TSM108

9/13

This component value is valid if the above

described characteristics are fixed... but in the

automotive field, the input voltage of the converter

is dependant of the car battery conditions. Also,

the frequency may vary depending on the

temperature, due to the fact that the frequency is

fixed by an external capacitor. Therefore, we must

calculate the inductor value considering the worst

case condition in order to avoid the saturation of

the inductor, which is when the battery voltage is

at it’s highest, and the switching frequency at it’s

lowest. Thanks to the OVLO function integrated in

TSM108, the operation of the DC/DC converter

will be stopped as soon as the voltage exceeds

the OVLO level. Let’s suppose the OVLO pin has

been left open, therefore, the maximum input

voltage of the DC/DC converter will be Vin max. =

32V. Frequency min stands in the range of 75kHz

In this case, D = 6 / 32 = 0.1875, therefore Lmin =

52µH.

If we allow a 25% security margin

Lmin = 68µH

9.2. Capacitor

The capacitor choice will depend mainly on the

accepted voltage ripple on the output

Ripple = DVout / Vout = (1-D) / 8LCF²

Therefore, C = (1-D) / 8LRippleF². If C = 22µF,

then

Ripple

=

0.4%

which

should

be

far

acceptable.

Here again, the worst conditions for the ripple are

set when the input voltage is at the highest (32V)

and the frequency at it's lowest (75kHz).

with C = 22µF, Ripple = 1.2%

9.3. Ratings for the Inductor, Capacitor,

Transistor and Diode

The inductor wire must be rated at the rms current,

and the core should not saturate for peak inductor

current. The capacitor must be selected to limit the

output ripple to the design specifications, to

withstand peak output voltage, and to carry the

required rms current.

The transistor and the diode should be rated for

the maximum input voltage (up to 60V in

automotive applications). The diode recovery time

must be in accordance with the time period and

the maximum authorized switching time of the

power transistor.

A

compromise

between

the

switching

and

conducting performances of the transistor must be

found, because choosing a very low ohmic Mosfet

aiming at the benefit of low conduction losses may

bring much higher switching losses than the

expected benefit.

Losses in the switch are:

Pswitch = Prise + Pfall + Pon

where Prise + Pfall represent the switching losses

and where Pon represents the conduction losses.

Prise + Pfall = Iout x Vin x (Trise + Tfall) x F / 2

Pon = Ron x Iout² x d

where Trise is the switching on time, and Tfall is

the switching off time, and where d is the duty

cycle of the switching profile, which can be

approximated to 1 under full load conditions.

With the two last equations, we can see easily that

what we may gain by choosing a performing low

Rdson P-channel MOSFET (for example) may be

jeopardized by the long on and off switching times

required

when

using

a

large

input

gate

capacitance.

10. Electromagnetic Compatibility

The small schematic hereafter shows how to

reduce the EMC noise when used in an EMC

sensitive environment:

EMC Improvement

The RC components should realize a time

constant corresponding to one tenth of the

switching time constant of the TSM108 (i.e. in our

example, the oscillator frequency is set to 10µs

corresponding to 100kHz, therefore, the RC

couple should realise a time constant close to

1µs).

Choosing the components must privilege a rather

small resistivity (between 10 to 100W). A guess

couple of values for RC in our example would be:

R= 22W, C= 47nF

11. Efficiency Calculations (rough estimation)

The following gives a rough estimation of the

efficiency of a car phone charger, knowing that the

exact calculations depend on a lot of parameters,

as well as on a wide choice

of external

components.

Let’s consider the following characteristics of a

classical car phone charger application:

L1

GD

D1

TSM108

Q1

MOSFET P