Searching \ for 'Switch Mode supply reliability [Was FAQ pointer]' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: www.piclist.com/techref/index.htm?key=
Search entire site for: 'Switch Mode supply reliability [Was FAQ pointer]'.

Truncated match.
PICList Thread
'Switch Mode supply reliability [Was FAQ pointer]'
1997\08\11@031222 by Mike

flavicon
face
At 04:19 PM 8/10/97 -0400, you wrote:

>FWIW, my primary desktop just needed a new power supply.  A transistor blew
>after 5+ years of being on almost continuously.  Anybody out there ever
>hear of "inrush current" and the damage it does?!?!?!

Almost every switch mode supply I've seen use Barium Titanate inrush
current limiters - these are great devices - we used them on the output
of a handheld inverter they protected the transistors very well indeed.

I still would not ever leave a switchmode supply unattended - especially
knowing that even switchmode transistors are susceptible to static
damage during assembly and you never know when a fire could be caused.

So if I wished to have a PC in a unattended environment I'd design my
own supply with a heck of a lot of safeguards...

Rgds

mike
Perth, Western Australia

1997\08\11@074914 by Kalle Pihlajasaari

flavicon
face
Hi Mike,

> >FWIW, my primary desktop just needed a new power supply.  A transistor blew
> >after 5+ years of being on almost continuously.  Anybody out there ever
> >hear of "inrush current" and the damage it does?!?!?!
>
> Almost every switch mode supply I've seen use Barium Titanate inrush
> current limiters - these are great devices - we used them on the output
> of a handheld inverter they protected the transistors very well indeed.

They are a serious liability if you have a 0.5 to 2 second drop out.
The typically very effective NTC thermistors that go close to short circuit
a few seconds after switch on.  Problem is when they are hot/short-circuit
and you have a dip, the primary switcher capacitor discharges and when the
power comes back the inrush current is now NOT moderated by the
thermistor and relies on the slo-blo of the fuse not to rupture.

Some switchers also enter a protection mode with dips of 0.1 to 0.5
seconds when the voltage does not reach zero and allow the switcher
IC to properly reset (perrenial PIC and other micro problem) and I have
had an occasion where 5 out of 30 computers latched up after one
particular dip.  Power cycling them clears the fault.

> I still would not ever leave a switchmode supply unattended - especially
> knowing that even switchmode transistors are susceptible to static
> damage during assembly and you never know when a fire could be caused.

There are hundresds of thousands of switchers that are unatended
all over the world in all sorts of industries, The Internet lives
on them and I would say that at least half of them are unatended
overnight and weekends.  I think you are beeing a bit too cautious
unless you have a national monument that cannot handle any risk of
fire.

We have some 30 PCs and they are on all the time and most of them
have not been even reset since the UPS was installed about 3 months
ago.  We run DOS, Win3.1, NT, Linux, SCO. Under heavy load the
Linux boxes do have resource leaks and need 60 day restarts, under
light load one cannot tell.  When running NT with non Microsoft
applications one sometimes also has resource leaks but stopping
and starting of offending services is sufficient to clear the problem.
The SCO box runs without reset unless The disk fills up with log
files, then I reboot it after making space.  My Win4W 3.11 crashes
at least 5 times per week, normally when using a web browser or
graphics application, I expect this and save before loading images.

> So if I wished to have a PC in a unattended environment I'd design my
> own supply with a heck of a lot of safeguards...

What I find is that the FAN is very important.  I have had switchers
fail in service but only those that do not have a fan or if the
fan has failed.  Dust is the main enemy of a fan.

Cheers
--
Kalle Pihlajasaari   spam_OUTkalleTakeThisOuTspamip.co.za   http://www.ip.co.za/ip
Interface Products   P O Box 15775, DOORNFONTEIN, 2028, South Africa
+ 27 (11) 402-7750   Fax: 402-7751    http://www.ip.co.za/people/kalle

DonTronics, Silicon Studio and Wirz Electronics uP Product Dealer

1997\08\11@124427 by Mike

flavicon
face
At 10:51 AM 8/11/97 +0200, you wrote:

>They are a serious liability if you have a 0.5 to 2 second drop out.
>The typically very effective NTC thermistors that go close to short circuit
>a few seconds after switch on.  Problem is when they are hot/short-circuit
>and you have a dip, the primary switcher capacitor discharges and when the
>power comes back the inrush current is now NOT moderated by the
>thermistor and relies on the slo-blo of the fuse not to rupture.

Good point. So I don't cycle the power on my systems within 60 seconds.
And where possible I'd use a UPS in the event the local kids throw
steel cable up onto those power poles - Yuk !

