QPI-8 - Vicor

QPI-8
 
QUIETPOWER
AdvancedTCA™ Hot-Swap with Active EMI Filter
Description:
Features:
The QPI-8 incorporates a total hot swap function with an EMI
filter for DC/DC converter applications. The product aligns
with the AdvancedTCA™ PICMG3.0 requirements for hot
insertion and board level conducted noise limitations. The
EMI filter provides active conducted common-mode (CM) and
differential-mode (DM) noise attenuation from 150 kHz to 30
MHz. The QPI-8 is designed for use on a 48 or 60 volt DC bus
(36-76VDC). The in-rush current limit (12A) and circuit
breaker (6A) are designed to satisfy the 200W per board
PICMG3.0 limit up to 70C (TA) before de-rating.













The under and over voltage thresholds can be trimmed
separately via the UVEN and OV inputs using external series
resistors. The Power Good active low output can be used to
disable a converter until its input bulk capacitors are fully
charged to the bus voltage. The QPI-8’s active filter greatly
reduces the amount of board space typically required for EMI
filtering.
The QPI-8 is available as a lidded
or an open-frame SiP (System-inPackage) with LGA mounting.
Evaluation boards are available to
allow for quick in-circuit testing of
the QPI-8LZ within an existing
system design.
Typical Application:
40 dB CM attenuation at 250 KHz (50Ω)
70 dB DM attenuation at 250 KHz (50Ω)
80 Vdc (max input)
100 Vdc surge 100 ms
1,500 Vdc hipot hold off to shield plane
6 A delayed circuit breaker rating, 12 A current limited
25.3 x 25.3 x 5.2 mm Lidded SiP (System-in-Package)
24.9 x 24.9 x 4.4 mm Open-frame SiP
Low profile LGA package
-40° to +125°C Ambient temperature (see Figure 19)
Efficiency >99%
Connect in series for higher attenuation
TÜV Certified
Applications






ATCA blades
Telecom base stations
IBA & distributed power
Network switches and routers
Optical line-cards
TD-SCMA wireless infrastructure
Figure 1 - QPI-8LZ
2
(~1 in area)
Figure 2 – Typical QPI-8 application schematic with a Vicor Mini brick converter.
(1)
Note 1: CSW capacitor, referenced in all schematics, is a 47uF electrolytic; United Chemi-Con EMVE101ARA470MKE0S or equivalent.
CY1 to CY4, referenced in all schematics, are 4.7nF hi-voltage safety capacitors; Vishay VY1472M63Y5UQ63V0 or equivalent.
CIN is the converter manufacturer’s recommended value for input capacitor. OPTO ISOL has a CTR of 200%, CEL PS2561 or equivalent.
The value of RPG is dependent on the selected Opto-isolator and QPI-8’s PWRGD current limit.
Picor Corporation · picorpower.com
QPI-8
Rev 1.5, Page 1 of 17
QPI-8
 
