ATX Power Supply Tester -ELEKTOR.pdf

ATX Power Sup

Ton Giesberts

PC power supplies can often be bought cheaply at

places such as computer fairs. But it isn’t that easy

to check if such a (second hand) power supply still

works properly. This

dedicated tester makes

that job quick and

straightforward.

pply Tester

Checks all voltages

New ATX 2.2 specification

ATX connector

20-pin

24-pin

pin 1

pin 11

pin 1

pin 13

This tester was designed for recent ATX power supplies, but it is

also ready for use with new power supplies described in version

2.2 of the ATX specification. These have a main connector with 24

pins instead of 20 (75 Watt extra for use by PCI Express cards).

There is a curiosity in the new specification regarding the -5 V con-

nection. According to version 2.2 of the specification it is no longer

used and the pin in question (20) is marked as NC (not connected).

However, according to the manuals of several motherboards with a

new 24-pin connector the –5 V is still present. So keep in mind that

when you test a power supply with a 24-pin connector the –5 V

output may or may not exist. The –5 V should always be present on

a 20-way connector.

The change from 20 to 24-pin connectors is compatible with the

older 20-pin connectors, with an extra +3.3 V, +5 V, +12 V and

ground added to one end. An older ATX power supply with a 20-

pin connector fits in a 24-pin socket and can only be inserted one

way, so mistakes aren’t possible.

+3V3

-12V

GND

PS - ON#

GND

-5V

+5V

+3V3

-12V

GND

PS - ON#

GND

+3V3

GND

+5V

GND

+5V

GND

PWR - OK

+5VSB

+12V

+3V3

GND

+5V

GND

+5V

GND

PWR - ON

+5VSB

+12V

+5V

pin 10

pin 20

+12V

+3V3

+5V

GND

pin 12

pin 24

040112 - 12

Figure 1. The pin-outs for 20 and 24-

pin ATX power connectors.

Apart from the power supply and this

tester, you’ll only need a mains cable

(and socket!). All outputs from the

power supply can be tested under load

and any deviations from the nominal

values are shown on 6 LEDs.

to the motherboard. A quick test

would be very useful then. The true

hobbyists may also want to investi-

gate the exact fault in a broken

power supply. But it isn’t a straight-

forward job to test a PC power sup-

ply with a multimeter.

The power supply tester described

here is a very useful and compact

tool. We have to admit that you prob-

ably won’t need it very often. But

once you have acquired one, word

will spread amongst your circle of

friends and you shouldn’t be sur-

prised when you’re called to ‘quickly’

check a PC power supply for them.

put voltages. The percentage devia-

tion of a selected output is shown on

6 LEDs. Two of these LEDs show

whether the deviation is positive or

negative and the other four indicate

the percentage difference from the

required output voltage.

For output voltages that are con-

nected to more than one pin only the

first pin is tested. (A power supply

generates only a single +5 V supply,

even though it is made available on

several pins.)

There is a 26-pin header (K2) on the

PCB that can be used to test each

pin individually. The outputs are con-

nected through 1 kΩ resistors to pro-

tect them against short circuits. If

you connect an extension lead to this

header you can use a multimeter to

take measurements from any pin.

Although the power supply in a PC

has little bearing on its overall

speed, there are times when it needs

to be replaced. This may be because

the old power supply has simply

given up the ghost, and sometimes

the internal fan has become too

noisy, or an upgrade of the PC has

increased the power requirements

above that what the old power sup-

ply can deliver.

ATX power supplies are available

from virtually every computer shop.

When you buy a new power supply

it is obviously safe to assume it will

be in perfect working order. But

when you buy a (used) power supply

at a computer fair or boot fair you

want to be sure that it works before

you fit it into the case and connect it

What is measured?

