0284.Boost Converter Design Tips[1]

8/12/2019 0284.Boost Converter Design Tips[1]

1/45

1

B o o s t C o n v e r t e r D e s ig n T i p s

Alan Martin

Field Appl ications EngineeringMarch 2005

MODERATOR:

National Semiconductor Online Seminar. Im Wanda Garrett and I will be your host. Before webegin Id like to go over the operation of your seminar interface. Slides will appear in the upper rightsection of your interface. If you wish like the slides to be larger, click the Enlarge button. Slides willautomatically advance or you can click on the slide title in the list on the left to jump to that place in

the presentation. At the bottom of your interface is an interactive Web browser set to a Web pagecontaining additional resources for this seminar. Questions may be submitted at any time and thepresenter will respond via email. To ask a question, click the Ask A Question button, then fill in theform that appears, and then submit the form. Finally, National Semiconductor owns and isresponsible for all content in this seminar. Todays topic is Boost Converter Design Tips. Todaysseminar will be given by Alan Martin, Field Applications Engineer. Welcome, Alan. Please go aheadwith your seminar.


2/45

2

2003 National Semiconductor Corporation

Who s h o u ld a t t e n d ?

Engineers designing a boost application

Engineers and technicians getting ready todebug a boost application.

MARTIN:

Thank you, Wanda. How do you do? My name is Alan Martin. Im a Field Applications Engineer hereat National Semiconductor. Ive been designing medium and low power switchers for nearly tenyears. This seminar is intended for engineers designing a boost application and engineers andtechnicians getting ready to debug a boost application.


3/45

3


S c o p e o f p r e s e n t a t i o n

Covering global topics that are missing fromIC data sheets that relate to the design andapplication of boosts.

Not covering component value selectionThese are covered in the data sheets.

Several on-line simulators assist withcomponent selection.

Give NSC WEBENCH a try!

Today well be covering global topics that tend to be missing from IC datasheets that tend to relate tothe application of boosts. Were not going to be covering value selections because these are wellcovered in datasheets as well as in simulation packages that are available for specific ICs. Werecommend you also take a look at National Semiconductors WEBENCH application package.Theres a Web link on one of the final slides of this presentation.


4/45

4


A g e n d a

Boost Converter BasicsTerminologyBoost Schematic FeaturesLimitations

Application Pitfalls and Gotchas Control section typesAdded circuitry for improved performance Component choices Instrumentation PCB Layout and routing guidelines

During todays seminar, we will be covering boost converter basics which includes some terminology,some schematic features, and some important limitations that youll be faced with, with boostconverters, and these are often not highlighted in datasheets. So this leads into some of theapplication pitfalls and gotchas that you will be faced with in using these circuits. Well cover controlcircuit types in the control section of the IC. Theres many types of those and you have to read thefine print in the datasheet to categorize which type of the IC control section is being used. Also,

added circuitry that you can get for improved performance and for bypassing some of the applicationpitfalls. Well cover component choices, instrumentation, and PCB layout and routing guidelines.


5/45

5


T e rm i n o l o g y

Boost = Step-Up Buck = Step Down Catch Diode = Rectifier

Manufacturers preference which terms to use

Switching Regulator -> Internal power switch Switching Controller -> External pwr switch

Synchronous = Higher efficiency w/ activerectifier P-channel mosfet w/ diode

Synchronizable = External osci llator Sync

First of all, theres terminology that varies among manufacturers. Boost converter is also known as astep-up converter. These are interchangeable terms; they mean exactly the same thing but its up tothe manufacturer in terms of which term that they choose to use in their literature. Likewise with buckconverters or step down converters, those are more common. These are, again, interchangeableterminologies. In IC circuit applications, catch diode is usually what is referred to but if you go to adiode manufacturer its always a rectifier. So these two terms are interchangeable yet its catch

diodes that you tend to find in the theoretical text of boost and buck converters.

Next is some subtle difference in the naming of an IC. If its a switching regulator, that implies theresan internal power switch. If its a switching controller, that implies theres an external power switch.So switching controllers tend to be used in the higher power applications where theyre more flexiblein terms of being able to scale the design for your particular application.

Next are the terms synchronous and synchronizable. Be very careful when you see these.Synchronizable implies theres an external input for oscillator synchronization so you can establish aknown operating frequency for a switching mode supply. Synchronous implies its a higher efficiencyversion of a boost or buck converter and in the boost case that theres a P-channel mosfet in parallelwith the rectifier diode to improve the efficiency. So these two terms can easily be confused.


6/45

6


Mo r e t e r m i n o lo g y

Junction of inductor diode and switch iscalled The Switch Node abbreviated onschematics as:VSWSWLXManufacturers preference which term to use

Next is that the term switch node is abbreviated on schematics and each manufacturer has their ownabbreviation: VSW, SW, LX. All of these imply the same thing and this is the junction of the inductor,the diode, and the switch.


7/45

7


B a s ic b o o s t s c h em a t i c

1

2

3

Swi t chNodeVIN

R2

FB

COUT

GND SNS

VOUT (>VIN)

CIN

R1

Rsns

VI N

D2 1

GATE

L

Next we come to the basic boost schematic. I recommend you print this page because well bereferring to it through the seminar. Beginning in the upper left-hand side is the input voltage. This isfiltered by the input capacitor. Going across the right, we come through the power inductor and outthe right-hand side through the rectifier diode to the output. Down the center is the switch node whichgoes to ground. Usually, there is a sense resistor built in there if it is a current mode boost. Mostboost converters are current mode; that type of controller is highly recommended as opposed to

voltage mode boost which omit this sense resistor. There is a feedback path through a divider of tworesistors and that enters the feedback input of the switching converter and that is what establishesthe regulated output voltage. Its important to note that these are truly step-up converters; that theoutput voltage must always exceed the input voltage for these devices to remain in regulation.


