
this stage, R
f0
increases abruptly with the current carrying time tin a
nonlinear mode to a level which is comparable to the series-resonant
impedance Z
r
of the converter. The nonlinear ageing stage is an ac-
celerated ageing state which is much shorter than the linear ageing
stage. Thirdly, the fusing and arc forming stage: only one of the fuse
narrow necks (called as the ageing neck) is fusing with forming electric
arcs under the instantaneous large pulse current, and the fuse resistance
fast drops to a low level. Fourthly, the arc extinguishing and cut-out
stage: the electric arc is extinguished as the 100 μs scale resonant cur-
rent dies out, and the fuse resistance recovers to a high impedance level
which corresponds to the cut-out state of fuse.
3. Linear ageing characteristics of the fuse conductor under the
overload pulse current carrying mode
As the carried 10 kHz/240 A sinusoidal current pulse train shown in
Fig. 2(c), the pulse current heating effect in each 10 ms of the carried
pulse train is equivalent to the heating effect of its root-mean-square
(RMS) current DC100A. In practice, the DC100A current with the duty
ratio of 1:2 in each 20 ms is employed to equate and substitute the
10 kHz/240 A sinusoidal current pulse.
3.1. The maximum temperature rise characteristics of the ceramic tube of
the ageing fuse under the DC100A current carrying
Usually, the tested temperature rise in the middle of the fuse
ceramic tube is higher than other places. Under the room temperature,
when the fuse carried the equivalent DC100A current in different
constant time modes (after the current carrying, the fuse cooled down
to the room temperature), the tested maximum temperature rise ΔT
max
in the middle of the fuse ceramic tube had linear increment relations
with the current carrying batch number m,asFig. 4(a) shows. In
Fig. 4(a), the same fuse orderly experienced 5 current carrying modes in
experiment as “DC100A/63s”,“DC100A/60s”,“DC100A/56s”,
“DC100A/53s”, and “DC100A/30s”. Because the fuse conductor of the
same fuse was ageing continuously, the 5 fitted ΔT
max
~mlines had
different “ΔT
max
intercepts”but almost the same straight slope. The
general linear relation of the 5 fitted ΔT
max
~mlines based on the
tested results is as follows.
=+m
T0.3ΔT
max max0 (3)
In (3), m is the current carrying batch number, and ΔT
max0
is the
general ΔT
max
intercept. The straight slope which also reflects the fuse
ageing speed, is about 0.3 °C/batch. The linear relation of ΔT
max
~m
shown in (3) had good recurrence in experiment.
3.2. The linear ageing resistance characteristics of the fuse conductor under
the DC100A/60s current carrying mode
Under the room temperature, the fuse worked under the equivalent
DC100A/60s current carrying mode. After each continuous 60 s for
current carrying, the fuse cooled down to the room temperature. This
carrying mode is equivalent to the effect of T
D
= 120 s (shown in
Table 1) with duty ratio of 1:2. Define the static resistance of fuse under
room temperature before each batch of current carrying as R
f0
, and
define the dynamic resistance of fuse when the fuse ceramic tube
temperature reaches ΔT
max
as R
f.
The fitted linear lines of R
f0
~mand
R
f
~mbased on the tested datum are shown in Fig. 4(b) and (c). The
linear expressions of R
f0
~mand R
f
~mare as follows.
⎧
⎨
⎩=+ = =
=+ = =
RkmRk mbatchR
RkmRk mbatchR
, 0.01 Ω/ , 6.1~6.2 mΩ
, 0.1 Ω/ , 13~15 mΩ
f
f
01 011 01
2022 02 (4)
The experimental results reveal when R
f0
reached ~7 mΩor R
f
reached 20–22 mΩ, the linear ageing stage was over and the nonlinear
ageing stage started. The linear ageing slopes k
1
and k
2
shown in (4)
had good recurrence in experiment.
