This is only a preview of the April 2014 issue of Silicon Chip. You can view 27 of the 96 pages in the full issue, including the advertisments. For full access, purchase the issue for $10.00 or subscribe for access to the latest issues. Items relevant to "40V Switchmode Bench Power Supply, Pt.1":
Items relevant to "USB-To-RS232C Serial Interface":
Items relevant to "A Rubidium Frequency Standard For A Song":
Purchase a printed copy of this issue for $10.00. |
By NICHOLAS VINEN
Compact Hybrid Switchmode
100W Bench Supply, Pt.1
. . . has dual voltage and current metering
This very compact bench supply can deliver 0-40V at up to 5A with
accurate and fast current limiting and has 3.5-digit 7-segment LED
readouts for simultaneous voltage and current display. You can power
it from any 12-24V DC supply such as a PC or laptop power supply
or lead-acid/lithium battery. It uses a combination of switchmode and
linear circuitry to obtain good regulation and low residual noise.
N
ORMALLY, YOU would expect
any adjustable power supply
capable of producing up to 40V and
5A to be a great deal larger than this
little unit. In fact, it fits into a tiny half
1U rack plastic case. It measures just
209 x 43 x 122mm (W x H x D), not in30 Silicon Chip
cluding the knobs and rear terminals.
So how have we managed this feat of
miniaturisation?
The first point is that it is really just
an elaborate regulator and is meant to
be powered by a laptop supply or similar. Second, the circuitry is housed on
one double-sided plated-through PCB
which employs some some surface
mount devices and Mosfets selected
for low-on resistance to produce very
little heat dissipation inside the case.
We have combined the benefits
of switchmode and linear regulator
siliconchip.com.au
Left: the unit is built into
a compact half 1U rack
plastic case measuring
just 209 x 43 x 122mm (W
x H x D), not including the
knobs and rear terminals.
It comes with the panel
meters and load switch
already fitted.
VR1 10k
SET OUTPUT
VOLTAGE
625Ω
TRACKING
FEEDBACK
12-24 V DC
INPUT
L2
F1 10A
12-24V
EMI FILTER
CON1
K
Q1
A
BUCK/BOOST
SWITCHMODE
DC/DC
CONVERTER
REVERSE
POLARITY
PROTECTION
-2.5V
L3
1-41V
RIPPLE FILTER
VOLTMETER
CON2
LOW-DROPOUT
LINEAR
REGULATOR
0-40V
OUTPUT
500mV
SET
CURRENT
LIMIT
VR2
+
–
CURRENT
LIMITING
0.1Ω
1%
CURRENT
MEASUREMENT
SHUNT
AMMETER
Fig.1: block diagram of the switchmode/linear bench supply. The output voltage is adjusted
using VR1 which provides feedback to the low drop-out linear regulator section, which
acts to maintain the feedback potential at 0V. VR2 sets the current limit to 0-5A while the
buck/boost switchmode section monitors the output voltage of the linear regulator and
adjusts its output to provide about 0.7V ‘headroom’ at the regulator’s input.
circuitry. Its output is adjustable over
the range of 0-40V and unlike some
designs, goes all the way down to 0V.
Its current limit is adjustable from 0-5A
and has a fine resolution so that low
currents can be accurately set. The dual
LED panel meters constantly display
the output voltage and current and the
current limit can be displayed and set
without having to short the outputs.
It also has a front-panel load switch;
this lets you set up the required voltage
before switching on power to the load.
Being a hybrid design (switchmode
+ linear), it has much lower output
noise (hash) than a pure switchmode
bench supply and also doesn’t need
a large output capacitor bank that
would then be dumped into the load
in case of a short circuit. In fact, when
the current limit kicks in, the output
voltage drops very rapidly and the unit
goes into current regulation mode. In
other words, it can also be used as a
near-ideal current source.
Since it runs off a low-voltage DC
input, it can even be used away from
230VAC mains and powered from a
car/truck/caravan battery or even a
portable battery pack.