QUIETPOWER
Absolute Maximum Ratings – Exceeding these parameters may result in permanent damage to the product.
Input Voltage, BUS+ to BUS-, continuous
Input Voltage, BUS+ to BUS-, 100ms transient
BUS+/ BUS- to Shield pads, hi-pot
Input to output current, continuous @ 70°C TA
Input to output current, Pulsed @ 25°C TA
Power Good Sink Current; Power Good asserted low
(2)
Power dissipation, @ 70°C TA, 5.5 A
Operating temperature - TA
(2)
Thermal resistance - RJ-A, using PCB layout in Figure 31
(2)
Thermal resistance - RJ-PCB
Storage temperature, JEDEC Standard J-STD-033B
Reflow temperature, 20 s exposure
ESD, Human body model (HBM)
-80 to 80 Vdc
-100 to 100 Vdc
-1500 to 1500 Vdc
5.5 Adc
12 Adc
5mA
1.6 W
-40 to 125 °C
20 °C/W
8 °C/W
-55 to 125 °C
245 °C
-2000 to 2000 V
Electrical Characteristics – Parameter limits apply over the operating temp. range, unless otherwise noted.
Parameter
BUS+ to BUS- input range
BUS+ to QPI+ voltage drop
BUS- to QPI- voltage drop
Common mode attenuation
Differential mode attenuation
Input bias current at 80 V
Load Current Prior to PWRGD
Under-voltage Threshold (UV)
Under-voltage Hysteresis
Over-voltage Threshold (OV)
Over-voltage Hysteresis
Power Good Low Voltage
Power Good Leakage Current
Notes
(2)
Measured at 5.5 A, 70°C ambient temperature
(2)
Measured at 5.5 A, 70°C ambient temperature
(2)
Measured at 5.5 A, 70°C ambient temperature
VBUS = 48 V, Frequency = 250 KHz, line impedance = 50Ω
VBUS = 48 V, Frequency = 250 KHz, line impedance = 50Ω
Input current from BUS+ to BUSMax load current before inhibiting Hot-Swap start-up.
Controller Disabled to Enabled, no external trim-up.
Controller Enabled to Disabled, no external trim-up.
Controller Enabled to Disabled, no external trim-up.
Controller Disabled to Enabled, no external trim-up.
IPWRGD = 1mA
VPWRGD = 90V
Min
UV
Typ
Max
80
100
185
40
70
15
25
34
UV - 2
76
OV - 4
0.2
0.6
1
Units
Vdc
mVdc
mVdc
dB
dB
mA
mA
V
V
V
V
V
µA
Note 2: See Figure 19 for the current de-rating curve.
Pad Descriptions
Pad Number
12, 13
Name
BUS+
Description
Positive bus potential
LGA Pattern (Top View)
1, 16
BUS-
Negative bus potential
7, 8
QPI+
Positive input to the converter
5, 6
QPI-
Negative input to the converter
4
Shield
Shield connects to the system chassis or to a safety ground.
10
PWRGD
Open drain output that asserts low when power in not good.
11
OV
High-side of Over-voltage resistor divider.
14
UVEN
2, 3, 15
SW
High-side of Under-voltage resistor divider.
Hot-Swap controlled negative rail of EMI filter.
9
NC
No internal connection, not used.
Ordering Information
Part Number
(3)
QPI-8LZ
QPI-8LZ-01
Description
QPI-8 LGA Package, RoHS Compliant
QPI-8 LGA Package, RoHS Compliant, Open Frame Package
Note 3: QPI-8LZ is a non-hermetically sealed package. Please read the “Post Solder Cleaning” section on page 16.
Picor Corporation · picorpower.com
QPI-8
Rev 1.5, Page 2 of 17
QPI-8
 
