The worlds simplest switching PSU

This
ultra simple switcher converts 12V to 24V for powering 24V appliances of a car
battery. The

regulation
is quite poor but efficiency is typically between 60 and 85%, and generally
goes up when

VIN is nearer VOUT, i.e. it"s more efficient
to convert 18V to 24V than 9V to 24V. As the input

voltage
rises the current consumption goes down, Tr1 is the first major power loss,
closely followed

by D2,
there"s nearly no loss in L1, in fact L1 is more efficient than a transformer
as there is no

secondary
winding.

It
cannot be run as a step-down regulator as when VIN gets near VOUT, VOUT will start to rise to an

unacceptable
level, this is because when VIN is > VOUT by more than about 600mV D2 will conduct

and VIN will charge C2 via L1. The
circuit I built gives 24.5V with the input between 9 and 18V, the

output
doesn’t vary more than 500mV, I
managed to draw more than 500mA before the output fell

below
24V.

Circuit Operation

I"m not
going to go in to too much detail here. L1, R1, C1 and Tr1 form a blocking
oscillator who"s

frequency
will depend on the characteristics of the core and the number of turns.

When Tr1
turns on a current will start to flow in the driver coil this will cause a
magnetic field to

build up
around the coil in the core. Tr1 turns off, the sudden change in magnetic field
will induce a

voltage
spike across the coil (as the inductor resists the change in current by
producing a high

voltage
to try and keep the current flowing). This spike it transferred to C2 via D2,
if the circuit is

left
running C2 would charge to a high voltage (this could be 100s if not 1000s of
volts and will

again
depend on the nature of L1). When the voltage on C2 is high enough to cause D2
to conduct

in
reverse, Tr2 will turn this will cut of the base of Tr1 and stop the
oscillator. The connected load

will
eventually suck enough volts of C2 to stop D2 conducting, Tr2 turns off and the
oscillator

continues.

Construction

Not
critical, I assembled my first prototype with fly leads and Tr1 mounted on a
heat sink.

Use a
current limited power supply when you first build this circuit, also mount Tr1
on a good large

heat
sink, if you find out you don’t need it later you can remove it or use a smaller one.

If is
doesn’t oscillate, reverse the
connections to the driver or feedback windings on L1, if it still

doesn’t work check all the connections.

When I
first built this circuit I connected up oscillator first (L1, R1, C1, Tr1)
ignoring the rest. I

could
tell it was working as I could hear a whining sound, if you are unable to hear
a noise you

could
connect the collector of Tr1 to an oscilloscope and view the output. You may
have to connect

a diode
in reverse across the driver winding of L1 (to protect Tr1 from the high
voltage pulse),

remove
it when you decide to build the rest of the circuit.

L1 Details

Non
critical, although you may have to do some experimentation, as not all
configurations will

work.

Wind the
following coils onto a 20mm ferrite E-core, the order is unimportant although I
do

recommend
winding the driver first, as you want a low resistance, the resistance of the
feedback is

less
important. To achieve maximum efficiency you will have to experiment with
different numbers

of turns
for the driver and feedback.

Driver:
18 turns of 22SWG copper glazed wire.

Feed
back: 10 turns of 35SWG copper glazed wire, doesn"t need to be thick as it only
carries a small

signal
to the base of Tr2.

Alun
Jones

CAUTION!

If Tr2 is bad or fails (which is unlikely
if the circuit is working properly) a high voltage

will build on C2 causing it to explode,
this will also happen if it is not connected

properly, the same also goes for R2 and
D1. To avert disaster connect a 5W zener

diode with the same or lower voltage than
C2 is rated for in reverse parallel with C2

see diagram (DCR (crowbar) drawn in grey).
Turn off the power if it starts to get warm.

R1

C1

L1

Tr1

Tr2

R2

D1

D2

C2

VIN

VOUT

DCR

Parts List

All
resistors are ¼W metal or carbon film

Part
Value Substation / Comment

R1 10K
Will vary depending on the nature of L1 and Tr1. Try values between 220R

and 22K.

R2 33K
10K to 100K will depend on R1, Feedback, Tr2 and, D2.

C1 4.7nF

Ceramic

Doesn’t seem to affect the frequency of the oscillator
much. May not be

needed,
as the parasitic capacitance of the feedback winding might be

sufficient
to start oscillation. 220pF to 220nF use the lowest value that works.

C2 220?F 35V

Aluminium

electrolytic

Try
different values, between 10 and 1000?F you may benefit from using a

tantalum
bead. The lower the frequency output the bigger value you will need.

Tr1
BD135

TIP3055

TIP31

TIP121

Faster
transistors maybe of some benefit. Try a high gain Darlington pair

power-transistor
so a large value for R1 can be used, this will increase the

efficiency
at lower currents, but the saturation voltage will be higher and it will

be less
efficient at higher currents.

Tr2
2N2222A Any general-purpose transistor, try BC547, BC337, 2N3904, ZTX540 etc.

DCR 30V 5W Only needed once.

D1
BZX79C24

(24V)

Any low
power zener will do try different voltages, ok to connect several

values
in series to increase the voltage rating.

D2
1N4001 General purpose rectifier, try a Schottky barrier e.g. MBRA120.