The dual voltage/current displays
are really handy for a bench supply
since you need to able to check that
you have set the right output voltage
and monitor the current draw while
you are performing your tests. It’s also
quite handy to be able to see what the
output voltage has dropped to, should
current limiting be activated.
siliconchip.com.au
One feature missing from some
cheap current-limited bench supplies
is the ability to view the current limit
setting without shorting the output
leads. This is especially useful if you
want to adjust the current limit while
the load is powered since otherwise
you really have no way to know what
you’ve set it to, as long as the load is
drawing less current than the limit.
Buck/boost converter
The switchmode-based bench supplies we have published in the past
have typically used a relatively large
mains transformer to charge a capacitor bank to around 50V. They then used
a step-down (“buck”) switchmode
converter to produce the required
output voltage efficiently. This means
the supply produces much less heat
than a linear design of an equivalent
power level.
In this case though, we wanted to
fit the supply into this neat case from
Altronics which comes with the panel
meters and load switch already fitted.
That ruled out using a large internal
transformer. So we had the idea of
powering it from a high-current DC
supply which constructors may already possess, such as an old PC power
supply or laptop charger. If you’re
like us, you have a few of these lying
around, just waiting to be used for
something grand.
PC supplies usually deliver the most
current from their 12V output while
laptop supplies normally give 15-24V
with the most common being 17V. This
means that our bench supply needs to
be able to step the incoming supply
voltage either up or down, depending
on the required output voltage. And
to be truly useful, it needs to do this
efficiently at a reasonably high power
level, matching that available from a
typical laptop supply (60-100W).
To achieve this, we are using a
“buck/boost” switchmode converter.
This is similar to the more common
“buck” type but it can produce an
output voltage that’s higher, lower or
the same as the input voltage. The particular chip we are using (the LM5118
from National Semiconductor, now
Texas Instruments) operates in buck
mode, boost mode or an intermediate
buck/boost mode, depending on the
ratio of the output to input voltages.
We’ll explain how this works in more
detail below.
As a result, this supply can deliver
plenty of current at lower voltages, up
to about 15V, and then a lesser but still
significant current up to the maximum
40V output (2.5A+, depending on the
input DC supply voltage & power).
Most bench supplies only go up to 30V
and while this is sufficient for many
tasks, we sometimes find it a bit limiting, hence the decision to go to 40V,
even with a reduced current capability.
Performance
As mentioned briefly above, switch
mode-based bench supplies always
have some of the high-frequency
April 2014 31
BUCK MODE (VOUT < VIN x 0.75)
VIN
VOUT
CURRENT
FLOW
S1
BOOST MODE (VOUT > VIN )
VIN
VOUT
CURRENT
FLOW
K
D2
S1
RLOAD
D2
A
K
K
L1
D1
S2
A
S2
A
PHASE 1
VIN
PHASE 1
VOUT
CURRENT
FLOW
S1
VIN
S1
RLOAD
VOUT
CURRENT
FLOW
K
D2
K
S2
RLOAD
A
L1
D1
K
D2
A
K
RLOAD
A
L1
D1
K
L1
D1
A
S2
A
PHASE 2
PHASE 2
S1
S1
S2
S2
I L1
Design concept
I L1
WAVEFORMS
WAVEFORMS
Fig.2: an illustration of how the switchmode converter works, in buck mode
(diagrams at left) and boost mode (diagrams at right). The mode of operation
is determined by whether S2 (actually a Mosfet) is switched with S1 or just
left open (ie, off). In buck mode, as the duty cycle approaches 100%, the
output voltage approaches the input voltage. In buck/boost mode, a 50%
duty cycle gives an output voltage that’s equal to the input, with higher duty
cycles boosting the output voltage above the input, approximately doubling
it at 75% duty cycle, quadrupling it at 87.5% and so on.
switching components present as
‘hash’ in the output, while an ideal
bench supply should have pure DC
with no noise or hash. In many cases, a
switchmode-based bench supply will
have an LC filter (or possibly a more
complex filter involving a differentialmode choke) at the output to attenuate the noise but this is only partially
effective and also adversely affects
output regulation.
Adding a linear regulator stage after
the switchmode stage is a better proposition. This can offer greater noise and
ripple rejection and depending on the
dropout voltage it operates with, can
also result in much better transient
load response. In other words, it can
cope better with sudden changes in
load impedance/current draw, resulting in smaller variations in output
voltage under these conditions.