QUIETPOWER
Applications Information
EMI Sources
pass the EN55022 class B limits.
Many of the components in today’s power conversion
modules are sources of high-frequency EMI noise generation.
Diodes, high-frequency switching devices, transformers and
inductors, and circuit layouts passing high dv/dt or di/dt
signals are all potential sources of EMI.
EMI is propagated either by radiated or conductive means.
Radiated EMI can be sourced from these components as well
as by circuit loops that act like antennas and broadcast the
noise signals to neighboring circuit paths. This also means
that these loops can act as receivers of a broadcasted signal.
This radiated EMI noise can be reduced by proper circuit
layout and by shielding potential sources of EMI transmission.
There are two basic forms of conducted EMI that typically
need to be filtered; namely common-mode (CM) and
differential-mode (DM) EMI. Differential-mode resides in the
normal power loop of a power source and its load; where the
signal travels from the source to the load and then returns to
the source. Common-mode is a signal that travels through
both leads of the source and is returned to earth via parasitic
pathways, either capacitively or inductively coupled.
The other component of the passive filter is the differential
LC network. Again, the inductor is chosen such that it will
present a high-impedance in the differential EMI loop at the
EMI’s base frequency. The differential capacitor will then
shunt the EMI back to its source. The QPI-8 was specifically
designed to work with conventional switching frequency
converters like Vicor’s Brick products; Micro, Mini and Maxi
modules; as well as converters from various vendors.
Active EMI Filtering
PICOR’s QPI-8 active EMI filter uses the same basic principles
for filtering as the passive approach, but its active commonmode filter can perform as well as a passive filter, when
filtering lower frequencies, in much less board area.
Figure 8 and Figure 9 are the resulting EMI plots, after
filtering by the QPI-8, of the total noise, both common and
differential mode, of a Vicor Brick converter. These
converters are mounted on a QPI-8 evaluation board and
tested under various loads. The red and blue traces represent
the positive and negative branches of total noise, as
measured using an industry standard LISN setup, as is shown
in Figures 6 and 7.
Figure 3 – Simplified Active EMI filter circuit.
Differential-mode EMI is typically larger in magnitude than
common-mode, since common-mode is produced by the
physical imbalances in the differential loop path. Reducing
differential EMI will cause a reduction in common-mode EMI.
Passive EMI Filtering
The basic premise of filtering EMI is to insert a highimpedance, at the EMI’s base frequency, in both the
differential and common-mode paths as it returns to the
power source.
Passive filters use common-mode chokes and “Y” capacitors
to filter out common-mode EMI. These chokes are designed
to present a high-impedance at the EMI frequency in series
with the return path, and a low impedance path to the earth
signal via the “Y” caps. This network will force the EMI signals
to re-circulate within a confined area and not to propagate to
the outside world. Often two common-mode networks are
required to filter EMI within the frequency span required to
Picor Corporation · picorpower.com
QPI-8
Typically, the lower the frequency the greater the needed
inductance would be to properly filter the EMI signal. This
means either a larger core or a greater number of turns on a
smaller core. A larger core requires more board space, where
a smaller core with more turns has a greater amount of
unwanted parasitics that can affect the filters ability to
attenuate EMI signals.
Figure 3 is a simplified schematic of the QPI-8’s active and
passive circuitry used for EMI filtering. The QPI-8’s active
filter uses a small high-frequency common-mode transformer
to filter the higher frequencies and adds a sensing element to
it so that the lower frequency common mode signal can be
sensed and a correction signal can be generated and inserted
into the shield connection. By this means, the QPI-8 is
capable of providing EMI filtering of converters in far less
space than standard passive filters and can provide filtering
over the entire EN55022 class B range.
Rev 1.5, Page 3 of 17
QPI-8
 
QUIETPOWER
EMI Management
favor one topology versus another. The EMI generated by
the “base-plate” configuration is much greater than that
generated by the “open-frame”. Selecting the right topology
will greatly reduce the amount of EMI signal that needs to be
filtered.
The more effectively EMI is managed at the source, namely
the power converter, the less EMI attenuation the filter will
have to do. The addition of “Y” capacitors to the input and
output power nodes of the converter will help to limit the
amount of EMI that will propagate to the input source.
Figure 2 shows the base-plate topology of re-circulating “Y”
caps. Here, CY1 to CY4 are connected to each power node of
the dc-dc converter, and then are commoned together on a
copper shield plane created under the converter. The
addition of the copper shield plane helps in the containment
of the radiated EMI, converting it back to conducted EMI and
shunting it back to its source.
The RY resistor, connected between the shield plane and the
QPI’s shield pin, provides an impedance that makes the QPI’s
common mode noise cancelation signal more effective at
removing the common mode noise that would normally
return to the shield/earth connection. It is important when
laying out the QPI that the RY resistor connects to the QPI’s
shield pin before making the connection to earth ground.
Figure 4 – An unfiltered converter’s response to “open-frame”
(light blue) and “base-plate” (purple) EMI configurations.
There are two basic topologies for the connection of the recirculating “Y” capacitors, referred to as “open-frame” and
“base-plate”. Figure 4 illustrates how a converter can
In Figure 5, the open-frame topology is shown where the “Y”
capacitors (CY1 and CY2) re-circulate the EMI signals between
the positive input and output, and the negative input and
output nodes of the power conversion stage.
Figure 5 - Typical 'open-frame" application using a converter with “low” enable.
Picor Corporation · picorpower.com
QPI-8
Rev 1.5, Page 4 of 17
 