8/45

8


B o o s t S c h em a t i c f e a t u r es

Input Capacitor Inductor Catch Diode (Rectif ier) - usually schottky Output capacitor Common ground input to output= (no dc isolation) = 3 terminal application

NPN bipolar or N-channel mosfet switch Integrated switch or control ler for ext sw Feedback divider Sometimes built-in

Internal reference and error amp (not shown)

Page 8 is merely a listing of the items I just covered with one exception that internal to the IC is avoltage reference and an error amplifier. Sometimes those are accessible but most converters thatsintegrated and really not shown except in the block diagram in the datasheet for the IC.


9/45

9


O t h e r c o n t r o l s e c t i o n fe a t u r e ss om e t i m e s a va i la b l e

Enable

Soft start limits inrush currentLimits output voltage overshoot

Compensation - RC or CRC networkCertain types of control need for stabilitySometimes internal compensationless flexibleCertain types of control inherently stable (May also provide extra control input for other features)

Current sense input for current mode contro l Reference voltage bypass Built-in rectifier diode Built-in synchronous rectifier

Fully integrated regulatorsTrue shutdown function

Negative feedback input for vo ltage inverting conf igurations Many boost ICs can be configured as Flyback, Sepic or Ck

There are other features which may be available on a boost IC such as enable input that will ceaseswitching activity on the IC. Soft start which are always a good idea but this is a relatively uncommonfeature. This will limit the inrush current and can limit the output voltage overshoot. There are certaintypes of controllers which require a compensation network. This is usually a series RC network or anRC network shunted with another capacitor. These are needed for control loop stability. Often this isa place inside the IC which may be less flexible but certainly occupies less space on the board if

thats done. There are types of controllers which do not require compensation in that theyreinherently stable. Those tend to be lower performance architectures. An advantage of having acompensation pin is that it can provide an extra control input for other features.

As shown in the schematic on page 7, there may be a current sense input for current mode controlassuming that this is a controller. If its a regulator where the switch is built-in, that sense resistor isintegrated inside of the IC. Some manufacturers have a reference voltage bypass for improving thenoise characteristics of the internal reference. They may build in the rectifier diode or, in fact, theymay go beyond that and make a built-in synchronous rectifier. This gives much higher efficiency inthat the forward-drop of the rectifier diode is shunted. That only exists in fully integrated regulatorsthough. And this may also include a true shutdown function which well be talking about.

Occasionally there are ICs which have a negative feedback input so you can put the IC in an inverting

configuration such as a Ck. And another point is that many boost ICs can be used in Flyback orSepic or Ck converters assuming that they have a high enough peak switch voltage.


10/45

10


Bo o s t C o n v e r t e r P i t f a l l s

No true shutdown = output disconnectDisable switcher and output falls to Vin, less adiode drop.

Read data sheet fine print - Parts with trueshutdown will brag about it.

True shutdown requires an extra switch eitherin series with input or output.

One of the major shortcomings of a boost converter is that they do not have true shutdown. Now,another name for true shutdown is output disconnect. In the case where you disable the switcher,theres a direct path from Vin to Vout through the inductor and diode. So when you disable theswitcher, the output voltage drops down to Vin less a diode drop. Now, if an IC does include trueshutdown, they will brag about it on the datasheet and thats something to look for in the fine print.But if youre looking at an IC thats similar to the schematic on page 7, it does not have true

shutdown so youll have to either add an extra switch either in series with the input or the output. Ifits in series with the input, it could be as simple as an on/off switch.


11/45

11


S i m p l e t r u e s h u t d o w n c i rc u i t w i t h p np b ip o l a r

GND

R1

L VOUT

FB

Swi t chNode

VI N1

2

3

GATE

CIN

1UF

R2

VIN 1

32

D2 1

COUT

On the following two pages, Ive drawn two applications which can provide true shutdown using anexternal transistor. So in the first circuit on page 11, you can insert a pnp transistor in series with theoutput. Now, notice that the feedback divider is still attached following the diode so even in the trueshutdown case, there will be some slight leakage current through the feedback divider so thoseresistor values need to be high.


12/45

12


S i m p l e t r u e s h u t d o w n c i rc u i t w i t h p -c h a n n e l m o s f e t

GATE

L

FB

R2

1

32

CIN

D2 1

COUT

GND

VIN

VI N

R1

1UF

VOUTSwi t chNode

1

2

3

The following circuit shows a similar way how to do it with a p-channel mosfet. This may be moreattractive because you dont have the VBE loss. You only have the loss with the RDS on of the p-channel transistor. This is suitable up to the breakdown voltage of the mosfet gain. So this is limitedto applications below about 20 volts.


13/45

13


S i m p l e t r u e s h u t d o w n c i rc u i t a c c o u p l e s w i t c h n o de

1

2

3

D22 1

FB

D1

2

1

L

CIN

R2

Cc VOUT

GATE

VIN

R1

GND 10

COUT

Swi t chNode

VI N

The following is a technique that does not exist in literature. It is a charge pump approach whereyoure AC coupling the switch node. The thing that is not obvious about this is that theres no net DCacross the coupling capacitor, only a diode drop. This provides true complete shutdown in that thevoltage feedback divider can be coupled directly from Vout.