Without getting into a lot of detail,
the reason for this is that switchmode
32 Silicon Chip
plement the current limiting feature
there. With the linear regulator’s high
bandwidth, that means it can provide
a smooth current flow even in the face
of rapidly varying load impedance
and it also means there can be a very
small output capacitance, so that there
isn’t much stored energy that will flow
through the load before the current
limit is effective. In this case, a 2.2µF
output capacitor delivers a maximum
of 1.75mJ of energy (with the output
set to 40V) into a dead short.
So you can see that combining a
switchmode and linear regulator gives
us the best of both worlds. Well, it isn’t
quite perfect – some switching noise
will still make it through the linear
regulator, for example. However, it’s
a highly effective combination and a
better compromise than either type
of regulator by itself. Implementing
it effectively is a bit tricky though, as
we shall see.
regulators tend to be quite heavily
“compensated”, ie, their “closed loop”
bandwidth is purposefully reduced
to a few kilohertz. This is necessary
because the inductor and capacitors
which are used to convert the switching output to a smooth voltage form a
low-pass filter which leads to a delay
between changes in the switching
waveform and changes in the output
voltage.
This delay is a form of phase shift;
quite a large one in fact. And feedback
systems with large phase shifts are
unstable unless the gain is limited at
higher frequencies. Linear regulators
(depending on design) can have much
smaller phase shifts, allowing more
feedback bandwidth and thus much
more rapid response to any changes in
the output voltage due to the behaviour
of the load.
Having a linear regulator at the
output also means that we can im-
Fig.1 is the block diagram which
shows the overall design of the supply. The output voltage and current are
controlled by the low-dropout linear
regulator section, with VR1 adjusting
the voltage and VR2 the current.
VR1 forms part of a voltage divider
between the output and a -2.5V reference voltage. If the feedback voltage
is above 0V, the regulator reduces its
output while if the feedback is below
0V, the output voltage is increased.
The values selected give the unit a
range of 0-40V.
The LDO regulator needs an input
voltage that’s slightly higher than its
output voltage for proper regulation
(at least 0.1V but ideally a bit more).
As a result, the switchmode converter
monitors this output voltage and attempts to maintain its own output at
a slightly higher voltage. The ‘headroom’ is set at around 0.7V, so the output of the switchmode regulator will
go slightly above 40V and normally
never drops to zero.
An LC filter between the two regulators reduces high-frequency ripple
fed to the linear regulator, as its input
supply/ripple rejection is best at lower
frequencies. There is a similar filter at
the input of the switchmode regulator
to stop too much noise coupling back
to the input and possibly radiating EMI
from the input wiring.
A 10A fuse protects the circuit
against serious faults, however if
siliconchip.com.au
the switchmode section is working
normally, its cycle-by-cycle current
limiting will mean that the fuse should
never blow. Q1 provides input reverse
polarity protection; while it operates
as a diode, it is actually a Mosfet to
avoid reducing the supply voltage too
much and wasting a lot of power, as a
standard diode would.
The voltmeter is wired across the
output terminals while the ammeter
displays the voltage across the shunt.
Note that the voltage across the shunt
is effectively subtracted from the output voltage but the way the feedback
network is connected automatically
compensates for this (as explained
later).
We have used 10-turn potentiometers for voltage and current adjustment
as this makes it easier to set these
values accurately; we recommend
constructors do the same however
there is nothing stopping you from
using the cheaper 270° rotation pots
should you wish.
Buck/boost operation
Most of the switchmode regulators
we have published in the past have
been one of three types, either “buck”,
“boost” or based around a transformer.
The buck and boost types are the simplest but the former can only reduce
the input voltage while the latter can
only produce an output greater than
the input. Hence the use of buck/boost
which gives a much wider range of
output voltages.
The LM5118 IC operates in buck
mode when the output voltage is less
than ¾ the input voltage and boost
mode when the output voltage exceeds
the input voltage. Between these, it
operates in an intermediate mode
which is partly buck and partly boost,
ie, buck/boost.