QPI-8
QUIETPOWER
Attenuation Test Setups:
Figure 6 - Open-frame EMI test setup using the QPI-8-EVAL1 carrier board with 48V converter.
Figure 7 - Base-plate EMI test setup using the QPI-8-EVAL1 carrier board with 48V converter.
In Figures 6 and 7, C1 is the required 47uF capacitor (United
Chemi-Con EMVE101ARA470MKE0S or equivalent), C2 is a
converter input cap (value dependant on converter), and CY
caps are 4.7nF ceramic (Murata GRM31BR73A472KW01L or
equivalent).
Attenuation Plots:
Figure 8 - V48B12C250BN using base-plate “Y” capacitors
with a 162W load.
Picor Corporation · picorpower.com
Figure 9 – V48B3V3C150BN using base-plate “Y” capacitors
with a 141W load.
QPI-8
Rev 1.5, Page 5 of 17
 
QPI-8
QUIETPOWER
Hot-Swap Function
charges the bulk capacitor by the third pulse of current, also
signaling that power is once again good. During the short
circuit the input current is limited to 12A peaks, with an “on”
duty cycle of about 1%.
Figure 10 - QPI-8 Block Diagram
The hot-swap of the QPI-8 sits before the active EMI filter and
™
is designed to align with the AdvancedTCA PICMG3.0
guidelines of 48V, 200W boards. Its 6A circuit breaker current
rating allows the QPI-8 to provide 200W of power down to
the minimum bus voltage of 36V. Its 12A maximum current
limit allows for fast charging of a converter’s input bulk
storage capacitance. Its ability to “retry” after a circuit
breaker fault time-out enables the QPI-8 to charge large
amounts of bulk capacitance through multiple 12A current
pulses before signaling that power is “good” via the PWRGD
pin. This signal should be used to enable the QPI-8’s load
converter.
The QPI-8 has a default under-voltage limit of 34V to turn on,
and 32V to turn off. The over-voltage limit is 76V to turn off,
72V to turn on. The high-side of each resistor string is
brought out to allow the user to trim up one or both of the
voltage limits, or the pins can be connected directly to BUS+
for the default under and over voltage limits. The equations
in Figure 12 provide the new trip point voltages for a given
added series resistor, RUVEN and/or ROV, as is shown in Figure
13.
Figure 12 – Resulting UVEN and OV voltage equations based
on selected series resistors.
Having access to the high side of the internal resistor divider
string enables the user to filter one or both nodes, allowing
the QPI-8 to pass various transient requirements of
PICMG3.0.
Figure 11 - 2 second short applied across QPI-8's outputs.
The waveforms shown in Figure 11 are of a running QPI-8
that has a short applied across its output for a 2 second
duration. The PWRGD signal (blue, CH1) is initially high, the
VSW voltage (light blue, CH2) is low with respect to BUS-, and
the input current (green, CH4) is at a minimum load.
With a short applied, PWRGD drops to a low value and the
voltage on VSW returns to the BUS+ value. The QPI-8 detects
a current fault and begins a cycle of trying to charge the bulk
capacitors, then waiting. Once the short is lifted, the QPI-8
Picor Corporation · picorpower.com
QPI-8
Figure 13 - UVEN filtering for Zero-Volt BUS transients.
The normal response of the QPI-8 to a BUS supply transient
which drops below the UV limit is to assert the PWRGD pin
low and shut off its pass FET. When the UV requirements are
Rev 1.5, Page 6 of 17
 