>There are hundresds of thousands of switchers that are unatended
>all over the world in all sorts of industries, The Internet lives
>on them and I would say that at least half of them are unatended
>overnight and weekends.  I think you are beeing a bit too cautious
>unless you have a national monument that cannot handle any risk of
>fire.

True - my main system is at my home, thats my monument - see my other
post in respect of sweat shop manufacture etc. If I had to fit out a lab
which NEEDED unattended operation then I'd quite happily manage the
setup to avoid a fire in the rare event of ignition and there are lots
of neat ways to approach this goal etc.

>> So if I wished to have a PC in a unattended environment I'd design my
>> own supply with a heck of a lot of safeguards...
>
>What I find is that the FAN is very important.  I have had switchers
>fail in service but only those that do not have a fan or if the
>fan has failed.  Dust is the main enemy of a fan.

I agree - I don't like Fans much - if I had to use one in my design, then
it would be in an 'open region' where dust could not accumulate - ie the
standard upside down smooth surface approach. The rest of the cooling I'd
arrange as far as possible to be convection, radiated, transmissive to
that large heatsink of a case. I've got some great large enclosure
designs using hefty aluminium extrusions for an inverter currently
being evaluated - no pesky continuous rated fans in that setup...

Rdgs

Mike
Perth, Western Australia

1997\08\11@164650 by Russell McMahon

picon face
> >FWIW, my primary desktop just needed a new power supply.  A transistor
blew
> >after 5+ years of being on almost continuously.  Anybody out there ever
> >hear of "inrush current" and the damage it does?!?!?!
>
> Almost every switch mode supply I've seen use Barium Titanate inrush
> current limiters - these are great devices - we used them on the output
> of a handheld inverter they protected the transistors very well indeed.
>
> I still would not ever leave a switchmode supply unattended - especially
> knowing that even switchmode transistors are susceptible to static
> damage during assembly and you never know when a fire could be caused.
>
> So if I wished to have a PC in a unattended environment I'd design my
> own supply with a heck of a lot of safeguards...
>
> Rgds
>
> mike
> Perth, Western Australia
> And others

I don't want to get into the OS system comparisons that some have been
making but I'm another who leaves a PC or 2 running continually. One has
now run for 4.5 years continuous (it acts as a facsimile receiver as well
as being used as a workhorse desktop) running Windows 3.1 with 8MB of RAM.
The power supply expired quietly after about 3.5 years and was swapped for
a new one. At the price (about $NZ40 including a case (approx $US23?) its
hardly worth repairing them. The biggest failure item I have seen is the
mains filter electrolytics which generally go out with a bang and split
longitudinally and blow their contents around the inside of the power
supply. As this country is universally 230 VAC and the supplies are usually
switchable 110 or 230 V ours are run in a different mode from most US ones.
On 110 AC the input electrolytics are run as a voltage doubler but on
230VAC they are simply placed in series across the rectified mains. I would
have thought that the voltage doubling mode would be harder on
electrolytics and lead to more failures. In my experience switching
transistor failures are less common. Even if a transistor does fail I think
it would be unusual for it to fail in such a way as to damage connected
equipment. of course, sudden loss of DC power can lead to damage or data
loss depending on configuration.

Regarding reliability of the system with time when run continually. I feel
that having the fax software in the background reduces the time the system
can be run before rebooting. Certainly, doing strange background tasks in
DOS can hasten the need for a reboot. Generally, a week is about the
longest I would expect to go before Windows started reporting low memory or
misbehaving. I have a log of startup times so I can check instantly how
long the system has been up. A Windows 95 machine (16MB RAM), which is not
running fax in the background, but which also runs continuously, seems more
stable than the 3.1 PC and much less prone to low memory messages after a
time so I suspect I could go several weeks at a time between reboots in
many cases. In practice I try to reboot every few days as I don't like to
risk losing a valuable work session on this PC due to unexpected problems.

No doubt some other OS's are much more stable but both of these systems
seem to be entirely adequate for practical purposes.

1997\08\11@174514 by Eric van Es

flavicon
face
Kalle Pihlajasaari wrote:
>
> What I find is that the FAN is very important.  I have had switchers
> fail in service but only those that do not have a fan or if the
> fan has failed.  Dust is the main enemy of a fan.
>
> Cheers
> --
> Kalle Pihlajasaari   .....kalleKILLspamspam@spam@ip.co.za   http://www.ip.co.za/ip
> Interface Products   P O Box 15775, DOORNFONTEIN, 2028, South Africa
> + 27 (11) 402-7750   Fax: 402-7751    http://www.ip.co.za/people/kalle

Yip - I agree about the fan. I my PSU had a very noisy fan - sounded
like a bearing (grinding noise) - although I doubt whether they actually
have real bearings.