Fig.2 shows the difference between
the two main modes. At left are the two
states used for buck mode. When S1
is on, current can flow from the input
straight to the output, via inductor L1
and Schottky diode D2. During this
time, L1’s magnetic field charges up
and the current flow smoothly ramps
upwards, at a rate determined by the
voltage across L1 and its inductance.
When S1 is switched off (below),
L1’s magnetic field continues to drive
current through the load via D2, but
this current can no longer come from
VIN, so it must flow through Schottky
diode D1 from ground. The dotted line
siliconchip.com.au
shows how current recirculates – the
only source of energy during S1’s off
time is L1’s magnetic field. As such,
the current flow smoothly drops, again
at a rate limited by the voltage across
L1 (now roughly equal to the output
voltage) and its inductance.
This cycle repeats and the ratio
of S1’s on-time to off-time, in combination with the load impedance,
determines the ratio of the output
voltage to the input voltage, but this
is always less than one. Some example
waveforms are shown below these diagrams, for a steady state (ie, constant
load and output voltage).
Compare these diagrams to those at
right, which show operation in boost
mode. The difference is that now S2
switches on simultaneously with S1.
This increases the voltage across L1 to
be the full input supply voltage and
this does not drop over time, so L1’s
magnetic field charges up much faster.
Thus, more current is delivered during the off-time (below) and hence the
output voltage is higher for the same
duty cycle as buck mode.
It stands to reason then that the ratio
of the output voltage to the input voltage can be greater than one and in fact,
it is inversely proportional to the duty
cycle. Thus the maximum output voltage is limited mainly by the maximum
duty cycle, which for the LM5118 is
related to the operation frequency (as
there is a fixed minimum off-time). In
our circuit, maximum duty cycle is
about 85%, giving a maximum boost
ratio of about 4:1, certainly sufficient
to get an output of over 40V from an
input of 12V.
When in the intermediate mode
mentioned above, the only difference
is that S2 switches off before S1, thus
giving three phases for each cycle,
equivalent to phase 1 for boost, followed by phase 1 for buck and then
phase 2 (same in either mode). Thus
the boost ratio is not as high as in pure
boost mode. This intermediate mode
means there is no discontinuity in
the converter’s operation or output
voltage.
Circuit description
Now let’s turn to the full circuit.
Fig.3 shows the main section. At its
heart is the buck/boost switchmode
converter, controlled by IC1 (LM5118).
First let’s look at the ‘output’ side of
IC1, ie, pins 12-20. These drive the
Mosfets which do the actual switching.
Pin 19 is the high-side driver output
which connects to the gate of Q2 (S1
in Fig.2). Pin 20 is connected to the
source of this Mosfet, which is the
‘floating’ node that switches between
ground and the incoming supply rail.
This pin is used as the ground return
for the discharge current from the
Mosfet gate and as a negative reference
when driving it high, charging the gate
to this voltage plus about 7V.
Thus if the input voltage is say 14V,
the gate of Q2 must be driven to 21V.
To generate this higher voltage, IC1 has
an internal charge pump and it uses
the 100nF capacitor between pins 18
& 20 to accomplish this. This capacitor
is charged to 7V from the input supply
when pin 20 is low and current then
flows back from it into the Mosfet gate
when pin 20 goes high, boosting the
pin 18 voltage to the required level.
This arrangement is known as a “floating high-side driver”.
The low-side (boost) Mosfet, Q3,
is driven from pin 15. This does not
require a charge pump as its source terminal is connected directly to ground
and thus the gate only needs to reach
about 5V for full conduction.
Q2 & Q3 are logic-level Mosfets
which are switched fully on with a
gate-source voltage of 5V. IC1 has an
internal 7V regulator with a 1µF output
filter capacitor connected from pin
16 to ground and this determines the
maximum gate-source voltage fed to
the two Mosfet gates. An external supply can be connected to pin 17 (VCCX)
but the internal regulator can supply
enough current to operate the Mosfets
at 350kHz without excessive dissipation (a maximum of about 650mW).