once again satisfied the QPI-8 then will require ~16ms to
restart and turn the pass FET back on, charging the bulk
capacitance. The PWRGD signal will be low during the whole
time of the transient, the 16ms delay and the time required
to restore the BUS voltage across the bulk capacitors. See
Figure 18 for a detailed description of the under and over
voltage fault event timings.
QPI-8
QUIETPOWER
up to the bus supply voltage is dependent on the value of RC
and CHOLD-UP. If the bus supply were to be removed, the
energy stored in CHOLD-UP will be released through the
diode D and the QHOLD-UP FET, which will conduct either by
being actively turned on by PWRGD or through its body
diode. Once the bus voltage is restored, CHOLD-UP will start
to re-charge back to the bus supply voltage.
The ATCA guidelines have a 5ms, zero-volt bus transient
requirement that must be met. The circuit in Figure 13 shows
that with the addition of a series diode (D), resistor (RUVEN)
and a capacitor (CE), the UVEN pin can be filtered from
reacting immediately to a transient loss of power, thereby
maintaining a high PWRGD status and allowing the converter
to run off of its bulk capacitors.
There is a practical limit to the amount of bulk capacitance
that can be used as an energy source during a low voltage bus
transient event. Upon recovery from the transient, the QPI-8
has to supply the current to charge the capacitors back to the
BUS+ voltage value and provide the current for the converter
in less time that the fault delay, about 1.2ms. If the caps
cannot be completely charged prior to the fault time-out,
then the QPI-8 will shut off its pass FET, assert the PWRGD
pin low, and then retry about 95ms later.
For example: A system with a bus voltage of 48V and a 4A bus
current has a drop in the bus voltage to 40V and then quickly
recovers back to 48V. While at 40V the bus current increases
to 4.8A to maintain the output load. To restore the bulk
capacitor back to the 48V bus voltage the amount of bus
current available, limited by the QPI-8’s current limit of 12A,
will be the 12A minus the average converter load current
(4.4A). Without regard to tolerances, the equation to
calculate the maximum amount of capacitance that can be
charged within the 12A pulse period (1.2ms) is:
The amount of capacitance required can be determined using
the following equation:
Where: E = Hold-up energy
VBUS = BUS supply voltage
VUVLO = converter’s UVLO limit
The 15V zener and 100V diodes are used to protect the FET’s
gate to source maximum voltage limit and to protect the QPI8’s PWRGD pin. At start-up, the voltage on QPI- is equal to
BUS+ with respect to BUS-. If the 100V diode were not
present, then the PWRGD pin would be pulled up to the BUS+
voltage minus the diode drop of zener. At start-up, the
PWRGD pin is in an active low state and will get damaged
being forced to the BUS+ voltage. The zener diode protects
against the FET’s maximum gate to source voltage being
exceeded.
Where I = 4.4A
Δt = 1.2ms
ΔV = 8V
The value of the bulk capacitor is 1140µF; to maintain some
margin for capacitor tolerance a 1000uF capacitor or smaller
should be used.
Another option is to use a current limiting charge circuit to
restore the capacitors. In Figure 14, the bulk storage
capacitor (CHOLD-UP) is charged through RC after PWRGD
has been released from its active low state and QHOLD-UP is
allowed to turn on. The time it takes for CHOLD-UP to charge
Picor Corporation · picorpower.com
Figure 14 - Powergood controlled, auxiliary bulk storage
capacitor charging circuit.
QPI-8
The alternative to adding large amounts of bulk capacitors to
the converter’s input is to create a voltage supply greater
than that of the bus supply. This takes advantage of the
increased stored energy of a capacitor at higher voltages.
Rev 1.5, Page 7 of 17
QPI-8
 
VBOOST
QPI-
QPI+
QPI-
QPI+
QUIETPOWER
Figure 15 - Simplified Hi-Voltage Boost Circuit for Charging Bulk Storage Capacitors.
The circuit in Figure 15 shows a simplified version of a
PWRGD enabled boost circuit. The bulk storage capacitance
(CBULK) is initially charged to BUS+, along with the CIN
capacitor, through the body diode of the PFET. Once these
capacitors are fully charged, PWRGD enables the boost
converter to slowly charge the bulk capacitance to the
desired hi-voltage. When a low QPI+ voltage is detected, the
BUS VOLTAGE MONITOR circuit enables the PFET to discharge
some of its energy into the load converter’s CIN capacitor and
to the load. When the QPI+ rail reaches a pre-determined
limit, it shuts off the PFET and the rail starts to drop again.
This circuit acts much like a ripple regulator and will maintain
the output load until either the BUS voltage is restored or the
energy of the bulk capacitor is depleted.
limit is set by the current limit level and the duty cycle of the
circuit breaker timer, which is about 99% off time. An
excessive load will discharge the bulk capacitors during the
off time to a level where the on time charging current will
never charge the capacitor to the full bus voltage. Only after
Powergood has been established should the load be enabled.
The QPI-8 has a current sensing circuit that shuts off the
internal pass FET if a high di/dt is sensed across its internal
inductor. To avoid triggering this function, the PFET should be
turned on in its linear region. Typically, a zener diode can be
used to regulate the VG-S of the PFET and the amount of
energy discharged from the bulk capacitance. The waveforms
in Figure 16 demonstrate a zero-volt bus transient and the
response of the hi-voltage boost circuit. The storage caps
provide energy to sustain the load, while the bus is at zero for
5ms. The input current (green, CH4) drops to zero while the
load current (violet, CH3) remains constant. The voltage
across the bulk storage caps (light blue, CH2) slowly decays
while input voltage to the converter (blue, CH1) is regulated.
Figure 16 - 5ms, 0V transient with 200W Load. CH1 and CH2
are offset by 50V DC.
It is critical to keep the load current on the converter’s input
capacitor to less than 25mA during the initial start-up. This
Picor Corporation · picorpower.com
QPI-8
Rev 1.5, Page 8 of 17
QPI-8
 
