ADDC02808PB

REV. A

–18–

FREQUENCY – Hz

30

20

–20

10

0

–10

Figure 39. Change in ADDC02808PB Differential Emis-

sions vs. ADDC02805SA Emissions with the Same Test

Setup

dB, by which the differential emissions would change if either of

these approaches were followed. Notice that the inductor solu-

tion provides substantial attenuation in the 1 MHz and higher

frequency range, while the larger capacitor solution has a more

uniform effect. These proposed solutions are suggestions; they

have not been tested.

FREQUENCY – Hz

30

20

–20

10

0

–10

Figure 40. Change in ADDC02808PB Differential Emissions

vs. the ADDC02805SA Emissions with External 30

µF

Capacitor

FREQUENCY – Hz

30

20

–20

10

0

–10

Figure 41. Change in ADDC02808PB Differential Emissions

vs. the ADDC02805SA Emissions with External 0.5

µH,

16 A Inductor

The peak in the radiated emissions in the 20 MHz–30 MHz

range of Figure 14 is dominated by common-mode noise. This

common-mode noise emission is changed only slightly between

the ADDC02808PB and the ADDC02805SA converters since it

does not depend on the power level or the differential input

filter. The turns ratio on the transformer has been changed, so

we expect the common-mode emissions might be 2–4 times

larger. This increase could be countered by increasing the 82 nF

common-mode capacitors of Figure 15 correspondingly. Again,

this solution is a suggestion; it has not been tested.

Finally, the pulsed nature of the load means there will be a

substantial ripple in the input current at the fundamental pulse

rate and its harmonics. This ripple can be calculated once the

power is known as a function of time by dividing by the input

voltage. For instance, if the load switches between zero and

200 W (260 W at the input) at 1 kHz with a duty ratio of 50%,

the current drawn by the converter will have a 9.3 A on, 0 A off,

50% duty ratio input current waveform (260 W/28 V = 9.3 A).

This waveform has an average of 4.65 A and a square wave of

plus and minus 4.65 A around this average. This square wave of

current has a fundamental component as well as odd harmonics

(3rd, 5th, 7th, . . .). The peak of the fundamental component is

(4/

π) 4.65. The rms value of this component is .707 times the

peak, or 4.2 A.

With the test setup in Figure 15, given the impedances of the

LISNs and the 100

µF capacitor with its 1 Ω series resistance, a

3.6 V rms waveform would result from this fundamental

component of the input current. The MIL-STD-461D limit

shown in Figure 11 calls for approximately 100 mV at the 1 kHz

frequency. If this limit is to be met, substantial filtering at the

lower frequencies will have to be added to the system.

RELIABILITY CONSIDERATIONS

MTBF (Mean Time Between Failure) is a commonly used

reliability concept that applies to repairable items in which

failed elements are replaced upon failure. The expression for

MTBF is

MTBF = T/r

where

T = total operating time

r = number of failures

In lieu of actual field data, MTBF can be predicted per

MIL-HDBK-217.

MTBF, Failure Rate, and Probability of Failure: A proper

understanding of MTBF begins with its relationship to lambda

(

λ), which is the failure rate. If a constant failure rate is assumed,

then MTBF = 1/

λ, or λ = 1/MTBF. If a power supply has an

MTBF of 1,000,000 hours, this does not mean it will last

1,000,000 hours before it fails. Instead, the MTBF describes

the failure rate. For 1,000,000 hours MTBF, the failure rate

during any hour is 1/1,000,000, or 0.0001%. Thus, a power

supply with an MTBF of 500,000 hours would have twice the

failure rate (0.0002%) of one with 1,000,000 hours.

What users should be interested in is the probability of a power

supply not failing prior to some time t. Given the assumption of

a constant failure rate, this probability is defined as

R(t )