ISL62881_14 Datasheet, PDF (20/35 Pages) Intersil Corporation – Single-Phase PWM Regulator for IMVP-6.5 Mobile CPUs and GPUs

ISL62881, ISL62881B
Figure 18 shows two optional circuits for reduction of the ring
back. R_ip and C_ip form an R-C branch in parallel with R_i, providing
a lower impedance path than R_i at the beginning of i_o change.
R_ip and C_ip do not have any effect at steady state. Through
proper selection of R_ip and C_ip values, i_droop can resemble i_o
rather than i_L, and V_o will not ring back. The recommended value
for R_ip is 100W. C_ip should be determined through tuning the
load transient response waveforms on an actual board. The
recommended range for C_ip is 100pF~2000pF.
C_n is the capacitor used to match the inductor time constant. It
usually takes the parallel of two (or more) capacitors to get the
desired value. Figure 18 shows that two capacitors C_n.1 and C_n.2
are in parallel. Resistor R_n is an optional component to reduce
the V_o ring back. At steady state, C_n.1 + C_n.2 provides the desired
C_n capacitance. At the beginning of i_o change, the effective
capacitance is less because R_n increases the impedance of the
C_n.1 branch. As Figure 15 explains, V_o tends to dip when C_n is too
small, and this effect will reduce the V_o ring back. This effect is
more pronounced when C_n.1 is much larger than C_n.2. It is also
more pronounced when R_n is bigger. However, the presence of
R_n increases the ripple of the V_n signal if C_n.2 is too small. It is
recommended to keep C_n.2 greater than 2200pF. R_n value
usually is a few ohms. C_n.1, C_n.2 and R_n values should be
determined through tuning the load transient response
waveforms on an actual board.
R_ip and C_ip form an R-C branch in parallel with R_i, providing a
lower impedance path than R_i at the beginning of i_o change. R_ip
and C_ip do not have any effect at steady state. Through proper
selection of R_ip and C_ip values, i_droop can resemble i_o rather than
i_L, and V_o will not ring back. The recommended value for R_ip is
100Ω. C_ip should be determined through tuning the load
transient response waveforms on an actual board. The
recommended range for C_ip is 100pF~2000pF. However, it
should be noted that the R_ip -C_ip branch may distort the i_droop
waveform. Instead of being triangular as the real inductor
current, i_droop may have sharp spikes, which may adversely
affect i_droop average value detection and therefore may affect
OCP accuracy. User discretion is advised.
Resistor Current-Sensing Network
PHASE
L
DCR
RSEN
RSUM
Vcn
ISUM+
Cn
Ri
ISUM-
Io
FIGURE 19. RESISTOR CURRENT-SENSING NETWORK
Figure 19 shows the resistor current-sensing network. The
inductor has a series current-sensing resistor R_sen. R_sum and is
connected to the R_sen pad to accurately capture the inductor
current information. The R_sum feeds the sensed information to
capacitor C_n. R_sum and C_n form a a filter for noise attenuation.
Equations 13 through 15 gives V_Cn(s) expressions:
V_Cn(s) = R_sen × I_o(s) × A_Rsen(s)
(EQ. 13)
A_Rsen(s)
=
----------¹------------
1
+
------^s------
ω_sns
(EQ. 14)
ω_Rsen
=
-------------¹-------------
R_sum × C_n
(EQ. 15)
Transfer function A_Rsen(s) always has unity gain at DC.
Current-sensing resistor R_sen value will not have significant
variation over-temperature, so there is no need for the NTC
network.
The recommended values are R_sum = 1kΩ and C_n = 5600pF.
Overcurrent Protection
Referring to Equation 1 and Figures 12, 13 and 19, resistor R_i
sets the droop current I_droop. Table 3 shows the internal OCP
threshold. It is recommended to design I_droop without using the
R_comp resistor.
For example, the OCP threshold is 20µA. We will design I_droop to
be 14µA at full load, so the OCP trip level is 1.43x of the full load
current.
For inductor DCR sensing, Equation 16 gives the DC relationship
of V_cn(s) and I_o(s).
V_Cn
=
⎛
⎜
⎝
-----------^R----ⁿ---^t--^c---ⁿ----^e---^t----------
R_ntcnet + R_sum
×
⎞
D C R⎟
⎠
× I_o
(EQ. 16)
Substitution of Equation 16 into Equation 1 gives:
I_droop
=
-²---
R_i
×
-----------^R----ⁿ---^t--^c---ⁿ----^e---^t----------
R_ntcnet + R_sum
×
DCR
×
I_o
(EQ. 17)
Therefore:
R_i
=
--------²----^R----ⁿ---^t---^c--ⁿ----^e---^t---^×-----^D----^C----^R-----^×-----^I--^o---------
(R_ntcnet + R_sum) × I_droop
(EQ. 18)
Substitution of Equation 8 and application of the OCP condition
in Equation 18 gives:
R_i
=
------²-----^×----^{(------RR--------nn------tt----cc------ss--------++----------RR--------nn------tt----cc------)--+------×----R------R--p------p-----×}-----^D----^C----^R------^×----^I--^o---^m-----^a----^x-----
⎛
⎜
⎝
-⁽--^R----ⁿ---^t--^c---^s----+-----^R----ⁿ---^t--^c---⁾----^×-----^R----^p-
R_ntcs + R_ntc + R_p
+
⎞
R _{s u m⎠⎟}
×
I_droopmax
(EQ. 19)
where I_omax is the full load current, I_droopmax is the corresponding
droop current. For example, given R_sum = 3.65kΩ, R_p = 11kΩ, R_ntcs
= 2.61kΩ, R_ntc = 10kΩ, DCR = 1.1mΩ, I_omax = 14A and
I_droopmax = 14µA, Equation 19 gives R_i = 1.36kΩ.
For resistor sensing, Equation 20 gives the DC relationship of
V_cn(s) and I_o(s).
20
FN6924.3
June 16, 2011