QPI-8 Timing Waveforms
The waveforms in Figure 17 are the typical start-up
waveforms that would be seen in a system where two current
pulses are required to completely charge the bulk
capacitance to the bus voltage. The GATE waveform is an
internal signal and cannot be measured externally; it is shown
only for reference.
Once VIN is established and the UV and OV thresholds are
satisfied, the internal FET’s gate will start to get driven high
after a 15ms delay. This delay allows for insertion debouncing. As the FET turns “on” and allows the passing of
current the magnitude of the current is monitored. Once it
reaches the 6A circuit breaker threshold an internal fault
timer is started and the current is limited to a 12A maximum
value. If the current does not drop to a value below the 6A
circuit breaker limit before the fault timer ends then the gate
is shut “off” and remains off for about 95ms. After this time
the gate is again turned on and current is allowed to pass.
Once the current is below the circuit breaker limit, the FET is
fully turned “on” to its minimum rds value and the Powergood
pin is released (open drain).
QUIETPOWER
The change in voltage across VSW is dependent upon the
value of the bulk capacitance that the QPI-8 is charging. The
circuit breaker fault time is 300us when the voltage on VSW is
greater than 40V, referenced to BUS-; or if there is no voltage
change across VSW during the 12A current pulse, indicating a
short across the bulk capacitor. Once below the 40V limit,
and a decreasing voltage across VSW, the fault time is
increased to 1.2ms to charge the bulk capacitors faster.
The timing waveforms for an OV (Over-voltage) and
(Under-voltage) events are shown in Figure 18. After
initial start-up delay, the internal FET’s gate is turned
Once its gate-to-source voltage reaches 8V it releases
PWRGD pin and allows it to get pulled up to VIN.
In the event of an OV fault, the QPI-8 will turn off its internal
FET and it will remain off until the OV fault is cleared. Once
cleared, the gate immediately starts to turn back on. With an
UV fault, the FET is again turned off, but the gate is not
turned back on until 15ms after the UV fault has been
cleared.
Figure 17 - QPI-8 current fault timing diagram.
Picor Corporation · picorpower.com
UV
the
on.
the
QPI-8
Rev 1.5, Page 9 of 17
QPI-8
 
QUIETPOWER
Figure 18 – QPI-8 UV and OV fault timing diagram.
6
1.8
5
1.5
4
1.2
3
0.9
2
0.6
1
0.3
0
Maximum Power Dissipation (Watts)
Maximum Current (Amps)
Current De-Rating: mounted to QPI-8-EVAL1 evaluation board.
0.0
-40
-15
QPI-8 Current
10
35
QPI-8 Power Dissipation
60
85
Ambient Temperature (°C)
110
Figure 19 - Current de-rating and power dissipation over ambient temperature range.
The de-rating curve in Figure 19 is based on the maximum
allowable internal component temperature and the 5.5A
maximum rating of the QPI-8. The power dissipation curve is
based on the current squared multiplied by the internal
resistance between the inputs and outputs of the filter. The
Picor Corporation · picorpower.com
QPI-8
internal resistance value is temperature compensated for the
power dissipation curve. The left axis (current) is in amps for
the solid trace, the right axis (power) is in watts for the
dashed trace.
Rev 1.5, Page 10 of 17
QPI-8
 
