IEEE Power Electronics Magazine - March 2017 - 33

20
10
60

40
20

FET Drain Voltage (V)

Laser Current (A)

vdrain(10 V/div)

EPC2016C
BSZ146N10LS5

ishunt
(10 A/div)

vgs(5 V/div)

0
0

vopt(0.2 V/div)

-20

t (ns)
FIG 11 The simulated waveforms comparing the EPC2016C to
the BSZ146N10LS5.

affect the waveform. However, the gate speed FOM is
more than two times higher due to the larger gate resistance. This, combined with the extra package inductance, will result in limited performance versus the
EPC2016C. We can estimate the difference using simulation since both the EPC2016C and the BSZ146N10LS5
have good quality models, and the BSZ146N10LS5 model
includes package inductances.
These high-voltage transistors are best suited for a
capacitive discharge laser driver. The simulation includes
an OSRAM SPL PL90_3 laser diode and 1.1-nF energy
storage capacitor ^C BUS h, with an initial capacitor voltage
of 58 V. The power loop inductance, including the laser
diode but excluding the transistor and its package, is
estimated to be 3 nH based on the EPC9126 lidar demo
board layout. A near ideal gate drive is used, modeled by a
voltage source with a 200-ps transition time and 300-mX
output resistance.
Figure 11 shows the results. It is clear that the pulse
from the EPC2016C is both higher in amplitude and shorter
in duration, both of which are desirable. The drain voltage
indicates that the FET turns fully on in 1 1 ns, resulting in
a laser current pulse resembling the desired half-sine. In
contrast, the current pulse in the MOSFET case shows a
slower rise and reduced peak amplitude. This results from

2 ns/div
FIG 12 The EPC9126 lidar demo board capacitive discharge
waveforms. The board is populated with an EPC2016C eGaN
FET and an OSRAM SPL PL90_1 laser diode. C BUS = 1.1 nF,
charged to 50 V.

the long turn-on time as indicated from the drain voltage
waveform. The drain parasitic resistance and inductance of
the Si MOSFET mean that it does not turn fully on for nearly
the entire current pulse.
Table 2 shows some numerical results for peak laser
diode current amplitude I LD,peak and pulsewidth measured
from the 0.1 : I LD,peak crossings (10% pulsewidth). On the
left-hand side of the table, we consider the case described
previously L stray = 3 nH, and on the right-hand side, we
consider a surface-mount laser diode with reduced stray
inductance L stray = 2 nH. For each case, the ideal values
based on (3) and (4), and the improvement of GaN over Si
is computed. From the table, we can see that, in the first
case, the eGaN FET shows a large improvement over the
Si MOSFET. More interesting is the case where the diode
inductance is reduced. The benefit of GaN over Si grows.
This is due to the fact that, with Si, the performance is
limited by the FET, and even a large reduction of laser
inductance will yield small benefits. In the case of GaN,
the inductance is still limiting the performance; hence,
improved laser diode packaging will yield large benefits.

Measured Performance
The EPC9126 demo board has been designed to operate as
a capacitive discharge laser diode driver and comes with

Table 2. A numerical comparison of lidar driver simulation results
for the EPC2016C (eGaN FET) and BSZ146N10LS5 (Si MOSFET).
L stray = 3 n (excluding FET)

L stray = 2 n (excluding FET)

Ideal

EPc2016c

Bsz146n10ls5

Benefit

Ideal

EPc2016c

Bsz146n10ls5

Benefit

I D,peak [A]

35.1

30.1

22.6

33%

43.0

35.9

23.7

52%

10% PW [ns]

5.34

5.39

7.07

31%

4.36

4.43

6.39

44%

PW: pulsewidth.

March 2017

z IEEE PowEr ElEctronIcs MagazInE

Table of Contents for the Digital Edition of IEEE Power Electronics Magazine - March 2017