repairing a switch mode plug pack

[Peter]

I have a couple of d-link wirless accesspoints on whose power supplies have
failed - they should deliver 5volts at 2 amps.
The power supply takes the form of a slim-line plug pack.
Inside the pack is a little switch mode power supply.
I'm looking for some tips on what to look for in repairing this item. there
is nothing obviously wrong on a visial inspection. Fuse is intact, no heat
stress on the board, no visually distressed components. Just nothing on the
output.
On the board there is a pwm chip - CM3842 - an 8 pin device. There is 8
volts on its Vcc pin (pin 7) - nothing appreciable on any other pins.
On there board there is also a 3 legged semiconductor - probably a FET -
Number is ST W55403 (then Morocco P4nK602 FP0) it's an all plastic device -
ferrite sleeve on its middle leg. It has 320 DC volts on its middle leg -
nothing on either of the other 2 legs.
Windings of the transformer show continuity.
I'd like to get this working again - I've been quoted $55 for a replacement.
Pic of the board can be seen at:
http://members.ozemail.com.au/~pebarhug/images/d-linkpsu.jpg
TIA
Peter

 

[[email protected]]

Peter:
Dust off your DMM or VOM and ESR meter... and test all the discrete
components like diodes, transistors which include bipolar & FET,
resistors, electrolytics, etc. If all of that checks out then it is
probably the chip.
electricitym
..
..
..

 

[Franc Zabkar]

I have a couple of d-link wirless accesspoints on whose power supplies have
failed - they should deliver 5volts at 2 amps.
The power supply takes the form of a slim-line plug pack.
Inside the pack is a little switch mode power supply.
I'm looking for some tips on what to look for in repairing this item. there
is nothing obviously wrong on a visial inspection. Fuse is intact, no heat
stress on the board, no visually distressed components. Just nothing on the
output.
On the board there is a pwm chip - CM3842 - an 8 pin device. There is 8
volts on its Vcc pin (pin 7) - nothing appreciable on any other pins.

The figure of 8V seems low. The datasheet for the UC3842 stipulates an
UVLO of 16V during startup and 10V during run mode. I'd check whether
the resistor that feeds Vcc is open or high.

- Franc Zabkar

 

[Peter]

The figure of 8V seems low. The datasheet for the UC3842 stipulates an
UVLO of 16V during startup and 10V during run mode. I'd check whether
the resistor that feeds Vcc is open or high.

- Franc Zabkar
--

Thanks Franc
resistor feeding it is 300K - and is on value. There is a zener from that
pin to ground - it's not an 8 volt zener though - I've checked it with a
current limited psu - at 8 volts it doesn't trigger. it's not open circuit
On it is just has the number 48

Peter

 

[Arfa Daily]

Thanks Franc
resistor feeding it is 300K - and is on value. There is a zener from that
pin to ground - it's not an 8 volt zener though - I've checked it with a
current limited psu - at 8 volts it doesn't trigger. it's not open
circuit
On it is just has the number 48

Peter

I was going to make exactly the same suggestion. 8v is definitely low for a
startup on a 3842. It's unusual to have a zener at that point. Normally, it
just relies on the current drawn by the chip and the value of the startup
resistor, to provide some kind of arbitrary voltage of about 12-14v. Once
the supply is running, a self regulated run voltage is provided from the
transformer. It is normal, however, to have a series diode at the pin, so
that when the run voltage comes up, this diode is blocked, so current ceases
to be drawn via the startup resistor. It may be that you've not followed the
circuit quite right at this point, and it is the series diode you're looking
at, and the " 48 " on it refers to it being a good old 1N4148.

Anyway, if the starup resistor's value is correct, then the next immediate
( and very common ) suspicion, is the decoupling cap for the supply pin.
It's usually quite a small value. 1 - 10uF would be typical. Check it with
an ESR meter. Capacitance value is unlikely to reveal anything amiss with
it. Failing that, just replace it.