QUIETPOWER
QPI Application Circuits:
Filtering Dual Converters
Figure 20 – The QPI-8 filtering dual supplies, using an open frame topology.
(4)
Filtering Parallel Converters
Figure 21 – Dual QPI-8’s filtering paralleled converters feeding a common load.
(4)
Note 4: In Figures 12 and 13; CIN1 and CIN2, CY1 through CY8, should be the value and voltage rating recommended by the converter’s manufacturer.
Picor Corporation · picorpower.com
QPI-8
Rev 1.5, Page 11 of 17
 
The shield plane under the two converters in Figure 20 should
be one contiguous plane under both. The circuit in Figure 20
is capable of filtering more converters than shown, up to the
maximum current limit of the QPI-8. In Figure 21, a separate
shield plane is required for each converter along with a
QPI-8
QUIETPOWER
separate RY resistor.
The QPI-8 is not designed to be used in parallel with another
QPI-8 to achieve a higher current rating, but it can be used
multiple times within a system design.
Output to Chassis Connection Using the QPI-8
Figure 22 – Connecting the converter’s output ground to chassis through an inductor.
(4)
Figure 23 -Total Noise EMI Plots of Vicor’s V48B12C250BN, connected as shown in Figure 22
The direct connection of the converter’s output to the
earth/chassis will degrade the EMI attenuation performance of
the QPI-8. Picor recommends that the connection to the earth
be made through a series inductor, rated to the maximum
output current of the converter, as shown in Figure 22. The
EMI plot shown in Figure 23 is of the same converter as in
Picor Corporation · picorpower.com
QPI-8
Figure 8, but uses an inductor in place of RY and has the
converter’s output ground connected to the shield plane.
The connection of the shield plane directly to the chassis/earth
will also degrade EMI attenuation by the QPI-8 and is therefore
not recommended.
Rev 1.5, Page 12 of 17
QPI-8
 
QUIETPOWER
QPI Insertion Loss Measurements
90
80
70
QPI Insertion Loss Equation:
Attenuation [dB]
60
50
40
30
20
10
0
0.01
0.1
QPI-8 CM
1
QPI-8 DM
10
Frequency [MHz]
Figure 24 - Attenuation curves into a 50Ω line impedance, bias from a 48V bus.
QPI Insertion Loss Test Circuits
Figure 25 – Test Set-up to measure Differential Mode EMI currents in Figure 24.
Figure 26 - Test Set-up to measure Common Mode EMI currents in Figure 24.
Picor Corporation · picorpower.com
QPI-8
Rev 1.5, Page 13 of 17
 
QPI-8
QUIETPOWER
Mechanicals
Figure 27 - Lidded Package Dimensions, tolerance of ±0.004”
Figure 28 - Open-frame Package dimensions, tolerance of ±0.004”. Pick and Place from label center.
Picor Corporation · picorpower.com
QPI-8
Rev 1.5, Page 14 of 17
 
QPI-8
QUIETPOWER
Pad and Stencil Definitions:
Figure 29 - Bottom view of open-frame (OF) and lidded (LID) products. (All dimensions are in inches.)
Figure 30 - Recommended receptor and stencil patterns. (All dimensions are in inches.)
Stencil definition is based on a 6mil stencil thickness, 80% of LGA pad area coverage. LGA Package dimensions are for both the OpenFrame and Lidded versions of the QPI-8.
Picor Corporation · picorpower.com
QPI-8
Rev 1.5, Page 15 of 17
QPI-8
 
