51 Research and Design, Electronic Engineers website - Embedded Systems, MCU, DSP, EDA, Test and Measurement, Components, Communications, Power, Microelectronics, Semiconductors

    Abstract: Introduced in big electric current portable DC/DC the MOSFET iterative flow and the power loss computation.

    Key word: Portable DC/DC; Power loss; Iterative flow; Thermal resistance; Junction temperature

Introduction

It is well known, today’s portable power source designer faces the most serious challenge is provides the power source for now high performance CPU. In recent years, essence CPU needed the supply current doubles every two years, namely the portable essence CPU supply current demand will reach as high as big 40A, but voltage between 0.9V and 1.75V. In fact, although electric current demand in growth with stability, but leaves power source’s space actually not to increase, this reality has achieved has even surpassed in the hot design aspect limit.

Regarding the so big electric current’s power source, usually its division is two or the heterogeneity, namely every one provides 15A to arrive at 25A, for example, turned a 40A power source two 20A power sources. Although may make primary device’s choice to be easier, but extra has not increased on the board or the environment space, regarding reduces the hot design the work basically not to have the big help. This is because when designs the big electric current power source, MOSFET is the component which most difficult to determine. This point is especially remarkable in the notebook, in this kind of environment, the radiator, the ventilator, the heat pipe and other radiation way has usually left CPU. But the power source design must face many disadvantage factors frequently, such as the narrow and small space and the static air current as well as its primary device send out quantity of heat and so on adverse circumstance, moreover, any other ways have not been possible to use for to assist to radiate.

How then to choose MOSFET? The reply is, when chooses MOSFET, must first choose has the enough electric current handling ability, and has the enough radiation channel, finally must from the quantification consider that the essential heat consumes with the guarantee enough radiation way, according to the above, calculates MOSFET the power loss, and determines their operating temperature. This article has analyzed a heterogeneity, the synchronized rectification, in the voltage dropping CPU power source the MOSFET power loss computational method.

Figure 1

1 MOSFET power loss computation

In order to determine whether MOSFET does suit in the specific application, must calculate its power loss, the MOSFET power loss (PL) mainly contains the resistance to lose (PR) and the switching loss (PS) two parts, namely the PL=PR PSMOSFET power loss relies on to a great extent its breakover resistance RDS(on), but, MOSFET RDS(on) concerns with its junction temperature Tj. But Tj relies on the MOSFET tube’s power loss as well as the MOSFET thermal resistance θJA. Because the power loss computation involves to certain interdependence factor, for this reason, may use the result which one kind of iterative process obtains us to need, like Figure 1 the flow shows.

The iterative process date from supposes Tj for each MOSFET, then, calculates each MOSFET respective power loss and the permission ambient temperature. When permits the ambient temperature achieves when or slightly is higher than in the cabinet the maximum temperature design value, this process then ended. This is one reversion design method, because, starts from hypothesis’s Tj to calculate first, must compared to start easy somewhat first from the ambient temperature computation.

Whether to enhance as far as possible this computation obtained ambient temperature? The reply is incorrect. Because, this will request to use more expensive MOSFET inevitably, and lays down the more copper membranes under MOSFET, or the request uses one to be bigger, a faster ventilator to produce the air current and so on, all these are impractical.

    Regarding the switch and synchronized rectification MOSFET, may choose a permission highest tube core junction temperature Tj(hot) to take the iterative process the starting point, the most MOSFET data books have only stipulated 25℃ under biggest RDS(on), but some products also had provided recently 125℃ under maximum value. MOSFET RDS(on) increases along with the temperature markup, the typical temperature coefficient in 0.35%/℃~0.5%/℃, as shown in Figure 2. If is in doubt, may use a more conservative temperature coefficient and MOSFET 25℃ the specification (or 125℃ specification), makes the approximate estimate under designation Tj(hot) by biggest RDS(on), namely

In the formula: RDS(on)SPEC is the MOSFET breakover resistance which the computation uses;

TSPEC is stipulates when RDS(on)SPEC the temperature.