Ground return for the low-side Mosfet driver is pin 14 (PGND) while pins
12 & 13 are used to sense the voltage
across a 15mΩ shunt connected in series with the buck recirculating diode,
D1. This sets the peak current limit to
125mV ÷ 15mΩ = 8.3A in buck mode
and 250mV ÷ 15mΩ = 16.6A in boost
mode (close to the inductor’s saturation current). The inductor current is
sampled just after the Mosfet(s) switch
off, when it is at its peak, just after D1
becomes forward biased.
Note that the switchmode arrangement is based largely on the sample
circuit in the LM5118’s data sheet,
which provides a design with similar
requirements to ours. We require a
maximum boost of 40V ÷ 12V = 3.3
times with an input current of around
April 2014 33
Fig.3: the main section of the bench supply circuit. IC1 is the buck/boost controller and this drives Mosfets Q2 & Q3
which form the switchmode converter in conjunction with inductor L1 and Schottky diodes D1 & D2. The output
then goes to the linear regulator (shown in detail next month) and then to the output terrminals. The output current
is checked using a 0.1Ω shunt resistor which is monitored both by the ammeter panel and the current limit circuitry.
REG1-REG3 and IC2 generate two extra supply rails for the linear regulator circuitry, one about 10V above the main
supply rail (VBOOST) and a -5V rail, using a charge pump driven by Q6 & Q7. VR1 & VR2 set the output voltage and
current limit while VR3-VR8 zero and adjust the meter displays, output voltage range and current limit range.
34 Silicon Chip
siliconchip.com.au
8A (100W ÷ 12V), while their design
is for a maximum boost of 12V ÷ 5V
= 2.4 times with an input current of
around 8A (36W ÷ 5V).
We are using 45V 15A Schottky
diodes for D1 and D2 since these are
the components in the switchmode
siliconchip.com.au
section which dissipate the most
power. They are relatively compact
devices which is important since the
PCB layout of this section is critical.
The BUK9Y6R0-60E Mosfets were
chosen due to their low gate charge
(minimising switching losses), low
on-resistance (minimising I2R power
dissipation) and ease of soldering.
While the LM5118 data sheet says a
snubber is not required, we have fitted
one – consisting of a 10Ω resistor and
10nF capacitor in series, across D1 –
to reduce voltage spikes and thus EMI
April 2014 35
The PCB assembly is designed to mate with integral pillars inside the half 1U rack plastic case. Pt.2 next
month has the full assembly details. Note that the final board will differ slightly from this prototype.
during Mosfet switching.
A parallel array of eight 10µF 25V
multi-layer ceramic capacitors is used
for input supply filtering. This arrangement has a very low ESR and is also relatively cheap. A similar arrangement
of nine 4.7µF 50V ceramic capacitors
is used for the output, where low ESR
is important to minimise ripple. These
form a low-pass filter in combination
with inductor L1. A couple of 47µF
low-ESR electrolytics are paralleled
for ‘bulk capacitance’, which helps
switchmode feedback loop stability.
The output of the switchmode regulator passes through another LC lowpass filter, consisting of a 3.3µH bobbin inductor (chosen for its low losses
and low price) followed by a 220µF
low-ESR electrolytic capacitor. This
attenuates the output ripple of the
switchmode regulator and this voltage
then feeds into the linear regulator (to
be described next month).
Feedback & control circuitry
Pin 1 of IC1 is the supply input for its
high-side and this is decoupled with a
100nF ceramic capacitor close to that
pin of the IC. The resistive divider connected to pin 2 sets the under-voltage
lock-out threshold at 11.3V [1.23V x
(82kΩ + 10kΩ) ÷ 10kΩ]; while the IC
can run from at little as 5V, we want
to avoid excessive input current draw
at low supply voltages.
The 100nF capacitor at pin 2 sets up
the ‘hiccup’ over-current protection;
if a prolonged over-current condition
is detected, pin 2 is pulled to ground
and this capacitor takes some time to
36 Silicon Chip
charge to the 11.3V threshold, preventing excessive current draw in case of
a prolonged short or other overload.
A resistor from pin 3 to ground sets
the operating frequency of the switchmode regulator, with 15kΩ giving
operation at around 350kHz. Higher
frequencies mean less RMS ripple
voltage at the output but in exchange
for that, switching losses are higher
(due to more frequent transitions at
the output). Also, inductors tend to be
lossier at higher frequencies.
Pin 4 is the enable input and must be
pulled high for the regulator to operate.
This is connected to power switch S1
via link LK1, with a 100kΩ pull-down
resistor. Thus, when power switch S1
is off, voltage is still applied to the
input of IC1 but it is disabled so the
output is at 0V. This avoids S1 having
to switch a high current. LK1 is used
to disable and bypass the switchmode
regulator for testing the rest of the
circuit independently.
The capacitor connected from pin
5 to ground sets the time constant for
the ‘emulated ramp compensation’.
This value is chosen to match the time
constant for the rate of change of current flow through the output inductor
(L1) and allows the IC to perform ramp
compensation without needing to
measure the current through L2. Ramp
compensation is required for feedback
loop stability at higher duty cycles.
Readers should refer to our article
on the LED Dazzler in the February
2011 issue for a detailed explanation
of ramp compensation. However, in
brief, it involves feeding back a volt-
age related to the output duty cycle
to the input of the error amplifier, in
order to avoid the duty cycle oscillating either side of the required stable
value without settling down.
Pin 6 is the analog ground pin, ie,
the ground return for the components
connected to pins 1-9. Pin 7 is the soft
start pin and the connected capacitor is
charged at power-on, with the output
duty cycle being limited until it is fully
charged, to prevent the IC from drawing very high input currents while the
output capacitor bank is charged.
Pin 8 is for voltage feedback and is
the input to the error amplifier. When
pin 8 is below 1.23V, the output duty
cycle increases and if it is above 1.23V,
the output duty cycle is reduced. Normally this is connected to the output
of the regulator via a resistive divider,
to set a fixed output voltage, or with
one resistor replaced by a rheostat or
potentiometer to give an adjustable
output voltage.
But in this case, we don’t want to
set the output voltage of the switchmode regulator directly. Instead, we
want its output to be slightly higher
than that of the linear regulator, so
that it has ‘headroom’ to operate and
deal with load transients but without
dissipating much power in the linear
pass element.
This is achieved using PNP transistors Q4 & Q5. These form a current
mirror and their emitters are tied to
the output of the switchmode regulator. The difference between this and
the output of the linear regulator
causes a current to flow through the
siliconchip.com.au
680Ω resistor at Q5’s collector. When
the difference is 0.75V, that current is
(0.75V - 0.5V) ÷ 680Ω = 0.37mA.
Being a current mirror, this also
flows through the 3.3kΩ resistor at
the collector of Q4 which gives a
voltage of 3.3kΩ x 0.37mA = 1.23V,
which is IC1’s internal reference voltage level. Thus, its negative feedback
will maintain the switchmode output
voltage about 0.75V above the linear
regulator’s output voltage. Being effectively a common-base amplifier, this
arrangement has very little phase shift
and thus does not affect IC1’s feedback
loop stability.
Note, though, that the output of
the switchmode regulator can’t drop
below 1.23V as this is the minimum,
so dissipation in the linear regulator
will be a little higher when its output
voltage is below 0.5V or so.
ZD7 and ZD8 prevent the output
of the switchmode regulator from
exceeding 45V in case of a feedback
fault. The 10kΩ current-limiting resistor in combination with zener diode
ZD9 and Schottky diode D19 protect
the feedback input (pin 8 of IC1) from
going outside the range of -0.3 to 5V,
which would risk destroying the IC.
Diode D18 protects Q4 & Q5 from
damage due to base-emitter junction
reverse breakdown.
Linear regulator supply rails
The linear regulator requires a positive supply rail (VPP) that is at least 7V
above the output in order to switch
its internal Mosfet on fully to supply
a high load current. It also requires a
siliconchip.com.au
well-filtered negative rail (VEE) several
volts below ground so that it can turn
that Mosfet fully off when required (allowing for the voltage drop of several
internal driving transistors).
These are supplied from a charge
pump shown in Fig.3. This is based
on REG1, a 12V low-dropout regulator and IC2, a 7555 CMOS timer IC.
IC2 is set up to provide a 12V square
wave at 100kHz at its output pin 3.
This drives a complementary pair of
bipolar transistors, Q6 & Q7, which
form an inverting buffer. The 100pF
capacitors across their base currentlimiting resistors speed up switch-off
to prevent cross-conduction.