Arfa

 

[Peter]

I was going to make exactly the same suggestion. 8v is definitely low for
a startup on a 3842. It's unusual to have a zener at that point. Normally,
it just relies on the current drawn by the chip and the value of the
startup resistor, to provide some kind of arbitrary voltage of about
12-14v. Once the supply is running, a self regulated run voltage is
provided from the transformer. It is normal, however, to have a series
diode at the pin, so that when the run voltage comes up, this diode is
blocked, so current ceases to be drawn via the startup resistor. It may be
that you've not followed the circuit quite right at this point, and it is
the series diode you're looking at, and the " 48 " on it refers to it
being a good old 1N4148.

Anyway, if the starup resistor's value is correct, then the next immediate
( and very common ) suspicion, is the decoupling cap for the supply pin.
It's usually quite a small value. 1 - 10uF would be typical. Check it with
an ESR meter. Capacitance value is unlikely to reveal anything amiss with
it. Failing that, just replace it.

Arfa
It's definitely a zener diode - ZD1 marked on the board

I've looked at a know good plugpack and compatred voltages - on the working
one there is 8.9 volts on pin 7.
I've swapped chips and the known good unit works with both pwm ICs
In the working unit there is 2.4 volts on pin 2 and 2 volts on pin 1
In the faulty unit there is no voltage on either pin.

I found a faulty 0.01 ceramic cap between pin 2 and 4 of the CM3842 on the
faulty unit. Replacement fixed the problem.
The unit uses a PC 817 IC to feedback some of the output to the pwm chip

Thanks all

Peter

 

[Sam Goldwasser]

Peter:
Dust off your DMM or VOM and ESR meter... and test all the discrete
components like diodes, transistors which include bipolar & FET,
resistors, electrolytics, etc. If all of that checks out then it is
probably the chip.
electricitym

Bad solder joints or high ESR electrolytic caps, chopper. Everything else is
much lower probability.

See the SMPS repair guides at the site below.

--- sam | Sci.Electronics.Repair FAQ: http://www.repairfaq.org/
Repair | Main Table of Contents: http://www.repairfaq.org/REPAIR/
+Lasers | Sam's Laser FAQ: http://www.repairfaq.org/sam/lasersam.htm
| Mirror Sites: http://www.repairfaq.org/REPAIR/F_mirror.html

Important: Anything sent to the email address in the message header above is
ignored unless my full name AND either lasers or electronics is included in the
subject line. Or, you can contact me via the Feedback Form in the FAQs.

 

[Ross Herbert]

I have a couple of d-link wirless accesspoints on whose power supplies have
failed - they should deliver 5volts at 2 amps.
The power supply takes the form of a slim-line plug pack.
Inside the pack is a little switch mode power supply.
I'm looking for some tips on what to look for in repairing this item. there
is nothing obviously wrong on a visial inspection. Fuse is intact, no heat
stress on the board, no visually distressed components. Just nothing on the
output.
On the board there is a pwm chip - CM3842 - an 8 pin device. There is 8
volts on its Vcc pin (pin 7) - nothing appreciable on any other pins.
On there board there is also a 3 legged semiconductor - probably a FET -
Number is ST W55403 (then Morocco P4nK602 FP0) it's an all plastic device -
ferrite sleeve on its middle leg. It has 320 DC volts on its middle leg -
nothing on either of the other 2 legs.
Windings of the transformer show continuity.
I'd like to get this working again - I've been quoted $55 for a replacement.
Pic of the board can be seen at:
http://members.ozemail.com.au/~pebarhug/images/d-linkpsu.jpg
TIA
Peter

I have just spent a couple of days trying to fix an almost identical
SMPS 6V plugpack but so far without success. It also had no physical
signs of component failure and the standard components each measured
ok. The PWM chip in mine was the 3843 but same pinout as the 3842. I
even stripped down the board and individually measured/tested all caps
resistors diodes opto-coupler and semiconductors. The 3843 worked fine
when substituted into a known good SMPS. The only possible faulty
component was a 7.2V zener connected directly across the DC output
which had gone leaky or changed specs. I then drew out the schematic
and rebuilt the lot with a new zener and got it to work a bit. It
produced 6V unloaded but this sagged when any load at all was applied.