QUIETPOWER
QPI-8 PCB Layout Recommendations:
Figure 31 - 3D view of paralleling planes underneath the QPI-8.
PCB Layout
When laying out the QPI-8 EMI filter it is important for the
designer to be aware of the radiated EMI field that all
converters emit and to place the QPI-8 outside of this field
area. It is also recommended that the bus lines feeding into
the QPI filter are not routed such that they lie between the
QPI and the converter, or that their copper planes over-lap on
inner layers. This can cause EMI noise to be coupled from
input to output via the parasitic capacitance between the
planes. In Figure 31, the QPI-8 is located ~1.5 inches from the
converter’s input pins to keep the radiated EMI from bypassing the filter and coupling directly to the BUS feeds.
Post Solder Cleaning
Picor’s LZ version QP SIPs are not hermetically sealed and
must not be exposed to liquid, including but not limited to
cleaning solvents, aqueous washing solutions or pressurized
sprays. When soldering, it is recommended that no-clean flux
solder be used, as this will ensure that potentially corrosive
mobile ions will not remain on, around, or under the module
following the soldering process. For applications where the
end product must be cleaned in a liquid solvent, Picor
recommends using the QPI-8LZ-01, open-frame version of the
EMI filter.
QPI-8 Mechanical Data
Datum
Units
QPI-8LZ
QPI-8LZ-01
Notes
FITS
Failure/Billion Hrs.
330
330
FITS based on the BellCore Standard TR-332
MTBF
Weight
MSL
Million Hrs.
grams
3
4.7
3
3
3.4
3
MTBFs based on the BellCore Standard TR-332
°C/20 seconds
245
245
IPC/JEDEC J-STD-020D
Peak reflow
Temperature
QPI-8 Evaluation Boards
Part #
Description:
QPI-8-EVAL1
A QPI-8LZ mounted on a small evaluation board with screw terminal blocks to allow for easy connection into an
existing system.
QPI-8-CB1
A QPI-8LZ mounted on a carrier board designed for use with DOSA compliant footprint dc-dc converters. Screw
terminal blocks to allow for easy connection into an existing system.
Picor Corporation · picorpower.com
QPI-8
Rev 1.5, Page 16 of 17
QPI-8
 
QUIETPOWER
Warranty
Vicor products are guaranteed for two years from date of shipment against defects in material or workmanship when in
normal use and service. This warranty does not extend to products subjected to misuse, accident, or improper
application or maintenance. Vicor shall not be liable for collateral or consequential damage. This warranty is extended to
the original purchaser only.
EXCEPT FOR THE FOREGOING EXPRESS WARRANTY, VICOR MAKES NO WARRANTY, EXPRESS OR LIMITED, INCLUDING,
BUT NOT LIMITED TO, THE WARRANTY OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.
Vicor will repair or replace defective products in accordance with its own best judgment. For service under this
warranty, the buyer must contact Vicor to obtain a Return Material Authorization (RMA) number and shipping
instructions. Products returned without prior authorization will be returned to the buyer. The buyer will pay all charges
incurred in returning the product to the factory. Vicor will pay all reshipment charges if the product was defective within
the terms of this warranty.
Information published by Vicor has been carefully checked and is believed to be accurate; however, no responsibility is
assumed for inaccuracies. Vicor reserves the right to make changes to any products without further notice to improve
reliability, function, or design. Vicor does not assume any liability arising out of the application or use of any product or
circuit; neither does it convey any license under its patent rights nor the rights of others. Vicor general policy does not
recommend the use of its components in life support applications wherein a failure or malfunction may directly threaten
life or injury. Per Vicor Terms and Conditions of Sale, the user of Vicor components in life support applications assumes
all risks of such use and indemnifies Vicor against all damages.
Vicor’s comprehensive line of power solutions includes high density AC-DC and DC-DC modules and
accessory components, fully configurable AC-DC and DC-DC power supplies, and complete custom power
systems.
Information furnished by Vicor is believed to be accurate and reliable. However, no responsibility is assumed by Vicor for
its use. Vicor components are not designed to be used in applications, such as life support systems, wherein a failure or
malfunction could result in injury or death. All sales are subject to Vicor’s Terms and Conditions of Sale, which are
available upon request.
Specifications are subject to change without notice.
Vicor Corporation
25 Frontage Road
Andover, MA 01810
USA
Picor Corporation
51 Industrial Drive
North Smithfield, RI 02896
USA
Customer Service: [email protected]
Technical Support: [email protected]
Tel: 800-735-6200
Fax: 978-475-6715
Picor Corporation · picorpower.com
QPI-8
Rev 1.5, Page 17 of 17