The buffered output drives two 1µF
capacitors, one of which is charged
to the switchmode output voltage by
D5 and the other which is clamped
to ground by D3. When the buffered
output of Q6/Q7 goes positive, D6
becomes forward biased and the connected 100µF capacitor is charged to
roughly 10V above the switchmode
supply rail. Similarly, when the collectors of Q6/Q7 go negative (to ground),
D4 becomes forward biased, charging
the connected 100µF capacitor to
about -10V.
These two new voltage rails are then
filtered using RC filters (10Ω/220µF
and 10Ω/100µF respectively) to remove most of the 100kHz component,
forming relatively smooth DC supply
rails. The -10V rail is then regulated
by REG3 to a stable and clean -5V for
the linear regulator’s VEE rail. This
regulation is necessary for two reasons:
(1) any noise or ripple on this line will
affect the regulator’s output; and (2)
this is also used as the reference for
setting the output voltage.
Adjustments & trimming
As stated earlier, the linear regulator
acts to keep the feedback voltage at
around 0V. This is determined by the
output voltage in combination with
the position of 10kΩ potentiometer
VR1 (Fig.3). This can be a 10-turn
potentiometer to give finer output
adjustment. It acts as a voltage divider
in combination with trimpot VR3 and
the 470Ω resistor connecting them.
VR3 is connected to a -2.5V rail,
derived from the -5V rail by voltage reference REG4. The circuit will operate
without REG4 however it will be subject to output voltage variations due to
thermal drift in -5V regulator REG3;
REG4 has much better thermal stability
and is not dissipating anywhere near
as much power either (typically <1mW
compared to ~100mW for REG2).
When VR1 is at minimum resistance, the output rail is effectively
connected directly to the feedback
point and so the output voltage is at
0V. As VR1 is turned clockwise and its
resistance increases, the output voltage must increase in order to keep the
feedback voltage at 0V. For example,
when the resistance of VR1 is around
625Ω, matching that of VR3 plus the
470Ω series resistor, the output is at
around 2.5V.
VR3 is used to trim out variations in
the other components, giving 40V at
the output with VR1 fully clockwise.
The output voltage is fed to VR1 via
link LK2 in parallel with diode D13.
This is intended to allow for some
wiring voltage drop compensation
to be used due to the need to run a
wire from the output terminal to the
off-board load switch, in which case
LK2 is removed and a wire is run from
the supply side of the load switch to
the lower terminal of LK2. In case this
connection fails, D13 limits the output
voltage from rising more than 0.6V.
The current limit is set using VR2,
another 10kΩ potentiometer which
can also be a 10-turn type. Its wiper
voltage is filtered with a 100nF capacitor and fed into the linear regulator
where it is compared with the voltage
across a 100mΩ shunt. There is 500mV
across VR2, derived from the +5V rail
by trimpot VR5. This +5V comes from
linear regulator REG2 which supplies
several reference voltages but also the
power for the two panel meters.
The voltmeter reads the voltage
across the output terminals but this is
a 200mV full-scale meter so the output
voltage is divided down by 1MΩ and
1kΩ resistors plus 500Ω trimpot VR7
for fine tuning. The panel meter’s input
impedance is 100MΩ so we are using
relatively high value resistors here.
Similarly, to get a reading of up to
5A on the ammeter panel, we need a
0-50mV signal and so the 0-500mV
from the shunt (at CurrSense) is
divided down by a factor of 10 by a
910kΩ resistor, 100kΩ+1kΩ resistors
and 20kΩ trimpot VR8. S2 allows
the voltage feeding this divider to be
switched from the current feedback
to the current limit setting, so that the
limit can be viewed without having to
short the output terminals.
continued on page 96
April 2014 37
Advertising Index
Altronics.................................. 75-79
Apex Tool Group............................. 5
Bitscope Designs......................... 13
Control Devices Pty Ltd.................. 7
Electrolube................................... 39
Emona Instruments...................... 11
Enertel Pty Ltd............................. 29
Front Panel Express....................... 8
Futurlec.......................................... 8
Gless Audio.................................. 95
Switchmode Bench Supply
. . . continued from page 37
The 10MΩ and 2.2MΩ resistors
provide a small bias current to the two
panel meters so that they do not give a
negative reading when the output voltage is 0V or no current is being drawn.