Am almost at the stage where I have to admit this has me beat but I
don't like letting this happen.

BTW, you can get an identical SMPS from Jaycar (MP3140) for AUD24.95.
You don't need to pay $55.
http://www.jaycar.com.au/
type MP3140 into the search window top left.

 

[Ross Herbert]

Thanks Franc
resistor feeding it is 300K - and is on value. There is a zener from that
pin to ground - it's not an 8 volt zener though - I've checked it with a
current limited psu - at 8 volts it doesn't trigger. it's not open circuit
On it is just has the number 48

Peter

The unit I was playing with also used a 300K from Vbulk - pin 7 and
the startup voltage was 8.6V which is ok for the 3843 but not the
3842. I suspect that your psu should also be fitted with the 3843
instead of the 3842.

 

[Jeff Liebermann]

I have a couple of d-link wirless accesspoints on whose power supplies have
failed - they should deliver 5volts at 2 amps. (...)
I'd like to get this working again - I've been quoted $55 for a replacement.

Such 5V 2A or 3A switching power supplies can be found for much less
than $55. See:
| http://www.allelectronics.com/cgi-bin/item/PS-5372/
| http://www.allelectronics.com/cgi-bin/item/PS-537/
| http://www.allelectronics.com/cgi-bin/item/PS-523/
| http://www.allelectronics.com/cgi-bin/item/PS-528/

 

[Ross Herbert]

I've looked at a know good plugpack and compatred voltages - on the working
one there is 8.9 volts on pin 7.
I've swapped chips and the known good unit works with both pwm ICs
In the working unit there is 2.4 volts on pin 2 and 2 volts on pin 1
In the faulty unit there is no voltage on either pin.

I found a faulty 0.01 ceramic cap between pin 2 and 4 of the CM3842 on the
faulty unit. Replacement fixed the problem.
The unit uses a PC 817 IC to feedback some of the output to the pwm chip

Thanks all

Peter

I should have read your last post first. It seems you have fixed the
problem.

 

[Peter]

The unit I was playing with also used a 300K from Vbulk - pin 7 and
the startup voltage was 8.6V which is ok for the 3843 but not the
3842. I suspect that your psu should also be fitted with the 3843
instead of the 3842.

Hi Ross - did you see my pic of the board? Is yours similar?

 

[Peter]

Such 5V 2A or 3A switching power supplies can be found for much less
than $55. See:
| http://www.allelectronics.com/cgi-bin/item/PS-5372/
| http://www.allelectronics.com/cgi-bin/item/PS-537/
| http://www.allelectronics.com/cgi-bin/item/PS-523/
| http://www.allelectronics.com/cgi-bin/item/PS-528/

Many thanks Jeff

here in OZ we have to pay a lot more for things - also the A$55 was for a
genuine d-link replacement - rip off!

Peter

 

[Dave D]

I have just spent a couple of days trying to fix an almost identical
SMPS 6V plugpack but so far without success. It also had no physical
signs of component failure and the standard components each measured
ok. The PWM chip in mine was the 3843 but same pinout as the 3842. I
even stripped down the board and individually measured/tested all caps
resistors diodes opto-coupler and semiconductors. The 3843 worked fine
when substituted into a known good SMPS. The only possible faulty
component was a 7.2V zener connected directly across the DC output
which had gone leaky or changed specs. I then drew out the schematic
and rebuilt the lot with a new zener and got it to work a bit. It
produced 6V unloaded but this sagged when any load at all was applied.

Am almost at the stage where I have to admit this has me beat but I
don't like letting this happen.