Generally speaking, the linear ageing stage of the Sielins DC1600V/
32A fuse exists a long time under the overload pulse current carrying
mode, which just corresponds to the total current carrying lifetime of
the fuse. Define the initial static fuse resistance under room tempera-
ture as R
f01
. In the linear ageing stage, R
f0
linearly increases from R
f01
and the stable recurrent variable scale of R
f0
is (R
f01
−R
f0
)/R
f0
≤10%.
As a result, the expended time for the increment of 10%R
f0
is the
normal prediction lifetime of the fuse under the overload pulse current
carrying mode.
=lifetime R t k10% /
fD01 (5)
4. Nonlinear ageing characteristics of the fuse conductor under
the overload pulse current carrying mode
4.1. Nonlinear ageing characteristics
At the last of the linear ageing stage of Sielins DC1600V/32A fuse,
once (R
f0
-R
f01
)/R
f01
≥10%, the fuse fast goes into the nonlinear ageing
stage. Firstly, R
f0
and R
f
respectively fast increase from 7 mΩand
20 mΩto hundreds of mΩor even Ωrange. After the fuse temperature
recovers to the room temperature, R
f0
and R
f
can not fall back.
Secondly, the transition between the linear and nonlinear stages is
usually completed through only one or a few current carrying batches.
Thirdly, though the fuse resistance is far away from its normal level, no
electric arcs are formed at the ageing narrow neck of the fuse con-
ductor. Pulse currents still flow through the series-resonant converter
circuit shown in Fig. 1(b), but the series-resonant characteristics of the
protected converter are changed due to the remarkable increment of R
f
.
The changes of the series-resonant characteristics can cause an un-
anticipated converter circuit failure.
4.2. The failure mechanism of the series-resonant converter caused by the
nonlinear ageing fuse resistance
As the typical series-resonant H bridge of converter shown in
Fig. 1(b), its soft-switching work mode is as follows. Firstly, the switch
pair S1-S3 is triggered simultaneously by the square voltage pulse
T_pulse1 with pulse width of T
r
/2, and the generated series-resonant
current I
r
(the positive half-period) in the converter circuit is as “a”
shown in Fig. 5. When t = T
r
/2, I
r
goes to 0, and S1-S3 turns offsi-
multaneously. During the stage T
r
/2 < t ≤T
r
, the negative half-period
of I
r
shown as “b”is generated by the freewheel diodes D1 and D3
which respectively have anti-parallel relations with S1 and S3. When
t=T
r
, the continuing I
r
goes to 0 and the first resonant period is over.
After that, S1–S4 and D1–D4 all turn off. During the stage
T
r
+ΔT
r
<t≤3T
r
/2 + ΔT
r
, the switch pair S2-S4 is triggered si-
multaneously by the square voltage pulse T_pulse2 with pulse width of
T
r
/2, and the generated series-resonant current I
r
(the negative half-
period) in the converter circuit is as “c”shown in Fig. 5. When t = 3T
r
/
2+ΔT
r
,I
r
goes to 0, and S2-S4 turns offsimultaneously. During the
stage 3T
r
/2 + ΔT
r
<t≤2T
r
+ΔT
r
, the positive half-period of I
r
shown as “d”is generated by the freewheel diodes D2 and D4 which
respectively have anti-parallel relations with S2 and S4. When
t=2T
r
+ΔT
r
, the continuing I
r
goes to 0 and the second resonant
period is over. After that, S1–S4 and D1–D4 all turn off. As aforemen-
tioned, the two triggering pulses T_pulse1 and T_pulse2 recur alter-
nately, with the interval of T
r
/2 + ΔT
r
.
In the series-resonant converter circuit shown in Fig. 1(b), the re-
sonant impedance Z
r
is as Z
r
=(L
r
/C
r
)
0.5
. As to the typical R-L-C circuit,
the resonant current I
r
contains some important characteristic para-
meters, such as the resonant frequency f
r
, resonant period T
r
, and time
constant τ. Under the underdamping state(R
f
<2Z
r
), critical damping
state (R
f
=2Z
r
) and the overdamping state (R
f
>2Z
r
), the theoretical
I
r
=I
r
(R
f
,t) is calculated as
Y. Zhang, et al. Electrical Power and Energy Systems 117 (2020) 105604
5