14/45

14


+ /- D u a l O u t p u t S u p p l y

D22 1

10

D1

2

1

Cc

L

10

COUT

R1

Swi t chNode

D1

2

1

GND FB

COUT

VOUT

VI N

Cc

CIN

R2

1

2

3

GATE

-VOUT

D22 1

VIN

An extension of this circuit is to provide a true shutdown circuit that provides both positive andnegative Vout so you can come from a battery input source and generate +/- 5 or +/- 12 from a lowerinput supply and this is very attractive for running analog op amp circuits. Be aware that only thepositive output is truly regulated and that the negative output goes along for the ride. And it will bewithin a few tenths of a volt of what the positive Vout is depending on loading.


15/45

15


Bo o s t C o n v e r t e r P i t f a l l s

No current limit on outputUncontrolled path from Vin to VoutEven when IC is in shutdownOutput short passes to input.A short term short on Vout or switch can

cause over voltage damage to switch. Usecare when probing with test leads!

Now, again referring to the schematic in figure 7, one of the other important limitations of a boostconverter is that there is no current limit on the output. So if you short the output, theres anuncontrolled path from Vin to Vout flowing through the inductor and diode. It doesnt matter if the IC isin shutdown when you do this. The output short is passed through to the input. There are caseswhere even if you momentarily short the output voltage, you can cause damage to the switchbecause you cause an escalating current in the inductor and when you release this short, the energy

in that inductor charges the output capacitor past the switch voltage reading. So its very easy todamage boost converters if you slip with a test lead so be very cautious with that.


16/45

16


Ov e r l o a d e d b o o s t c o n v e r t e r

GATE LOW

D2 1L VOUT < VINVIN

R2

GND

1

2

3

COUT

R1

CIN FB

Swi t chNode

VI N

Theres an intermediate condition. Lets suppose youve overloaded a boost converter and youvemuscled the output voltage down below Vin through a low impedance path. The switch converter maystill be operating so while you have this escalating current through the inductor, every time the switchturns on, it is attempting to shunt this high current through the switch to ground. This can causeoverheating in the boost converter and it may, in fact, cause damage.


17/45

17


Mo r e b o o s t c o n v e r t e r p i t f a l ls

Input voltage surges pass through to outputAs Vin rises up to Vout ( a regulated voltage

level), the switcher stops switching. Vinpasses through inductor and diode to Vout.

No input-output isolation Look at flybackconverter with opto-isolated feedback if youneed isolation.

Another important point is that as the input voltage surges, that this will be passed through to theoutput. So as Vin rises up past the regulated Voutput voltage level, the switcher stops switching andVin passes through the inductor and the diode to Vout. So thereyou can cause damage to yourload circuitry if theres an input surge occurring. Another shortcoming is that theres no isolationbetween input and output. This is a three terminal connection. Circuitry input and output grounds arecommon. If you need an isolated converter where theres no DC continuity between input and output

grounds, take a look at the flyback converter using opto-isolated feedback.


18/45

18


B o o s t s c h em a t i c v a r i a t i o n s :S e p a r a t e V I N, Vb i a s

GATE

VOUT (>VIN)

0.1UF

VbiasVI N

R2

GND

1

2

3

VIN

FB

D2 1

COUT

CIN

R1

L

Now lets look at some variations in the schematic. Notice that the IC controller or regulator itself hasits own input power supply terminal labeled Vin. This can come from a separate Vbias supply. Thismay be above or below Vin. So there are cases where its advantageous to run your controller fromone supply voltage yet the energy that youre converting in the boost converter comes from adifferent voltage source.


19/45

19


B o o s t s c h em a t i c v a r i a t i o n s :V b ia s b o o t s t r a p pe d f r om V OUT

CIN

R1

0.1UF

COUT

GND

VOUT (>VIN)

GATE

R2

L

VI N

FB

D2 1

VIN

1

2

3

A further variation on that is you can bootstrap the Vin voltage of the regulator from the output. Thiswill give improved regulation in applications where your Vin is declining towards zero. So this is oftenused in battery-operated products such as a double A cell source where your supply begins at 3 voltsbut runs down toward 1.8 at the end of discharge.


20/452

20


B o o t s t r a p p in g V b ia s f ro m V o u t

Controller bias supply can be bootstrappedfrom Vout, providing higher drive voltage forthe switch gate.

Permits switcher to continue operation tolowest possible input voltage. Common for 2AA cell battery supply devices that start at 3Vbut run down to 1.8V. (Wont start at 1.8V!)

Allows complete 3 terminal f ixed voltageboost regulators. Low complexity - No

features - low performance.

Now, in many applications, the boost converter wont restart if your batteries are dead and sitting at1.8. But, you know, that is a convenient way in integrated ICs to extend the operating range of theswitching converter.


21/452

21


3 T e rm i n a l B o os t

VOUT (>VIN)D2 1

CIN

L

12VFB

VI N SWI TCH COUT

GND

VIN

Theres a final variation on this where you connect the Vbias terminal to the output and this permits athree terminal integrated switcher. Now, these are only available with six output voltages but you docome across these periodically. This configuration has really not caught on but it existed when PO92310 packages were common and theres still some Asian manufacturers that build these in threeterminal SOT-23 packages. So thats how they accomplish this converter with so few pins.


22/452

22


C om m o n t y p e s o f b o o s t c o n t r o ls e c t i o n s

Gated OscillatorSimple control sectionUsually fixed on-timeDifficult to filter low freq ripple

Pulse Frequency mode - PFMGood light load efficiencyDifficult to filter low freq ripple

Constant frequency current mode PWMMost modern parts use thisRequires a current sense method

SenseFET Sense resistor internal or external

PFM / PWM dual modeBest of both approaches

Next well cover common types of boost control sections. The simplest type is the gated oscillator.These are rather hysteretic in their performance. Theyre low performance overall and its difficult tofilter the low frequency ripple on their output. These are often fixed-on-time architectures so the dutycycle does not vary while theyre operating. More recently, theres a type called pulse frequencymode or PFM. These give excellent light load efficiency but at low currents its difficult to filter the lowfrequency ripple out of them.

The most common type of high performance part is the constant frequency current mode PWM. Mostmodern parts use this. It does require a current sense mechanism for current sensing the peak switchcurrent. This can either be some sort of sense strap or a grounded sense resistor at the bottom of theswitching transistor either internally or externally. And then theres combinations of these two that arecombined PFM/PWM dual mode. So those give excellent low load efficiency and good performanceat full load current.


23/452

23


Cu rr e n t m o d e b o o s t c u r r e n ts e n s i n g

1

2

3

Swi t chNodeVIN

R2

FB

COUT

GND SNS

VOUT (>VIN)

CIN

R1

Rsns

VI N

D2 1

GATE

L

As I was mentioning, theres a sense resistor in series with the switch itself as drawn here in figure23. This diagram is actually identical to figure 7 earlier. I guess the important point is that senseresistor is a low value, low inductance device and

MODERATOR:

I think, Alan, I think that also part of your point is that the sense resistor has to be there.MARTIN:

Well, yeah, okay. Thank you for helping me recover there. In that this is an important part of thecontrol loop and the operation of it.


24/452

24


A t t a c h i n g t e s t e q u ip m e n t t o ab o o s t c o n v e r t e r (1 )

Apply a current probe to the input side of theinductor from Vin. Otherwise, the straycapacitance of the current probe head maycause switching performance problems.

Added inductance of the current probe isminor effect.

Measure output vol tage ripple directly acrossterminals of last output capacitor. PCBtrace inductance wil l add to reading.Eliminate the scope probe ground leadduring measurement.

Lets move on to attaching test equipment. This is one of my favorite areas because Im a testequipment junkie. And if we look at that schematic, theres two places where you can measureinductor current. You can either do it on the input side of the inductor or the output side of the inductorat the switch node. Well, the input side is the recommended point and the point there is that the ACvoltage on the input side of the inductor is bypassed by the input capacitor and thats the appropriateplace to put your current probe. If you place it on the switch node, the AC signal at the switch node

will couple into the current probe and increase losses and may cause a performance malfunction.Adding a current probe does increase the value of the inductor but its insignificant and it has noeffect because youre merely adding inductance to the current probe in series with the inductor thatsthere so it becomes part of the circuit.

The next thing is when measuring the output voltage ripple, do so directly across the place of theoutput capacitor. And its important to eliminate the scope ground lead if youre making thismeasurement. Also, the output capacitor may be several capacitors in parallel and it should be thelast of the output capacitors, the one thats farthest away from the output diode. And its important towatch out for the PCB trace inductance because that can add to the ripple reading.


25/452

25


A t t a c h i n g t e s t e q u ip m e n t t o ab o o s t c o n v e r t e r (2 )

Feedback input is a sensitive node. Avoidmeasurement with a multi-meter withunshielded leads. Use a 10X scope probe andscope or attach a series 10K resistor beforeconnecting the multi-meter lead.

NEVER short the feedback input while theswitcher is running. Vout will increase untilsomething breaks.

NEVER open the feedback loop either.

The next issue is that the feedback inputs a very sensitive node and you may be tempted tomeasure this with a multi-meter to see if, in fact, it agrees with what the datasheet spec isusually,its a bandgap 1.25 volt. But I assure you, thats not a good idea because you may inject signal fromyour voltmeter leads and interfere with the performance of the switcher. So using a 10X scope probeor, at the very least, a 10K resistor in series with a multi-meter lead is recommended.

One of the big gotchas on the bench is never short the feedback input while the switcher is runningbecause this opens the loop, the switching duty cycle will increase, and the output voltage willincrease until something breaks. So you could have an output capacitor fail, the diode can fail, theswitch can fail. Basically, youve got to start over when this occurs. And likewise, never open thefeedback loop. Thats R1 in the schematics; thats the voltage divider. Never open that either; thats

just as destructive as shorting the feedback input.


26/452

26


E ff i c i e n c y M e a s u r em e n t

Connect voltmeter leads directly to benchcircuit . (Kelvin connection)

Then have separate leads for applying powerand connecting to load.

Efficiency =VOUTx IOUT

VINx IIN

D.U.T.