Forget about testing the capacitors, replace them. This sounds like a
classic SPMS electrolytic capacitor problem to me. There is no convenient
100% conclusive and reliable way to test electrolytic capacitors IMO,
especially when they are used in an abusive environment like an SMPS.

Try warming up each cap in turn and see if it restores normal operation
temporarily, that should pinpoint which cap is weak.

Dave

 

[Arfa Daily]

Forget about testing the capacitors, replace them. This sounds like a
classic SPMS electrolytic capacitor problem to me. There is no convenient
100% conclusive and reliable way to test electrolytic capacitors IMO,
especially when they are used in an abusive environment like an SMPS.

Try warming up each cap in turn and see if it restores normal operation
temporarily, that should pinpoint which cap is weak.

Dave

Gotta disagree with you on that one Dave ! I don't think I can recall ever
having had an electro that was causing trouble on a switcher, that didn't
show bad on my trusty ESR meter, which is definitely convenient to use, and
in my opinion totally conclusive when you've learnt to correctly interpret
what it's telling you. In that respect, I suppose, it is a bit of a black
art, so I might concede 100% conclusive for those people not using an ESR
meter all the time ;~}

Arfa

 

[Dave D]

Gotta disagree with you on that one Dave ! I don't think I can recall ever
having had an electro that was causing trouble on a switcher, that didn't
show bad on my trusty ESR meter, which is definitely convenient to use,
and in my opinion totally conclusive when you've learnt to correctly
interpret what it's telling you.

I disagree. A shorted or leaky electrolytic cap will pass an ESR test ;-) An
ESR meter is not an absolute indication of the serviceabilty of an
electrolytic cap. In many cases, the only 100% effective proof of a faulty
or marginal cap is to replace it. An ESR meter does not simulate the working
conditions of an SMPS capacitor, and an intermittant or marginal problem may
not arise when under test.

In that respect, I suppose, it is a bit of a black art, so I might concede
100% conclusive for those people not using an ESR meter all the time ;~}

I sort of agree with that, I'd say that combined with other techniques, an
ESR meter is a very useful tool indeed. However, the OP has reached a dead
end and IMO the most likely culprit is capacitors, regardless of what his
ESR meter is saying.

Dave





Dave

 

[Ross Herbert]

Hi Ross - did you see my pic of the board? Is yours similar?

Sorry Peter, I didn't see it before now. And yes, mine is exactly the
same in layout and components.

 

[Arfa Daily]

I disagree. A shorted or leaky electrolytic cap will pass an ESR test ;-)
An ESR meter is not an absolute indication of the serviceabilty of an
electrolytic cap. In many cases, the only 100% effective proof of a faulty
or marginal cap is to replace it. An ESR meter does not simulate the
working conditions of an SMPS capacitor, and an intermittant or marginal
problem may not arise when under test.


I sort of agree with that, I'd say that combined with other techniques, an
ESR meter is a very useful tool indeed. However, the OP has reached a dead
end and IMO the most likely culprit is capacitors, regardless of what his
ESR meter is saying.

Dave

All academic now Dave, as the OP has posted his final finding, which was a
defective 0.01uF ceramic cap across two pins of the 3842 ( posted on 20th
oddly, the same as the original post ) so neither your technique of blitzing
all the electros, or mine of applying an ESR meter and a bit of black magic,
would have found it anyway !!

Arfa

 

[Dave D]

All academic now Dave, as the OP has posted his final finding, which was
a defective 0.01uF ceramic cap across two pins of the 3842 ( posted on
20th oddly, the same as the original post ) so neither your technique of
blitzing all the electros, or mine of applying an ESR meter and a bit of
black magic, would have found it anyway !!

Arfa

Arfa, I was replying to Ross Herbert, not the OP! Check out his post on
21/05/2006 at 03:33.

Dave

 

[Tom MacIntyre]

Arfa, I was replying to Ross Herbert, not the OP! Check out his post on
21/05/2006 at 03:33.

Dave

The thread breakdown in my newsreader indicates otherwise...

Tom