Our tester doesn’t require a separate

power supply, as it takes its power

from the PC power supply under

test. All you need to do is plug the

power supply into the tester and

then use a mains lead to connect it

to the mains. A rotary switch is then

be used to quickly check all the out-

A look at the circuit

An ATX power supply has a total of

6 output voltages, which all have to

be tested: +3.3 V, +5 V, +5 V for

1/2005 - elektor electronics

pply Tester

+3V3

R28

3k3

R29

28k7

IC1 = TS922IN

+2V5

+5VSB

R49

R30

100k

+5VSB

IC3 = TS924IN

+5V

220n

R45

R31

100k

+5VSB

S1.A

S1.B

100k

D 4

R34

IC2

R32

15k

R33

365k

+12V

5...10%

IC1.B

R55

R35

200k

R37

453k

IC3.B

-5V

R36

27k

R38

100k

MDX

R46

R51

R50

-12V

3X1

3X2

+5V

+5VSB

220n

D 5

D 3

R39

10k0

+5VSB

R13

10...20%

<+/–5%

R59

R53

IC3.C

+5VSB

+2V5

R40

10k0

R47

R52

D 2

R60

12k

74HC4053

POWER

D 7

IC4

IC1.A

D 6

R42

R43

R11

10k

neg.

IC3.A

>20%

100p

R48

IC3.D

BC547B

R41

R12

D 8

R61

12k

+2V5

R44

+5VSB

LM4041

DIZ_ADJ

pos.

+2V5

R54

PWR_ON

PS_ON

250 Ω

IC1

IC2

IC3

100n

+3V3_3

R14

R16

+3V3_3

+3V3

+3V3_2

+3V3

R15

R27

-12V

+3V3_2

-12V

+3V3

+5V

+12V

+5VSB

R18

R57

+5V

R10

R56

+5V_2

R19

D 1

-5V

R58

R26

-5V

R23

R20

+5VSB

+12V

+12V_2

+3V3_4

+5V_3

+5V_4

+5V_5

+5VSB

+5V_3

10W 10W

R24

R21

+5V_4

+12V

STANDBY

+12V_2

R25

R22

+5V_5

+3V3_4

R17

ATX

PSU

10W

-5V

-12V

040112 - 11

Figure 2. The measurement circuit itself is fairly small. A lot of room is taken up by the power resistors (R1-R9), which load the

power supply.

standby, +12 V, –5 V and –12 V. The

standby voltage (+5VSB) is always

present as long as the mains is con-

nected. This voltage is therefore

used as the supply for the tester

( Figure 1 ). LED D1 is driven directly

from the +5VSB supply and hence

indicates that the mains is turned on

and that the power supply has at

least a working standby voltage.

The power supply is turned on by

closing switch S2. This pulls pin

PS_ON sufficiently low via R56.

According to the specification this

pin should be <0.8 V at 1.6 mA. A

value of 470

A at a minimum voltage of

2.4 V, a buffer stage consisting of

R11, R12 and T1 has been added.

Once the mains is turned on (and D1

and D2 are lit), S1 is used to select

the voltage that is connected to the

input of amplifier IC1b.

S1 is a 2-pole 6-way rotary switch (it

has to be a break-before-make type,

otherwise you’ll introduce shorts in

the outputs). The first switch selects

the supply voltage to be tested. The

common output of this switch is also

connected to a PCB pin (via a 100 Ω

resistor for protection). It is possible

to connect a small voltmeter module

to this pin, so that the absolute value

of the selected voltage can be seen.

Next to the connection for the meter

(M1) is an extra PCB pin with +5 V

for the voltmeter module.

The selected voltage makes its way

via the common of S1b to one of the

potential dividers connected to the

inputs of IC1b.

Each resistor combination gives the

right amount of attenuation to the

chosen voltage such that the output

of IC1b will be a nominal 2.5 V at

every position of S1. There is no need

for a symmetrical power supply to

measure negative voltages because

IC1b is a rail-to-rail type opamp.

With positive voltages IC1b func-

tions as a non-inverting buffer. The

two negative supply voltages are

inverted and attenuated.

We now take a small jump to the tol-

erance LEDs in the circuit (D3-D8).

According to the ATX specification

all voltages should be within

for R56 achieves this.

The PWR_ON output, also called

PWR_GOOD or PWR_OK, is used by

the power supply to show that the

most important outputs (+12 V, +5 V

and +3.3 V) are within their limits

and can supply a nominal current.