Using RDS(on)hot which calculates may determine the synchronized rectification sum

Switch MOSFET power loss. Therefore, how will discuss to further calculate each MOSFET under the tube core temperature power loss which assigns, as well as completes iterative process the following step, its entire process specification as shown in Figure 1.

1.1 synchronized rectification power loss

Besides the most underloading, synchronized rectification MOSFET leaks, the source voltage to clear and in the closure process can by the after flow diode clamp. Therefore, the synchronized rectification does not have the switching loss nearly, its power loss PL only need consider the resistive loss then. In the worst situation’s loss occurs in the synchronized rectification work when the biggest dutyfactor, is also the input voltage achieves is lowest time. Using synchronized rectification’s RDS(on) and the work dutyfactor, through the ohm’s law may the approximate calculation its power loss, namely

1.2 switch MOSFET power loss

The switch MOSFET resistance loses the PR computation and the synchronized rectification is similar, must use its dutyfactor (, but is different with the former) and RDS(on)hot, namely

The switch MOSFET switching loss calculates is quite difficult, because it relies on many with difficulty quantifications, and does not have the standard factor, these factors simultaneously affect the clear and the shutdown process. Therefore, may first carry on the appraisal with the below sketchy approximate formula to some MOSFET, then carries on the confirmation through the experiment to its performance, namely

In the formula: Crss is the MOSFET reverse transmission electric capacity (the data book

A parameter);

fs is the turn-on frequency;

Igatb is the MOSFET electronics grid driver when MOSFET is in the critical breakover (Vgs located at electronics grid charge pattern flat site) the absorption/source to leave the electric current.

If considered from the cost element, will choose the scope to reduce (the different generation of MOSFET cost difference is very big) to specific some generation of MOSFET, might find one in this generation of component to be able to cause the wattage dissipation smallest component. This component should have the balanced resistance and the switching loss, the use is smaller the resistance which, a quicker component increases to lose will surpass it in switching loss aspect reduction, but uses in a big way (, but RDS(on) is lower) the component increases the switching loss will surpass it reduction which loses regarding the resistance.

If Vin is the change, needs distinguishes compute switch MOSFET under Vin(max) and Vin(min) the power loss. The worst situation will possibly appear under lowest or the highest input voltage. This power loss is sums of the two kind of factor: When Vin(min) achieves the highest resistive diffusion (dutyfactor to be high), as well as when Vin(max) achieves the highest switching loss. A good choice should have approximately the same power loss in the Vin two kind of extreme cases, and maintains in the entire Vin scope the balanced resistance and the switching loss.

If loses in Vin(min) time is higher than obviously, then the resistance loses the leading role. In this kind of situation, may consider that is bigger with electric current a spot MOSFET (or above MOSFET parallel) reduces RDS(on). But if in Vin(max) time loses obviously is higher than, then should consider with electric current small spot MOSFET (, if is multibarreled parallel, or removes M0SFET), with the aim of causing its shutter speed to be quicker a spot. If the resistance and the switching loss have reached balanced, but the total power loss excessively was still high, also has many kinds of means to be possible to solve:

- - the change or redefines the input voltage scope;

- - cuts the turn-on frequency to reduce the switching loss, or selects RDS(on) lower MOSFET;

- - increases the electronics grid drive current, has the possibility to reduce the switching loss;

- - uses a technical change MOSFET, with the aim of simultaneously obtaining the quicker shutter speed, lower RDS(on) and the lower grid resistor.

What needs to point out mistakes, is separated from the condition which some assigns to the MOSFET size to make a finer adjustment not greatly possible, because component’s choice scope is limited. The choice agent is MOSFET must be able to fall in the worst situation power loss by the diffusion.

2 about thermal resistance

According to shown in Figure 1, continues to carry on the iterative process next step, with the aim of seeking for appropriate MOSFET to take the synchronized rectification and switch MOSFET. This step is must calculate around each MOSFET the ambient temperature, under this temperature, the MOSFET junction temperature will achieve our assumed value. Therefore, first needs to determine that each MOSFET ties to the environment thermal resistance θJA.