Volts

LoadSource

VOUTVoltsVIN

AmpsAmps

IIN IOUT

Next we come to the subject of efficiency measurements and proper connection of equipment. Itsvery important to Kelvin connect the voltmeter directly to your circuit. It is surprising how muchvoltage drop you get across bench type leads. Likewise, its surprising how much voltage drop youget across the internal shunts of ammeters. Now, sort of the standard in voltmeters, 6.5 digit benchvoltmeters, is a HP 34401 meter. Its got an excellent ammeter section but I believe the ammetershunt internally is 4 ohm and in low voltage circuits, that can lead to substantial errors. So its wise to

invest in some ammeter shunt if youre doing currents above 2 amps or specifically if youredesigning power supplies made to drive LEDs because the ammeter shunts will cause significanterrors when youre trying to calculate efficiency.


27/452

27


L o a d i n g a s u p p l y u n d e r t e s t

Need a load that is easily varied Power rating of load must be greater than the supply undertest

Clarostat 240-C Power resistor decade box 50W/5A limit on lowest range Useful to 100s of volts Hard to break Not attractive

Electronic loads Kikusui HP / Agilent Chroma Home brew Recommend using Constant Resistance mode with boost

converters Constant Current mode will have problems with start-up on

boost converters.

Internal load steppers usually too slow to properly testtransient response

Next comes the subject of applying a load. You need a load thats easily varied and the power ratingof the load must be greater than the supply under test, of course, otherwise youll overheat the load.

And really one of the best products for a boost converter is an old, old product called a Clarostat 240-C. This is a big power resistor decade box. Its got a 50 watt limit and it goes from 1 ohm to, I believe,10 meg ohms in 1 ohm steps. Now, its not particularly accurate but thats really not important. Butone of the other features is that its useful to hundreds of volts. Its really hard to break. Theyre

particularly unattractive and that keeps the resale price down. You can find these on the auctionWebsites for less than $100 or even less at swap meets.

If you want to be a little more advanced, you could get an electronic load. Theres numerousmanufacturers that make electronic loads. But with a boost converter, youre going to have start-upproblems if you use them in constant current mode. So be sure and get an electronic load that has aconstant resistance mode so you can evaluate the start-up time of your boost converter. Many ofthese electronic loads have an internal load stepper so you can look at transient response to thesupply. I find that the load steppers are always too slow to really properly test transient response soyoull have to build something with a mosfet and 555 timer and another load resistor and get it placedvery close to your power supply under test if you really want to do a good job of load step transientresponse.


28/452

28


F re q u e n c y R e s p o n s e A n a l y ze rc o n n e c t i o n

CIN

Swi t ch

Node

COUT

VIN VOUT (>VIN)

12

COMP

SRC

R2

L

C1

D2 1

C2

VI N

R1

GND

R3

FB

1

2

3

GATE

Rinj

100

Next, comes the subject of connecting a frequency response analyzer. This is so you can evaluateyour loop response for stability. Theres a whole science to this and its something that Ive spent alot of time studying. Frequency response analyzers are unique instruments. There arent a lot of themunder manufacture presently but they permit you to measure the open loop frequency response ofthe power supply loop in a closed loop configuration. And looking at the circuit in figure 28, theres anadded resistor in the feedback path and this serves as a signal injection point. The value of that

resistor is unimportant but Ive drawn it as 100 ohms because thats kind of an industry standardnumber but it could be 10, it could be some other value that youve used on your schematic.

One of the keys of frequency response analysis is that the analyzer itself have high impedance, inputchannelsthese are drawn as items 1 and 2and that theres a floating source out of the frequencyresponse analyzer that is a tracking sine wave source that is applied across the signal injectionresistor. This creates an error signal and over a particular range of frequencies of analysis you canplot a bode plot which will give you information about the gain and phase margin of your circuit. Thissame injection topology can be applied to buck converters. The key is having access to a frequencyresponse analyzer to do this.


29/452

29


FRA M a n u f a c t u r e r s

Venable Industries - 2.2 MHz max freqModel 350 HPIB/GPIB/IEEE488 Big Japan NF Instruments

Model 3120 - Portable Ridley Engineering (AP Instruments)

Model AP200 15MHz maxUSB and parallel port models - Portable

HP 4194A or 4195 discontinued10 Hz to 100 MHz Very Big

Dynamic Signal Analyzers 100kHz maxHP 3562A - HPIB

Stanford Research

A quick review of manufacturers. Venable Industries has been doing this for years. Theres severaldownsides to their equipment in that its particularly large and the frequency response limit is 2.2MHz. They make two modelswell, they make more than that, but the common industry standardmodel is the model 350. It was actually manufactured for them by a company in Japan called NFInstruments. Its an HPIB controlled instrument and software available from Venable Industries drivesthat. One downside is that its an older HPIB interface and you may have difficulty supporting thatinterface on a modern computer. Its also very large and non-portable. Theyve updated that recentlywith a model 3120. Thats either a printer interface or a USB so thats suitable for running with alaptop computer. And its nice and its compact. It still has the 2.2 MHz frequency limit.

Ridley Engineering is a new company with an offering. The instruments actually made by APInstruments. The model number is an AP200. Very nice thing about this is its a 15 MHz max. Now,going to 1 MHz is actually adequate for any of the loop responses youre going to find on commonregulators. But being able to go to 15 MHz permits you to evaluate parasitic of inductors andtransformers and input and output caps. So these are very handy for analyzing AC characteristics.

Theres some older HP products that are available, 4194. These are discontinued. Theyre still veryexpensive on the used market but they are out there and maybe theres one available in yourequipment pool. Some of these HP instruments are designed with 50 ohm input so youll be requiredto use fet voltage mode probes on them and you may have some difficulty getting the floating dryvoltage to impress across the signal injection resistor.