When this signal is active, D2 lights

up. Since this output can only source

5%,

with the exception of -12 V, which

may be ±10%. We have therefore cho-

sen four tolerance ranges that are

covered by the LEDs: <5% (green

LED D3), 5-10% (yellow LED D4), 10-

20% (red LED D5) and >20% (second

red LED D6). The range division at

10% was used to give you the choice

whether to accept that deviation or

elektuur - 1/2005

200

Circuit details

The potential dividers for IC1b have been designed as accurately as possible through the use of resistors

from the E96 series. Three of the dividers are made with a (large) E96 and a (small) E12 resistor to get as

close to the theoretical value as possible. Since the value of the E12 resistor is much smaller than that of

the E96 resistor connected in series, it only has a small effect on the total tolerance. Hence a resistor from

the E12 series is suitable here.

Although capacitor C6, which is connected in parallel to reference zener IC4, is not essential according to

the data sheet, a little bit of HF decoupling never does any harm with a switched mode power supply.

R41 reduces the effect of the input bias current of opamp IC1a, keeping any error limited mainly to that

from the tolerance of resistors R39 and R40.

A small amount of hysteresis is required around IC3a to make it switch cleanly. This does introduce a

small error near the zero point as far as a positive or negative deviation concerns (

0.1%), but this is very

small compared to the tolerance levels we’re looking at.

For IC3b-d, which are used as comparators, we have intentionally used opamps rather than real com-

parators because these usually have open-collector outputs. These wouldn’t be suitable for this purpose.

The reference voltages (via R45-R48 and P1) for the comparators are 5%, 10% and 20% lower than the

main 2.5 V reference (2.375 V, 2.25 V and 2 V respectively). Resistors R45 and R46 in the potential

divider should of course have been exactly 500

, but 499

is a difference of only 0.2%, which is much

less than the tolerance of the resistors themselves.

not. A difference of more than 20% is

not acceptable in any case.

These LEDs are driven by compara-

tors IC3b-d, which have their invert-

ing inputs connected to a potential

divider (R45-R48 and P1). This deter-

mines the tolerance ranges with

respect to the 2.5 V reference volt-

age. P1 is used to set the reference

levels as accurately as possible.

This just leaves the section that joins

the output signal from IC1b to the

LEDs. This output signal is nominally

2.5 V and may be a bit more or less

when it deviates. But the comparator

circuit built round IC3b-d can only

indicate negative differences. To get

round this problem IC1a inverts the

output signal from IC1b. This is fol-

lowed by an analogue switch that

can be controlled using a digital sig-

nal. This switch is part of IC2 (a

triple analogue multiplexer). The out-

put signal from IC1b and the

inverted one from IC1a are con-

nected to inputs Y0 and Y1 of an

analogue switch (pins 2 and 1 on

IC2). The output of IC1a is also con-

nected to opamp IC3a, which acts as

a comparator and compares the sig-

nal with the 2.5 V reference voltage.

The output of IC3a acts as the con-

trol signal for the analogue switch.

When the deviation is negative

(<2.5 V), IC3a switches pin 2 of IC2

to the output (pin 15), which is con-

nected to the comparators. When the

deviation is positive (>2.5 V), the

inverted signal (pin 1) is connected

to pin 15. In this way LEDs D3-D6

always show the deviation com-

pared to the nominal value. The out-

put of comparator IC3a is also con-

nected to two LEDs, which indicate

if the measured voltage is greater or

smaller than the nominal value. The

yellow LED (D7) is lit when the volt-

age is lower and the red LED (D8)

indicates that the voltage is higher

than the reference voltage.

The 2.5 V reference voltage men-

tioned a few times previously is sup-

plied by an LM4041DIZ-ADJ (IC4)

made by National Semiconductor.

This voltage can be adjusted to

exactly 2.5 V with preset P2.

All outputs from the ATX power sup-

ply are provided with a resistive

load, where some outputs are loaded

more than others. The +3.3 V and

+5 V outputs often require a mini-

mum load for the power supply to

operate correctly, and are therefore

loaded more heavily. To avoid exces-

sive heat generation we haven’t

taken the maximum power from the

supply, but have limited it to some

45 W (R1 to R9).

Construction

The PCB designed for the tester is

shown in Figure 3 . The dimensions

of the PCB have been kept as small

as possible and are not based on any

particular enclosure. The ATX power

supply connector is on the edge of

the PCB, so that this can stick out

through the side of an enclosure.