The thermal resistance estimate will be quite possibly difficult. Sole component on a simple printing plate θJA the reckoning is relatively easy some, but must in a system forecast that the actual power source’s thermal properties are very difficult, because, there has many heat sources to compete for the limited radiation channel. If has many MOSFET to use parallel, its overall thermal resistance’s computational method, with calculates two above shunt resistances the equivalent resistances to be the same.

We may from MOSFET θJA the specification start. Regarding the sole tube core, 8 pin seal’s MOSFET says, θJA usually approaches in 62℃/W. Other type’s seal, somewhat has the radiator fin or the exposed heat conduction piece, its thermal resistance in 40℃/W to 50℃/W (will see Table generally 1 row). May use the following formula to calculate MOSFET the tube core to be opposite in environment temperature rise Tj(rise), namely

Tj(rise)=PL×θJA    (5)

Then, the computation causes the tube core achieves when predetermined Tj(hot) ambient temperature Tambient, namely

Tambient=Tj(hot)-Tj(rise)    (6)

If calculates θJA is lower than cabinet’s most large quantity to decide the ambient temperature, must use following or the many measures:

- - elevates predetermined Tj(hot), but do not surpass the data book stipulation the maximum value;

- - chooses more appropriate MOSFET to reduce its power loss;

- - through increases around the air current or the MOSFET copper membrane reduces θJA.

Redesigns Tambient (to use again fast calculated that table may simplify computational process, passes through only then selects an acceptable design repeatedly many times). But Table 1 is the MOSFET seal typical thermal resistance.

Table 1 MOSFET seal typical thermal resistance

Explained: Because seals reasons and so on physical characteristics, tube core size and installment and binding method, therefore the similar seal type does not use the component, as well as the different manufacturer product’s similar seal’s thermal resistance is also various, for this reason, should consider in the MOSFET data book carefully the hot information.

If calculates Tambient is higher than cabinet’s most large quantity to decide the ambient temperature to be many, may take following or the complete measure:

- - reduces predetermined Tj(hot);

- - reduces uses in the copper membrane area which specially MOSFET radiates;

- - uses more inexpensive MOSFET.

These steps are may elect, because of overheated will not damage under this situation because of MOSFET. However, through these steps, so long as guaranteed that Tambient is higher than cabinet maximum temperature certain allowance, then may reduce the line board area and the cost.

In above computational process biggest erroneous source from θJA. Should in the careful reading data book related θJA specification all annotations. Generally the standard supposes the component to install on the 4.82g/cm2 copper membrane. The copper membrane diffusion majority of powers, the different quantity’s copper membrane θJA difference has been very big. For example, will have 4.82g/cm2 the copper membrane D-Pak seal θJA to achieve 50℃/W. But if only lays down the copper membrane in pin, θJA will be higher than two times (to see Table 1). If will be many MOSFET parallel use, θJA is mainly decided the copper membrane area which installs in them. Two components equivalent θJA may be single component’s half, but must simultaneously double the copper membrane area. That is, increases parallel MOSFET not to increase the copper membrane the words, may cause RDS(on) to halve, but will not change many θJA. Finally, θJA the standard usually does not suppose any other components to the copper membrane radiation area transmission quantity of heat. But in big electric current situation, on power circuit’s each primary device, even is the printing plate line can have the quantity of heat. In order to avoid MOSFET overheated, must estimate in the actual situation carefully θJA, and takes the following measure:

- - the dwell upon designates the MOSFET existing thermal properties aspect the information;

- - inspects whether to have the enough space, with the aim of establishing the more copper membranes, the radiator and other components;

- - determined whether to have the possibility to increase the air current;

- - observes in the hypothesis radiation way, whether to have other remarkable radiation component;

- - estimates from the periphery part either the spatial surplus quantity of heat or the cold quantity.

3 conclusions

The hot management is in the big electric current portable DC/DC design one of difficulty big domains. This kind of difficulty forces us to have the necessity to use the above iterative flow. Although this process can eagerly anticipate the thermal properties designer to approach the optimum design, but must through the experiment determine finally the design cycle is whether precise enough. Should calculate MOSFET the thermal properties, provides the enough diffusion way for them, then examines these computations in the laboratory, like this is helpful in obtains one durable and the safe hot design.