Theres some other instruments called Dynamic Signal Analyzers which can be applied. This maywork but, again, they only go to 100 kHz. But HP 3562, very commonly available on the used marketand quite low cost. One difficulty with it is HPIB, it does not have a floppy drive and its very difficult toget information out except to an HPIB plotter. Theres newer equipment from Stanford Research thatmay be applicable also.


30/453

30


I n du c t o r C o n d u c t i o n m o d e s

Discontinuous conduct ion mode DCM Continuous conduction mode CCM

Observe inductor current with an AC-DCcurrent probe. You need one if you are goingto work on switching mode supplies.

Tektronix & LeCroy produce them Expensive Fragile - Dont drop or exceed ratings Keep them locked up and hidden

Dont lend them out

Next, were going to talk about inductor conduction mode. Most current mode controllers at full loadcurrent are running in continuous conduction mode but at lower load currents they run atdiscontinuous conduction mode. And the loop response is different for these two modes and thatssomething thats covered very well in literature from Middlebrook and Erickson. Touching again oncurrent probes, if you apply a current probe to the inductor input, you can view these waveforms thatare in the next drawings. And current probes themselves are also expensive and if youre going to

work on switching mode supplies, you absolutely need to have access to one. In a voltage domain,youre rather crippled in terms of seeing whats going on. Tektronix and LeCroy are both commonmanufacturers of current probes. The LeCroy ones are specifically made to tie directly to their brandof oscilloscope. The Tektronix ones come in two styles, some that are only compatible with Tektronixscopes and some that are general purpose that have a 50 ohm output cable. All current probes areexpensive, theyre very fragile, and I really cannot highlight enoughdont drop them and dont betempted to exceed their ratings because its a very expensive mistake. And also keep them locked upand hidden. Dont lend them out, they tend to walk off.


31/453

31


D is c o n t i n u o u s c o n d u c t i o nm o d e

VI N

VOUT

DCMI NDUCTORCURRENT

B

A'

SWI TCHNODEVOLTAGE

C'B'

E

BA

DC

I pk

0A

0V

D'

So looking at our first current herethis is sort of a poor graphic I have to apologize for in terms ofthe section E. So from point A to B, the switch node is attached to ground. Inductor current isclimbing, the switch opens between point C and D, and the inductor current decreases all the waydown to zero and this is what signifies discontinuous conduction mode. Now, in point E, this issupposed to represent a sinusoidal ring out around Vin. And a lot of people look at this in the voltagedomain and think, Oh God, theres something wrong with the power supply. I need to do something

to damp out that ringing.And, in fact, theres nothing wrong with that ringing with a possibleexception if its so large that it rings out and goes all the way down to zero volts. Theres some ICsthat may malfunction if thats the case. But generally, that sinusoidal ring out in discontinuousconduction mode is not an issue at all and its just something you should be aware of as you adjustthe loading of your power supply.


32/453

32


C on t i n uo u s c o n d u c t i o n m o d e

C

CCMI pk

I NDUCTORCURRENT

SWI TCHNODEVOLTAGE

0V

VOUT

0A

The next diagram is continuous conduction mode and notice that the inductor current zero line at thebottom of the page is below the lower tips of the inductor current. So thats the important distinctionbetween those two conduction modes.


33/453

33


Ca p a c i t o r C h em i s t r y C h o ic e s

Tantalum Do not use for Cin if hot plugging input Derate voltage by 2X 35V max -> ~20V max in applications Dont exceed ripple current rating Dont exceed vol tage rating Dont reverse polarity

AVX TPS ser ies Many others

Next, lets cover capacitor chemistry choices. Tantalums used to be very popular because you canget such a high value in a small package. Theyve become very expensive due to tantalum shortagesand they do have some limitations that you should be aware of. For example, its really notrecommended using tantalum as the input capacitor in these designs if youre going to hot plug froma powered wall adapter. This can exceed the current rating, the surge current rating of the capacitorand cause failure. Its also recommended, just through experience, to derate the voltage by 2X. So

youve got a 20 volt capacitor, dont use it in an application above 10. Or if youve got a 5 volt rail,always use a 10 volt cap. Tantalums are limited in the available ratings to about 35 volts so 20 voltsis probably the highest application voltage you would use a tantalum in. Its important not to exceedthe ripple current rating with a current probe and a true RMS meter you can see what the actualripple current is in either the input or the output cap. Its important never to exceed the voltage ratingunder transient conditions and dont ever reverse the polarity. Both of these can lead to failure.Theres many manufacturers; the most common in power supplies is the AVX TPS seriesthatstands for tantalum power supplies so its very easy to remember those when youre looking for data.Theres many other manufacturers but thats the one Im most familiar with.


34/453

34



Aluminum electrolyt icAlways parallel w/ at least 1uFd of ceramic Look for low ESR in ti tle and 100kHz data Voltage ratings from 6.3V up to 450VDC Derate voltage to ~ 80% Values from 1 uFd up to many 1000ufd. Lowest cost Least reliable if hot environment

SanyoPanasonic

Many others

Next are aluminum electrolytics. Its important in a switching mode power supply that if youre goingto use aluminum electrolytic that you should always parallel with at least 1 uFd of ceramic. And thishelps bypass the internal inductance. When selecting aluminum electrolytics and viewing datasheets,always look for a brag phrase in the title where theyre saying that it is low ESR and its imperativethat there be data at 100 kHz. If theres only data at 120 Hz then those are really not appropriate forswitching mode power supplies. Aluminum electrolytics are available in voltage ratings from 6.3 to

450 volts. Its probably appropriate to derate them about 80% instead of 2X as with the tantalum. Andyou can find values from 1 uFd to many thousands of uFds but probably the 10 to 100 uFd range iswhat youll find mostly in switching mode supply. These are the lowest cost but theyre also the leastreliable because internally they are wet and they will eventually dry out if used in an extended periodof a hot environment. Sanyo and Panasonic are common manufacturers. Theres many, manyothers.