The two remaining trimpots, VR4
and VR6, are used to trim out any offset error in the voltage feedback and
current limiting circuitry respectively.
These inputs have a low impedance to
ground so the adjustment ranges span
just a few millivolts either side of 0V.
Remaining circuitry
The circuit is protected from a reversed input supply polarity by Mosfet
Q1. When the supply is connected the
right way, Q1’s gate is pulled positive
by the 100kΩ resistor and clamped at
a safe level by the 15V zener diode.
This switches it on and allows ground
current to flow from the circuit back
to the supply.
If connected backwards, the gate is
pulled negative and so Q1 remains off.
Its body diode is also reverse-biased
and thus very little current will flow.
The 100nF capacitor from its gate
to ground slows its turn-on to avoid
large current spikes charging the input
capacitor bank when power is first
supplied; IC1 has a soft-start feature, so
it’s just this input bank that can draw
a high current initially.
A 10A fuse protects the circuit
against serious faults while 27V zener
diode ZD2 conducts if the input supply voltage becomes too high. If that
excessive voltage is maintained for
very long, it will blow the fuse. The
clamping voltage is above the 25V
rating of the input capacitor bank but
they are unlikely to fail due to a brief
96 Silicon Chip
Harbuch Electronics..................... 91
DOWNLOAD OUR CATALOG at
www.iinet.net.au/~worcom
Hare & Forbes.......................... OBC
High Profile Communications....... 95
WORLDWIDE ELECTRONIC COMPONENTS
PO Box 631, Hillarys, WA 6923
Ph: (08) 9307 7305 Fax: (08) 9307 7309
Email: worcom<at>iinet.net.au
Icom Australia................................ 9
Notes & Errata
LD Electronics.............................. 95
Soft Starter for Power Tools (July
2012): the 10mΩ SMD shunt
was left off the parts list. This
should be a 6332 (metric)/2512
(imperial) size SMD chip resistor
with a rating of at least 2W, such
as CRA2512-FZ-R010ELF (element14 Cat. 2394421).
Microchip Technology................... 71
Jaycar ......................... IFC,45-52,92
Keith Rippon ................................ 95
KitStop.......................................... 10
Master Instruments...................... 41
Mikroelektronika......................... IBC
Ocean Controls............................ 12
Quest Electronics......................... 95
RF Modules.................................. 96
Rohde & Schwarz.......................... 3
Sesame Electronics..................... 95
Silicon Chip Binders..................... 72
over-voltage of just a few volts and
we don’t want ZD2 to conduct any
significant current with the supply
below 25V.
A 4.7µF capacitor and 3.3µH inductor L2 prevent much switching noise
from passing back through the input
leads, which could lead to electromagnetic interference being radiated
from them. Power switch S1 enables
the switchmode regulator and at the
same time, applies power to the rest
of the circuit.
When LK1 is moved to the “Test”
position, the linear regulator remains
off and power can bypass it from S1
straight to the output. This is so that
the constructor can check the linear
regulator and other circuitry is working before activating the switchmode
portion; otherwise troubleshooting
could be very difficult.
Finally, there is a Schottky clamp
diode (D16) at the output of the switchmode regulator so that its output can
not be pulled very far below ground by
Silicon Chip Online Shop........ 88-89
Silicon Chip Subscriptions........... 75
Television Replacements............. 95
Verbatim....................................... 55
Vicom Australia............................ 43
Wiltronics........................................ 6
Worldwide Elect. Components..... 96
the linear regulator at start-up. There
is also a clamp consisting of two 27V
zeners (ZD5 & ZD6) in series after filter
inductor L3, so that if the switchmode
regulator feedback fails (including the
ZD7/ZD8 voltage clamp), its output
will not go high enough to damage the
63V filter capacitors or any part of the
linear regulator circuitry.
LDO operation & construction
That’s all we have room for this
month. Next month, we’ll describe
the linear regulator section and begin
the construction.
SC
siliconchip.com.au
|