Replaced the entire mini tower - now I'm stuck with one I hate...
--
Eric van Es               | Cape Town, South-Africa
Mailto:vanesspamKILLspamilink.nis.za | WWW: http://www.nis.za/~vanes/
TEMPORARY/HOLIDAY ACCOMMODATION? http://www.nis.za/~vanes/accom.htm

1997\08\11@175322 by Mike Keitz

picon face
On Mon, 11 Aug 1997 23:55:38 +1200 Russell McMahon <.....apptechKILLspamspam.....CLEAR.NET.NZ>
writes:

>The power supply expired quietly after about 3.5 years and was swapped
>for
>a new one. At the price (about $NZ40 including a case (approx $US23?)
>its
>hardly worth repairing them.

Repair is largely an academic pursuit unless it is a unit with an unusual
physical configuration.  Even then, the board from a new unit can often
be fitted into the unusual case.

The biggest failure item I have seen is
>the
>mains filter electrolytics which generally go out with a bang and
>split
>longitudinally and blow their contents around the inside of the power
>supply.

They are supposed to "vent" quietly by blowing out either a thin place in
the rubber plug at the bottom or a scored thin spot in the top of the
can.  If the pressure rose rapidly enough to make the side blow out it is
probably because the other capacitor became shorted and suddently exposed
the one that exploded to the full line voltage.  Or something else pulled
the midpoint of the two capacitors out of balance.

As this country is universally 230 VAC and the supplies are
>usually
>switchable 110 or 230 V ours are run in a different mode from most US
>ones.
>On 110 AC the input electrolytics are run as a voltage doubler but on
>230VAC they are simply placed in series across the rectified mains. I
>would
>have thought that the voltage doubling mode would be harder on
>electrolytics and lead to more failures.

The series connection has the serious problem that the voltage
distribution between the two capacitors is not well controlled.   In the
120 V configuration it is impossible for either capacitor to charge to
more than the peak line voltage.  In the 240V mode the center point of
the capacitors is either open or connected to a blocking capacitor to one
end of the power transformer.  This blocking capacitor is necessary in
the 120 V mode to prevent DC from flowing in the transformer primary if
the positive and negative supply rails or the driver duty cycle isn't
balanced.  In 240V operation it may work to remove it and let the almost
balanced duty cycle of the drive transistors force the center point of
the capacitors to nearly the halfway point.  But, without the blocking
capacitor, a (single) shorted power transistor would make the opposite
filter capacitor explode.

Typically there are large 100K resistors in parallel with each capacitor
but I think these are mostly for safety reasons rather than voltage
balance.  They will leak the charge out of the capacitors in case some
fool ignores the dire warning on the label and takes the supply apart.
(The fool would have to be smart enough to disconnect the line cord of
course).

In my experience switching
>transistor failures are less common. Even if a transistor does fail I
>think
>it would be unusual for it to fail in such a way as to damage
>connected
>equipment. of course, sudden loss of DC power can lead to damage or
>data
>loss depending on configuration.

In the US with 120V operation, filter capacitor failure is uncommon.
Lightning or surges can take out the fuse, input diodes and/or the power
transistors, usually in that order as the severity increases.  A lot of
lightning will blow them right off the board (Parts unit!).  Most
failures though are minor.  Many have no problem other than a bad switch,
bad connection, or worn-out fan.  IBM spec'd Panasonic ball-bearing fans
which I have never seen wear out.  Two other trouble spots are the
startup resistors (usually 330K) on the power transistors and the output
capacitors.

A particularly nasty failure mode results from the diode that taps power
for the control and driver circuit failing shorted.  This places a
capacitor where there shouldn't be one and causes the +12V output to soar
to about +26V, while the 5V line stays right at 5V.  Very bad for hard
drives.  There is no overvoltage protection in any of these units, even
the major brand ones.  The only protective circuit per se is a current
sensor in series with the input of the power transformer.  This is
usually included in the very cheap units as well.  It appears that they
are all built to the same application note somewhere.

1997\08\11@190245 by nino.benci

flavicon
picon face
To add to the discussion, I have some some cheap Korean / asian ???,
power supplies built for the Australian market here in my workshop
with failed flyback snubber protection for the main stiching
transistor / fet.

As an academic exercise i calculated the peak flyback curent for this
snubber design (app note design most likely) and discovered to my
horror that the units were designed to operate with a maximum of
120VAC input. What the supplier had done was to series wire the
input caps for 240VAC and not bother to check anything else.
Therefore the snubber network was under-rated. End result over a
period of time was stressed components and eventual failure.

Nino.
******************************************************
* Antonio (Nino) Benci                               *
* Chief Technical Officer                            *
* Monash University - Dept of Physics                *
* Wellington Rd, Clayton. 3168                       *
* Victoria, Australia.                               *
* TEL - 61 3 9905 3649, FAX - 61 3 9905 3637         *
* EMAIL - EraseMEnino.bencispam_OUTspamTakeThisOuTsci.monash.edu.au               *
* WWW - http://www.physics.monash.edu.au/services/ews.html   *
******************************************************

1997\08\12@020412 by nigelg

flavicon
picon face
In message  <3.0.1.16.19970811152933.3ba77f02spamspam_OUTmail.wantree.com.au>> @spam@PICLISTKILLspamspamMITVMA.MIT.EDU writes:
> Almost every switch mode supply I've seen use Barium Titanate inrush
> current limiters - these are great devices - we used them on the output
> of a handheld inverter they protected the transistors very well indeed.
>
> I still would not ever leave a switchmode supply unattended - especially
> knowing that even switchmode transistors are susceptible to static
> damage during assembly and you never know when a fire could be caused.
>
> So if I wished to have a PC in a unattended environment I'd design my
> own supply with a heck of a lot of safeguards...

This seems to be more 'scare-mongering' than anything else, I've repaired
many thousands of switch-mode PSU's over the years - including the worlds
first domestic use of an SMPSU, the Thorn 3000 colour TV. During all this
time I've never seen an SMPSU cause a fire, certainly in the UK units are
subjected to stringent safety tests to avoid such an occurance.

Having seen mains plugs from Australia and the USA, I would think they
are a lot more likely to give rise to spurious fires than faults in domestic
equipment :-).

During this same period I have however seen quite a few houses struck by
lightning, certainly into double figures in the relatively small service
area we cover. Now this does make a mess of an SMPSU :-).

I would certainly consider the likelyhood of your house been struck by
lightning many times greater than an SMPSU burning it down.

Nigel.

       /--------------------------------------------------------------\
       | Nigel Goodwin   | Internet : KILLspamnigelgKILLspamspamlpilsley.demon.co.uk     |
       | Lower Pilsley   | Web Page : http://www.lpilsley.demon.co.uk |
       | Chesterfield    |                                            |
       | England         |                                            |
       \--------------------------------------------------------------/

1997\08\12@085251 by Mike Smith

flavicon
face
-----Original Message-----
From: Nigel Goodwin <RemoveMEnigelgTakeThisOuTspamlpilsley.demon.co.uk>
To: spamBeGonePICLISTspamBeGonespamMITVMA.MIT.EDU <TakeThisOuTPICLISTEraseMEspamspam_OUTMITVMA.MIT.EDU>
Date: Tuesday, 12 August 1997 15:35
Subject: Re: Switch Mode supply reliability [Was FAQ pointer]

<snip>

>This seems to be more 'scare-mongering' than anything else, I've repaired
>many thousands of switch-mode PSU's over the years - including the worlds
>first domestic use of an SMPSU, the Thorn 3000 colour TV. During all this
>time I've never seen an SMPSU cause a fire, certainly in the UK units are
>subjected to stringent safety tests to avoid such an occurance.

I've seen one in 'el cheapo' monitor(brand was Teco) - at a TAFE (college to
some)  Caused a fair mess.  Mike will no doubt chastise me for my
inconsistency here (leaving CPU on, and letting 'green' monitor sleep, but
it is a name brand monitor)

>
>Having seen mains plugs from Australia and the USA, I would think they
>are a lot more likely to give rise to spurious fires than faults in
domestic
>equipment :-).
>

Hey, leave our plugs alone!  They're nicer than some of the UK monstrosities
I've seen.  Tackle the UK one's if you must (after all, at a lower voltage,
they need to support higher current...)

The monitor fire originated in the switch, from memory.

>During this same period I have however seen quite a few houses struck by
>lightning, certainly into double figures in the relatively small service
>area we cover. Now this does make a mess of an SMPSU :-).
>
>I would certainly consider the likelyhood of your house been struck by
>lightning many times greater than an SMPSU burning it down.

Esp in some of the more storm-prone areas.

MikeS
<RemoveMEmikesmith_ozspamTakeThisOuTrelaymail.net>

1997\08\12@091937 by Mike

flavicon
face
At 09:54 PM 8/12/97 +0930, you wrote:

>I've seen one in 'el cheapo' monitor(brand was Teco) - at a TAFE (college to
>some)  Caused a fair mess.  Mike will no doubt chastise me for my
>inconsistency here (leaving CPU on, and letting 'green' monitor sleep, but
>it is a name brand monitor)

Actually that reminds me, there was a news report on OZ TV in the last
year or so about a fire caused by a monitor being left on in some attic
above a shop (around NSW somewhere) where someone slept - I remember they
died from smoke inhalation.

Although its rare - turning the monitor off before going to sleep might
have been helpful on that particular occasion.

Rgds

Mike
Perth, Western Australia

More... (looser matching)
- Last day of these posts
- In 1997 , 1998 only
- Today
- New search...