fuses Are SlowEven the Fast Ones eFuse Performance I LIMIT is programmable, predictable, and stable over temp Bus droop and supply stress reduced by tight over current tolerance 47

Fuses are Lossy Higher resistance -> more energy -> more heat -> higher OPEX 13x more power lost with fuse! 800 mV/2A = 400 m vs. eFuse @ 30 m Lifetime cost of 1 Watt = $2 to $18 ( customer supplied numbers) Includes energy cost, distribution infrastructure, HVAC, product life Little Fuse 231Series 0.12 0.19 0.30 0.48 0.75 1.19 Lower Losses using TPS2590 ( 30 m ) 48

Fuses are Inaccurate Fuse makers recommend the I NOMINAL < 75% I FUSE_RATED Power supplies must be overspecd Accommodate fuse derating, fuse tolerance, P FUSE Bigger supplies = more CAPEX, more OPEX Seconds 49

Fuses Behavior is Sloppy and Stressful 50 During OverloadAfter Overload Much slower than eFuse No active current limiting Uncontrolled turn off time Bus droop likely More stress on supplies & load High I 2 R losses 10x+ nom. trip current for 3 ms No auto reset Inoperative system Module, fuse, or system must be replaced Repair costs Field returns Unhappy Customers

Fuse Summary 51 ChallengesBenefits Slow Lossy Inaccurate Load unpowered after event Low Cost Can provide Safety Compliance -UL, IEC, CSA

Fuses DO Excel in Some Apps! 52

Thank you! 53

Circuit Protection Fundamentals 11.5c eFuse vs. Polyfuse

eFuse vs. Polyfuse 55 eFuse (USB Power Switch)Polyfuse Current based I LIMIT Stable, accurate (20% - 30%) I LIMIT Fixed or Programmable I LIMIT Repeatable I LIMIT0 Fast ( < 1.5 us typ) Wide temp range -40 to +125 C Temp based I LIMIT Sloppy, variable I LIMIT No Programmable I LIMIT R ON Increases with each event Slow to trip (several ms) Not usable above 85 C Auto-resets after trip event

Polyfuses (PTC Devices) Require Derating Curve D TPS2420/21, TPS2590/910 D D

eFuse vs. Polyfuse Brand conscious Tier 1, 2, 3s use USB Power Switch Low cost bottom end apps may use Polyfuse True story #2 Major ODM experiencing Power supply resets during STB short testing. TPS2066C with faster response got designed in and no more resets. True story #1 Low end desktop maker melted wireless datacard during a short condition. Three times. Now using USB switches.

Polyfuse Summary 58 ChallengesBenefits Slow Lossy 2x regular fuse Inaccurate Each OC event increases resistance Not suitable for high temp. R increases with Temp.- Resets after OC event Low Cost Can provide Safety Agency compliance UL, IEC, CSA

Thank you! 59

Documents

Module Flow 11.1 Circuit Protection Overview 11.2 Circuit Protection Device Features and Options 11.3 Common Design Errors 11.4 Common Test Errors 11.5