35/453

35



Ceramic Use X7R, X5R, X5S, never Z5U or Y5V Values up to 100UF at 6.3V Large sizes may crack wi th PCB flexing Max voltage about 50V for high C units (they

tolerate voltages much higher than rating)Taiyo YudenMurataMarcon / United ChemiconAVX

Modern switching mode supplies, the switching frequencies have moved up and this has permittedthe use of ceramic capacitors. Now, with ceramic capacitors theres choices of dielectrics and this ishighlighted in the datasheets. They always have a three letter abbreviation such as X7R or X5R andthose are fine for switching mode power supplies or even X5S. The latter two are attempts at puttingever higher amount of capacitance in a smaller package. Theyve achieved this in dielectrics calledZ5U and Y5V. We dont recommend those two capacitor types; those have bad voltage coefficients

so as you increase the voltage across the cap the capacitance drops and they also have badtemperature characteristics. So really it may be tempting to try those, we recommend you stay awayfrom them and stick to the first three.

You can now get compact values up to 100 uFd at 6.3 in a 1210 size surface mount package. Thereare also larger sizes of capacitors but be aware that some of those crack with PCB flexing. This isrelated to soldering practices as well, but just in general youre better off putting several smallercapacitors in parallel to end up with a total capacitance than trying to get just one huge capacitor inyour circuit. Max voltage ratings are about 50 volts for these high capacitance units. I think in a 12 pinpackage you can get 2.2 uFd at 50 volts is a fairly common package. And an important note is thatwhile these may have a 6.3 volt rating or a 50 volt rating, theyll tolerate surge voltages much, muchhigher than the voltage rating. So that can be real handy if youve got say an automotive applicationwhere youve got to withstand 100 volt surge. My favorite manufacturers of ceramics are Taiyo

Yuden, Murata, and Marcon, and AVX. There are other ones out there but this is a good place tostart.


36/453

36



Poscap (Sanyo)or SPcap (Panasonic) polymer Excellent performance Low ESR High ripple current rating Limited choice of voltage ratings (


37/453

37



Sanyo OSCON Excellent performance Low ESR High ripple current rating Limited choice of voltage ratings (


38/453

38


D io d e s e l e c t i o n

Schottky available up to 100VPIVDerate ~80%Diodes Inc Best selection of packagesCentral SemiconductorOn Semiconductor (formerly Motorola)

Ultra Fast recovery Small signal sil icon 1N4148/1N914

Low cost / low current 1N4007 type far too slow

In terms of diode selections, in most boost converters youre going to be using a Schottky diode.These have very low capacitance, very low forward drop. You should derate their voltage to about80% or more of the rated voltage. So if youve got a 20 volt output converter, you know, use a 30 voltdiode. You can now get Schottkys up to 100 volts. My favorite manufacturer is Diodes Incorporated.Theyve got the best selection of packages. Central Semiconductor puts out some and then the triedand true numbering system from On Semi will cover current ratings from about half an amp past 15

amps. So thatll cover all of the bases that youll encounter in a standard boost converter. If youre inan application where you need beyond 100 volts, you may be forced to use ultra fast recovery diodes.

All these manufacturers offer product selections that are ultra fast recovery. If its a low currentapplication, you can get by with small signal diodes; the common 1N4148, the surface mountequivalent to those. Those are good low cost, low current. The switching speed is appropriate andthose can be used in low power supply. Dont ever get tempted to use standard slow 1N4000 typediodes. Your power supply just wont work correctly.


39/453

39


PCB l a y o u t g u i d e l i n e s (1 )

The function of any dc-dc converter is tocreate an output voltage with low voltagenoise. The point of lowest noise is directlyacross the output capacitor terminals.

The output capacitor may actually be severalcapacitors in parallel

Distribute Vout from the plate of the lastoutput capacitor as close to the + terminal aspossible.

Similarly attach to the ground plane from the terminal.

Finally, were going to cover PCB layout guidelines. A lot of these rules apply to any type of converterwhether its a linear or a switching mode. Linear you tend not to have any layout problems withbecause theres no switching currents in it but many of these things apply to buck converters, too. Sothe most important function of any DC to DC converter is to create an output voltage with low voltagenoise. And I cant emphasize enough that the point of lowest noise is directly across the outputcapacitor terminals.

And its common that the output capacitor may actually be several in parallel but when youredistributing to the Vout plane and to the ground plane, you want to do this from the terminals of thelast output capacitor. And this is very important. Now, this would be true in buck regulators as well.


40/454

40


PCB l a y o u t g u i d e l i n e s (2 )

Keep discontinuous currents OFF the Vcc and grounddistribution p lanes. Switch currents and diodes currents arediscontinuous. Route these currents on copper floods on thesurface of the PCB.

Ground terminal of the input capacitor is routed on the surfaceflood of copper from Cout.

Consider adding an input fil ter inductor of several uH. (Nevermentioned in the data sheet.) Can be stability issues. ReadMiddlebrook / Erickson / Venable

One layer below the switcher should be an image planegrounded at Cout ground point.

If output ripple is too high add an output LC filter.Voltage Feedback comes from the input side of the output

filterBeware of high-Q LC filter on output . Big C w/ ESR.

So the following is a list of rules in terms of guiding this layout. You want to keep the discontinuouscurrents off of the Vcc and ground planes which are, you know, distribution planes lower in the PCboard. The switch currents and diodes currents are the discontinuous currents you need to watch for.So you need to route these currents on copper floods on the surface of the PCB and probably thebest way to do this is you really want all of the components of the boost converter on one side of theboard. It may be tempting due to placement trying to make real estate as small as possible to put the

diode on the backside of the board or something of that sort and I really discourage that. If you canget all of the power components on one side of the board, its very easy to route and follow the firstdistribution guideline which is to distribute from the output capacitor.

Now, as Ive mentioned previously, the ground terminal of the input capacitor is common with aground terminal of the output capacitor. So at the input capacitor, you dont want a connection intothe ground plane even though thats the same net name. And this should be routed on the surface ofthe board of the copper flood. One drawback to that is now since the input capacitor is not at ACground any more but, in fact, influenced by switching currents on a copper flood on the surface of theboard, that now the positive terminal of the input capacitor is noisy. So you may consider putting aninput inductor as an input to the input capacitor. You have to be cautious about that. This can lead toinstability if the inductor value is too large. So theres papers by Middlebrook and Erickson andVenable on input filter inductor stability and how to go about measuring that with a frequency

response analyzer to determine if its an issue.

Moving on, one layer below the switcher should be an image plane thats also grounded at Cout andthis will shield the switching currents from coupling into the ground plane below. If in your applicationyou need a real low output ripple, an output LC filter may be appropriate. And well highlight theseitems later on a schematic.


41/454

41


PCB l a y o u t g u i d e l i n e s : TO26 3

On many integrated boost regulators inTO263 packages the metal tab connects tothe ground net. It is tempting to sinknumerous vias into the ground plane for heatsinking.

Beware:This applies discontinuous switch currents

into the ground plane and makes swi tchernoise filtering very difficu lt.

Attachment into the ground plane should onlybe at Cout.

Route ground as a flood of copper on thesurface.

A very common integrated boost regulator is housed in the TO263 package. A lot of the simpleswitchers are provided in that package. In the center terminal or the big power tab on that package istied to the ground net. And its real tempting to sink a bunch of vias into the ground plane for heatsinking but this violates one of these previous rules of trying to keep discontinuous currents off of theground plane. So just be aware of this. Its appropriate to have a big copper flag for heat sinkingabilities but tying it down into the ground plane is going to increase the noise of the converter and

make emissions control more of an issue.


42/454

42


PCB l a y o u t g u i d e l i n e s : o n v i a s

When routing floods of copper that areconnected the ground and Vout nets, viasmay make inadvertent connections to groundor Vout in the ground and power planes. Mayneed to construct special vias to preventattachment to planes with the same netname.

I mentioned that its probably a good idea to keep all of the power parts on one side of the board andtheres a reason for that. Lets suppose you move the diode to the back of the board. The cathode ofthe diode is also tied to the output capacitor and because the net name is Vout, when there is a viathat comes from the cathode of the diode through the board to the top to the output capacitor,because of the net name, its going to attach to the Vcc output plane on the way through the boardand this is something you want to avoid. You may have to construct special vias if youre forced to

place power components on opposing sides of the board to prevent attachment to planes on the waythrough with a via.


43/454

43


PCB l a y o u t g u i d e l i n e s ,s c h em a t i c v ie w

R2

FB

Swi t chNode

VI N VOUT

VIN

1UF+22UF

D2 1

GNDPLAN

1

2

3

GATE

L

1UF

CIN

LoutputLinput

GND

R1

COUT

Finally, weve got a schematic ofsort of a schematic view of what the PCB layout would be iftheres an output filter and if theres an input filter. So Im very specific with where Im placing groundterminals here on the output side. Theres connection to the VCor excuse me, the ground planeand the positive terminal has connection to the Vout plane. Intermediate to the converter, we haveattachment to the internal output capacitor of the boost converter and up at the input side weve got asmall value ceramic and an input inductor.


44/454

44


Bo o s t R e g u l a t o r De s i g nR e s o u r c e s

Power.national.com online resources

Power WEBENCH Online Designwww.national.com/appinfo/power/webench

General-purpose Boost IC datasheetsLM5000 High Voltage Switch Mode RegulatorLM3478 High Efficiency Low-Side N-Channel

Controller for Switching RegulatorLM2577 SIMPLE SWITCHER Step-Up Voltage

Regulator

Following are some design resources available from National, as well as some recommended ICsyou might look at in the National product line. And that concludes our presentation and Ill turn it overto Wanda.


45/45

45

T h a nk Y o u

Thank you, Alan, and thank you, everyone, for joining us for the seminar, Boost Converter DesignTips, brought to you by National Semiconductor. This concludes todays online seminar. When youclose your seminar window, a survey form will appear. Please fill out and submit the survey form,your answers to this survey will help us in the development of new products as well as futureseminars. Thank you for attending and have a great day!(END OF PRESENTATION)

Documents

0284.Boost Converter Design Tips[1]