1/2005 - elektor electronics

COMPONENTS

LIST

Resistors:

R1,R2 = 2

2 10W

R3,R4 = 3

3 10W

R5,R6 = 22

10W

R7 = 33

R8 = 33

10W

R10,R13-R27,R42,R43,R49,R51-

R54,R57,R58,R59 = 1k

R11,R12 = 10 k

R28 = 3k

R30,R31,R34,R38 = 100 k

Ω

R33 = 365k

R55

R35 = 200k

R59

R60

Ω

R37 = 453k

Ω

R39,R40 = 10k

R41 = 4k

R61

IC1

Ω

R45,R46 = 499

R43

R47 = 1k

R45

R44

R42

R48 = 7k

IC3

IC2

R50 = 820

R55 = 100

R47

R52

Ω

R60,R61 = 12k

R50

R49

R51

P1 = 250

preset

P2 = 1k

preset

Capacitors:

C1,C2 = 220nF

C3...C5 = 100nF

C6 = 100pF

Figure 3. There is room on the PCB for all components. The power resistors are

mounted on top of each other.

Semiconductors:

D1,D2,D5,D6,D8 = LED, red, low-

current

D3 = LED, green, low-current

D4,D7 = LED, yellow, low-current

T1 = BC547B

IC1 = TS922IN (ST Microelectronics,

Farnell # 332-6275)

IC2 = 74HC4053

IC3 = TS924IN (ST Microelectronics,

Farnell # 332-6299)

IC4 = LM4041DIZ_ADJ (National

Semiconductor, Farnell # 271-263)

This makes it much easier to insert

the connector from an ATX power

supply.

There are no ‘special’ parts on the

PCB. As long as you take care with

the polarity and values of all compo-

nents, and solder neatly, you should-

n’t have any problems with the con-

struction.

All the power resistors are also

mounted on the PCB. Due to the heat

these generate they should be

mounted at least 2 or 3 mm above

the PCB, otherwise the PCB will give

off smells. (The resistors will do that

in the beginning anyway). Resistors

R1, R3 and R5 are mounted another

2 to 3 mm above R2, R4 and R6. This

method of construction leaves

enough air around the power resis-

tors for ventilation.

Before you mount the board into an

enclosure or drill any holes, you

should make a careful note of the dis-

tance between the rotary switch and

the ATX power supply header. The

wiring for the LEDs and the on/off

switch can be made with thin

stranded wire.

Since this circuit generates a fair

amount of heat, it is advisable to use

a metal enclosure with sufficient

(possibly even forced) cooling. A

miniature 5 V fan will be essential if

you use a small enclosure. This can

be connected to the +5 V pin for the

voltmeter module. Make sure that

you have enough ventilation holes in

the enclosure.

To give the tester a professional look,

and make it easier to use, we have

produced a front panel, which is

shown at a reduced size in Figure 5 .

Miscellaneous:

K1 = 24-way angled ATX header, PCB

mount (Molex 39291248, Farnell #

413-8508)

K2 = 26-way boxheader (2x13)

S1 = 2 pole 6 position rotary switch,

PCB mount

S2 = on/off switch, 1 contact

Optionally:

M1 = 3 1 / 2 -digit LCD voltmeter module,

range 0-20 V (e.g., Farnell # 422-

0146)

Enclosure: e.g., type 1455L1601BK

(Hammond Manufacturing)

Calibration and

operation

There are two presets on the PCB

that can be used to set the tester up

accurately, although the circuit

works perfectly well when they are

set to their mid-position. For those of

you who want to set the tester up as

accurately as possible we’ll explain

the calibration procedure.

PCB, order code 040112-1 , see

Readers Services page

elektuur - 1/2005

R9 = 10

R29 = 28k

R32 = 15k

R36 = 27k

R44 = 1M

R56 = 470

Plik z chomika:

Inne pliki z tego folderu:

Inne foldery tego chomika: