Power Supply keep blowing

[Blue Rival]

Hi Everyone - I have a home build system and have had trouble with
blowing power supplies. My problem started when I moved in to a new
apartment with crappy power. I had determined that at least and
purchased a decent APC Surge Protector. This seamed to solve my problem
for a while but over the past 8 months I have had 4 power supplies blow
on me. All where basically cheep - 2 were 400W one was 450W and then
most recently blew out a 500W. I do not think it is power because
nothing else plugged into the APC device has shorted out or blown. It
is strange because the power supplies will run fine for a while, 1-2
months with no problems then all of the sudden the PC shuts off and I
can smell them burning. When the 420W blew there was a flash bang and a
lot of smoke. The 500W I most recently put in only lasted about 3
weeks.

I think the problem is from within my computer itself. I have looked
for thinks that might be shorting out the MOBO but have not found
anything. Does anyone else have any ideas? I have thought about buying
a really good power supply but am reluctant to buy a good one that will
also blow because of a problem somewhere else in the system.

Luckily so far none of my other components have fried.
Please help!

Basic Configuration
Gigabyte &VT600P-RZ MOBO
AMD Athalon XP 1600
1GB DDR3200 (2x 512 chips)
200 GB HD
80 GB HD
GFORCE 6600GTOC (BFG)
DVD ROM (CDRW)
2 case fans
Wireless NIC
Game Theatre XP sound card

Game Theatre XP sound card

 

[John P. Dearing]

Blue Rival wrote:

-=[ deletia about power supplies frequently failing ]=-

Basic Configuration
Gigabyte &VT600P-RZ MOBO
AMD Athalon XP 1600
1GB DDR3200 (2x 512 chips)
200 GB HD
80 GB HD
GFORCE 6600GTOC (BFG)
DVD ROM (CDRW)
2 case fans
Wireless NIC
Game Theatre XP sound card

This hardware doesn't look to be all that power hungry, so unless you've
got some really defective hardware, I tend to discount the stuff in the
box as the cause of your problems.

I believe that the "new apartment with crappy power" is probably the
ultimate root cause.

Get a Digital Multimeter. Set it on ACV and measure the AC voltage at
your outlet. My guess is that it is probably too high. Anything much
over 120 VAC is probably too high. Higher than normal voltage will
stress the components in the power supply.

If the voltage is too low that can stress the power supply too.

In either case, you need more than just a good surge surpressor, you
need a voltage regulating transformer like the Line-R series from APC.
There are others from companies like Tripp-Lite, etc...

What they do is for a large change in input voltage generate a small
change in output voltage. If the volatage is low, it will b boosted, if
the voltage is too high, it will be cut back.

In an A+ class I taught, we had some systems that we were having all
kinds of problems with, some would boot and some wouldn't. It turned out
the voltage in the classroom was too high (nearly 130VAC) and some of
the power supplies were immediately shutting down for overvoltage or
would simply refuse to start (voltage was too high). I got hold of a
voltage regulating transformer and all the systems booted just fine. We
let the Plant Maintenance dept know of the problem and it was fixed in a
few days.

Try checking your voltage, I suspect that it is not what it should be.

Cheers!

John

 

[David Maynard]

Hi Everyone - I have a home build system and have had trouble with
blowing power supplies. My problem started when I moved in to a new
apartment with crappy power. I had determined that at least and
purchased a decent APC Surge Protector. This seamed to solve my problem
for a while but over the past 8 months I have had 4 power supplies blow
on me. All where basically cheep - 2 were 400W one was 450W and then
most recently blew out a 500W. I do not think it is power because
nothing else plugged into the APC device has shorted out or blown. It
is strange because the power supplies will run fine for a while, 1-2
months with no problems then all of the sudden the PC shuts off and I
can smell them burning. When the 420W blew there was a flash bang and a
lot of smoke. The 500W I most recently put in only lasted about 3
weeks.

'Cheap' power supplies almost never put out as much power the EMBLAZONED
ACROSS THE BOX 'watt' label suggests. It's an utterly meaningless 'peak
power' number. Uh huh. 'Peak' for how long? 12 nanoseconds?

I think the problem is from within my computer itself. I have looked
for thinks that might be shorting out the MOBO but have not found
anything. Does anyone else have any ideas? I have thought about buying
a really good power supply but am reluctant to buy a good one that will
also blow because of a problem somewhere else in the system.

A proper power supply should not 'blow' regardless. Over current
protection, over voltage protection, or thermal shutdown should kick in.

I suspect your previous 'cheap' power supplies could not handle the 12 volt
drain, overheated, and self destructed as a result of it.

 

[John Weiss]

Hi Everyone - I have a home build system and have had trouble with
blowing power supplies. My problem started when I moved in to a new
apartment with crappy power. I had determined that at least and
purchased a decent APC Surge Protector.

I suspect your "cheap" power supplies can't hack low voltage from your wall
socket.

A surge protector isn't going to do squat for you. You need a "line
conditioning" UPS that will maintain the line voltage to reasonably tight
tolerances, plus a good power supply that will hack any variations.

You probably already spent more $$ on 5 "cheap" power supplies than 1 good Antec
or equivalent would cost. If you look hard, you can find a reasonable Belkin
UPS that won't cost much more than the "decent" but useless surge protector...

 

[John Doe]

Blue Rival wrote:

-=[ deletia about power supplies frequently failing ]=-

Basic Configuration Gigabyte &VT600P-RZ MOBO AMD Athalon XP 1600
1GB DDR3200 (2x 512 chips) 200 GB HD 80 GB HD GFORCE 6600GTOC
(BFG) DVD ROM (CDRW) 2 case fans Wireless NIC Game Theatre XP
sound card

This hardware doesn't look to be all that power hungry, so unless
you've got some really defective hardware, I tend to discount the
stuff in the box as the cause of your problems.

I believe that the "new apartment with crappy power" is probably
the ultimate root cause.

Get a Digital Multimeter. Set it on ACV and measure the AC voltage
at your outlet. My guess is that it is probably too high. Anything
much over 120 VAC is probably too high. Higher than normal voltage
will stress the components in the power supply.

If the voltage is too low that can stress the power supply too.

In either case, you need more than just a good surge surpressor,
you need a voltage regulating transformer like the Line-R series
from APC. There are others from companies like Tripp-Lite, etc...

What they do is for a large change in input voltage generate a
small change in output voltage. If the volatage is low, it will b
boosted, if the voltage is too high, it will be cut back.

In an A+ class I taught, we had some systems that we were having
all kinds of problems with, some would boot and some wouldn't. It
turned out the voltage in the classroom was too high (nearly
130VAC) and some of the power supplies were immediately shutting
down for overvoltage or would simply refuse to start (voltage was
too high). I got hold of a voltage regulating transformer and all
the systems booted just fine. We let the Plant Maintenance dept
know of the problem and it was fixed in a few days.

I am enjoying a 1200 watt voltage regulator on a high-end personal
computer. It might be overkill but maybe that enhances the
protection. I have experienced good stability over the last year or
so.

You can hear the relay(s) clicking on and off when power is
fluctuating badly. Most of the time it doesn't, but one afternoon
during a cold rain it was clicking every minute or so.

It is only about $60 (USD) and it can tell you whether bad house
current is the problem. If so, you did good. If not, at least you
know and maybe it will be useful in some other dwelling.

I like knowing what's going on with the electricity.

Good luck.

 

[larry moe 'n curly]

I have had 4 power supplies blow on me. All where basically cheep -

2 were 400W one was 450W and then most recently blew out a 500W.

AMD Athalon XP 1600
200 GB HD
80 GB HD
GFORCE 6600GTOC (BFG)
DVD ROM (CDRW)

Unless you're getting those cheapos for free, most likely a single
high-quality PSU would have cost you less, and Newegg.com and
Directron.com have excellent Fortron-Source-made units for just $25-45.
A 350W Fortron can put out more than many 500W cheapos. JonnyGuru
managed to get 500W from a 300W Fortron, but he fried a pair of 500W
Powmax Assassins at just 200W and 300W, and a 500W Deer PSU he tested
would shut down at 300W and couldn't start with a load of more than
150W.

If none of your other AC powered stuff blows, I'd suspect the PC PSUs,
but most PSUs are rated to work over 90-130VAC.

Don't buy a PSU that looks something like this on the inside ("580W"
Powmax):

http://static.flickr.com/43/73954319_bc8f9de6d0_o.jpg (photo by
JonnyGURU.com)

A good PSU will look heftier and more crowded inside:

http://static.flickr.com/38/75117436_1de7a3142d_b.jpg

This one looks beefier than the 580W Powmax, but it's rated for only
300W.

 

[DaveW]

If you keep blowing "cheap" PSU's, which is a common occurence among all
users, you might consider buying one more expensive PSU that will last.
Antec PSU's are the best selling and among the most conservatively rated
heavy duty supplies on the market. And with all your components I would
recommend at least the 480 watt True Power model. That's what I've used for
years.

 

[w_tom]

First can those power supplies even support the load? For
example, when marketing to 'bean counter' types, then inferior
supplies hype a 'watts' number that is technically accurate
but actually deceptive and not useful. IOW a power supply
rated at 350 or 400 watts may be equivalent to another that is
only 250 watts. Use of a 3.5 digit multimeter for but two
minutes would better answer a question , "Is it undersized?"
Meanwhile, if power supply manufacturer does not provide a
long list of written and numerical specifications, then
suspect a supply manufactured for 'bean counters' who then
sell home built machines hyping their lower costs.

With sufficient experience, then one might tell the
difference between heftier components. More likely, few to no
one can tell which power supply has heftier components and
which is using larger, less expensive and less powerful
components. Better to demand numbers you don't understand.
Maybe 1% know what those numbers means. But inferior supply
manufacturers fear to empower that 1%; therefore provide no
numerical specs. Written numerical specs will better suggest
a minimally sufficient supply.

Second, nothing inside a computer can damage a properly
constructed power supply. Furthermore, a power supply failure
must never cause computer component damage. If either failure
happens, well again, was it a sufficient supply or something
marketed to 'bean counter' types?

Third, anything accomplished by a plug-in protector or line
conditioner should already be inside that supply.
Unfortunately, myths promote these 'miracle plug-in solutions'
without first learning what those solutions even claim to
accomplish. If power supply is damaged, then no plug-in
protector or line conditioner is going to solve a problem that
is also damages bathroom GFCI, dishwasher, TV, smoke
detectors, or incandescent bulbs.

Fourth, a dead body exists. Best responses are provided
with specific details such as what inside a supply actually
failed. But then, this assumes a power supply is even
minimally acceptable (see above first point). Another
indication of an inferior supply is a full retail price of
less than $60. How to sell at lower price and higher profit?
Forget to provide other essential internal functions.

Fifth, numbers with that power supply should claim to
operate OK at voltages so high that incandescent bulbs burn
out
three times faster; at voltages so low that incandescent bulbs
glow at less than 40% intensity. Furthermore, low voltages
never cause a minimally acceptable supply to be damaged - as
was even industry standard required 30+ years ago.

 

[bryanlove]

Another indication of an inferior supply is a full retail price of
less than $60. How to sell at lower price and higher profit?
Forget to provide other essential internal functions.

All but the cheapest supplies contain circuitry meant to protect
against overloads, overvoltage, and undervoltage because the controller
ICs contain those functions, but the protection often doesn't work
correctly because the design wasn't debugged thoroughly enough or built
with adequate components.

High volume (1M-12M units/annum) and Chinese labor allow good supplies
to retail for under $60, and Fortron-Source Power proves this
constantly with some of the best products on the market. OTOH some
high priced supplies have low quality because their marketers emphasize
cosmetics (windows, glowing fans, sheathed cables) over functional
design

 

[w_tom]

If all but the cheapest supplies contain circuitry meant to
protect against overvoltage, then why do so many supplies not
have it? The Intel spec is quite blunt about this function
that was defacto standard 30 years ago:

The overvoltage sense circuitry and reference shall reside
in packages that are separate and distinct from the regulator
control circuitry and reference. No single point fault shall
be able to cause a sustained overvoltage condition on any or
all outputs. The supply shall provide latch-mode overvoltage
protection ...

How does one know overvoltage protection (OVP) is missing?
The 'separate and distinct' packages don't exist. If a power
supply does not specifically and in writing claim to have OVP,
then it's a sure bet, the overvoltage protection does not
exist. Missing OVP is common in cheapest supplies that only
cost $60 full retail - a very low price. To get prices lower,
essential functions must be 'forgotten'.

Did they forget an essential function? Well, where is that
function listed in numerical specs? If the function (ie
overvoltage protection) is not specifically listed, then it
probably is not provided. Those written specifications - what
they do and do not say - are damning facts. No long list of
written numerical specs? Then that supply may not have OVP
and a long list of other essential functions.

This specifically not possible - does not meet Intel specs
because it is not overvoltage protection:

... circuitry meant to protect against overloads, overvoltage,
and undervoltage because the controller ICs contain those
functions ...

One major reason for OVP: controller IC may fail. A failed
controller IC therefore IS a reason for separate OVP circuitry
- as Intel demands. Intel is blunt for good reason:

The overvoltage sense circuitry and reference shall reside
in packages that are separate and distinct from the regulator
control circuitry and reference.

One way to reduce that retail price below a so competitive
$60? Forget the OVP.

Why would power supplies keep failing? Without technical
details such as meter readings and what internal part is
damaged, then every answer is only speculation. But if that
supply is missing essential functions, then it also will not
provide that long list of written and numerical specs.

 

[bryanlove]

If all but the cheapest supplies contain circuitry meant to
protect against overvoltage, then why do so many supplies not
have it? The Intel spec is quite blunt about this function
that was defacto standard 30 years ago:

How does one know overvoltage protection (OVP) is missing?
The 'separate and distinct' packages don't exist. If a power
supply does not specifically and in writing claim to have OVP,
then it's a sure bet, the overvoltage protection does not
exist. Missing OVP is common in cheapest supplies that only
cost $60 full retail - a very low price. To get prices lower,
essential functions must be 'forgotten'.

The protection circuit often isn't in a 'separate and distinct'
package, regardless of the supply's price or quality, examples being
high quality Fortrons using the Fairchild KA3511 chip and relying on
its over/under voltage sense inputs (this manufacturer claims its ATX
supplies do have OVP) and low quality generic supplies built around the
SG6105 ATX controller. I've also seen many generics where the OVP was
in a 'separate and distinct' LM339 quad comparator chip. And I've
never see an ATX supply containing crowbars as part of the OVP
circuitry to block voltage spikes generated by inductor saturation, and
I doubt even PC Power & Cooling supplies have them. Can you cite an
ATX power supply brand that does or that meets your minimum standards
of acceptable quality?

 

[w_tom]

Even power supplies in original IBM PC and PC-AT had OVP.
However, this is irrelevant. A consumer must see that
function, in writing, on a numerical spec sheet. If not
specifically stated, then OVP probably does not exist. No OVP
suggests a supply marketed to naive 'bean counters'.
Meanwhile, OVP was missing in a latest Gateway power supply
that I recently repaired. Another example of a product that
sells on price rather than quality.

LM339 quad comparators are typically part of the Power Good
circuit; not large enough to latch output power and act as an
OVP circuit. Should any critical voltage be too low, the
LM339 signals all voltages to shut off. This is not
acceptable for OVP protection. LM339 circuits here typically
trigger on voltage too low; not on voltage too high. The
latter (too high) is necessary for an OVP circuit.

One final point. For an OVP to be in the regulator, then
the regulator needs optocouplers or equivalent to monitor
output voltage. The only optocoupler in most supplies is for
voltage regulation. Just another reason why the OVP would be
separated from regulator circuit. If power supply has OVP,
then it is not part of the regulator circuit. But again, the
relevant bottom line: are OVP, overpower protection, and those
other essential functions listed on a manufacturer numerical
specification? Minimally acceptable supplies claiim to
provide those essential functions - in writing.

 

[bryanlove]

(e-mail address removed) wrote:

Even power supplies in original IBM PC and PC-AT had OVP.
However, this is irrelevant. A consumer must see that
function, in writing, on a numerical spec sheet. If not
specifically stated, then OVP probably does not exist. No OVP
suggests a supply marketed to naive 'bean counters'.

The existence of such spec sheets is meaningless because almost every
marketer of power supplies has them available, whether they're based on
reality or are complete fabrications.

Meanwhile, OVP was missing in a latest Gateway power supply
that I recently repaired. Another example of a product that
sells on price rather than quality.

What brand? Gateway has often used Newtons.

LM339 quad comparators are typically part of the Power Good
circuit; not large enough to latch output power and act as an
OVP circuit. Should any critical voltage be too low, the
LM339 signals all voltages to shut off. This is not
acceptable for OVP protection.

A chip like an LM339 is typically used only to turn off the PWM
controller, and that's about all the OVP we get now, even in the most
expensive supplies - no crowbar.

One final point. For an OVP to be in the regulator, then
the regulator needs optocouplers or equivalent to monitor
output voltage.

Not if the regulator is on the secondary side.

 

[bryanlove]

(e-mail address removed) wrote:

Even power supplies in original IBM PC and PC-AT had OVP.
However, this is irrelevant. A consumer must see that
function, in writing, on a numerical spec sheet. If not
specifically stated, then OVP probably does not exist. No OVP
suggests a supply marketed to naive 'bean counters'.

The existence of such spec sheets is meaningless because almost every
marketer of power supplies has them available, whether they're based on
reality or are complete fabrications.

Meanwhile, OVP was missing in a latest Gateway power supply
that I recently repaired. Another example of a product that
sells on price rather than quality.

What brand? Gateway has often used Newtons.

LM339 quad comparators are typically part of the Power Good
circuit; not large enough to latch output power and act as an
OVP circuit. Should any critical voltage be too low, the
LM339 signals all voltages to shut off. This is not
acceptable for OVP protection.

A chip like an LM339 is typically used only to turn off the PWM
controller, and that's about all the OVP we get now, even in the most
expensive supplies - no crowbar.

One final point. For an OVP to be in the regulator, then
the regulator needs optocouplers or equivalent to monitor
output voltage.

Not if the regulator is on the secondary side.

 

[w_tom]

Missing numbers on a spec sheet is damning information.
Power supply vendors that have 'forgotten' essential functions
routinely don't provide them. Some provide spec sheets that
are really jokes; a few lines so that the naive also will
'feel' that Saddam has WMDs. We call them 'color glossies' so
that the boss will 'feel' he knows what is happening.

In switching power supplies, regulator (called pulse width
modulator) is on AC side of galvanic isolation. It oscillated
maybe 300+ volts. Galvanic isolation between those 300 volts
and secondary (3.3, 5, and 12 V) is a critical function of
every AC power supply design. Galvanic isolation is why
voltage output is typically monitored using an optoisolator.
Regulator cannot be on secondary side and create those 300+
volt pulses. Pulse width modulator is on AC side where 300+
volts is oscillated at 10+ Khz according to information
received from secondary side via that optoisolator.

If the LM339 turns off a pulse width modulator (the
regulator chip), then a connection from LM339 to regulator
must be via galvanic isolation (ie. another optoisolator). So
where is that galvanic isolator? Regulator for ATX switching
power supply would not be on secondary side.

Standard design uses an LM339 (or equivalent) to drive a
Power Good (gray wire) signal. Then power supply controller
responds accordingly and orders the power supply to power off
(via green wire). LM339 typically causes system power down if
any critical voltage is too low. A function completely
different from OVP which responds to voltage too high.
Whereas LM339 continuously monitors voltage, the OVP must do
something different: latch. Again, a function demanded by
Intel specs and a function that the LM339 does not perform.

Correctly noted; that Gateway supply, missing OVP, was a
Newton.

 

[John Doe]

w_tom wrote:

Not if the regulator is on the secondary side.

Are you new to USENET? Arguing with w_tom in an off-topic group will
get you nowhere. He scours USENET for post which contain his pet
keywords and then jumps in and tries to baffle 'em with BS (because
he can't dazzle 'em with brilliance). Best is to catch him in
sci.electronics.design if you want to learn something from the reply
authors.

 

[David Maynard]

Missing numbers on a spec sheet is damning information.
Power supply vendors that have 'forgotten' essential functions
routinely don't provide them. Some provide spec sheets that
are really jokes; a few lines so that the naive also will
'feel' that Saddam has WMDs. We call them 'color glossies' so
that the boss will 'feel' he knows what is happening.

Everybody makes 'glossies' because most people aren't engineers and
wouldn't know what to do with a spec sheet.

In switching power supplies, regulator (called pulse width
modulator) is on AC side of galvanic isolation. It oscillated
maybe 300+ volts. Galvanic isolation between those 300 volts
and secondary (3.3, 5, and 12 V) is a critical function of
every AC power supply design. Galvanic isolation is why
voltage output is typically monitored using an optoisolator.
Regulator cannot be on secondary side and create those 300+
volt pulses. Pulse width modulator is on AC side where 300+
volts is oscillated at 10+ Khz according to information
received from secondary side via that optoisolator.

That's one way to do it but by no means the only way and it's common to
have the PWM on the secondary side driving the high voltage switching
transistors via a pulse transformer.

If the LM339 turns off a pulse width modulator (the
regulator chip), then a connection from LM339 to regulator
must be via galvanic isolation (ie. another optoisolator).

No it 'mustn't'. Only has to be galvanic isolated if the PWM is on the
primary but it needn't be.

So
where is that galvanic isolator? Regulator for ATX switching
power supply would not be on secondary side.

Sure it could, and many are.

Standard design uses an LM339 (or equivalent) to drive a
Power Good (gray wire) signal. Then power supply controller
responds accordingly and orders the power supply to power off
(via green wire). LM339 typically causes system power down if
any critical voltage is too low. A function completely
different from OVP which responds to voltage too high.
Whereas LM339 continuously monitors voltage, the OVP must do
something different: latch. Again, a function demanded by
Intel specs and a function that the LM339 does not perform.

It can if you know how to wire it right.

 

[bryanlove]

(e-mail address removed) wrote:

Meanwhile, OVP was missing in a latest Gateway power supply
that I recently repaired. Another example of a product that
sells on price rather than quality.

One final point. For an OVP to be in the regulator, then the
regulator needs optocouplers or equivalent to monitor output voltage.

In switching power supplies, regulator (called pulse width
modulator) is on AC side of galvanic isolation. It oscillated
maybe 300+ volts. Galvanic isolation between those 300 volts
and secondary (3.3, 5, and 12 V) is a critical function of
every AC power supply design. Galvanic isolation is why
voltage output is typically monitored using an optoisolator.
Regulator cannot be on secondary side and create those 300+
volt pulses.

It can and often is, as shown by all those PC supplies containing 3
transformers. In addition to the usual large power transformer and
small transformer for the standby +5V is another small transformer to
couple the control output of the low side controller to the bases or
gates of the primary side power transistors.

If the LM339 turns off a pulse width modulator (the
regulator chip), then a connection from LM339 to regulator
must be via galvanic isolation (ie. another optoisolator). So
where is that galvanic isolator? Regulator for ATX switching
power supply would not be on secondary side.

Again, the need for isolation is eliminated if the PWM controller is on
the secondary side.

Standard design uses an LM339 (or equivalent) to drive a
Power Good (gray wire) signal. Then power supply controller
responds accordingly and orders the power supply to power off
(via green wire). LM339 typically causes system power down if
any critical voltage is too low. A function completely
different from OVP which responds to voltage too high.
Whereas LM339 continuously monitors voltage, the OVP must do
something different: latch. Again, a function demanded by
Intel specs and a function that the LM339 does not perform.

But comparators like the LM339 can be used to sense overvoltage, and
low side PWM controllers merely have to stop the oscillation to latch
the shutdown.

Correctly noted; that Gateway supply, missing OVP, was a Newton.

In my Gateway's 90W Newton are a UC3843 high side controller and a
daughter board on the low side with a 358 (LM358?), a house marked
16-pin DNA1002D (often found in Delta brand supplies), a small
transistor, several resistors, and 4 glass diodes. Diodes like this
are often involved in either OVP sensing or in generating the
Power_good signal, but I didn't trace out the circuit because removing
the PCB would require removing a snap-in switch that I didn't feel like
breaking. I don't know how similar your supply is to mine or if mine
has OVP, but it's at least as elaborate as the supplies that I know
contained such protection.

 

[w_tom]

Yes more inductors can be added to the galvanic isolation.
But that is on the order of something under ten dollars per
inductor. Unnecessary since the PWM works just fine on the
high side without another expensive transformer. PWM is
placed on the high side and uses only one optocoupler as
feedback to control output voltage.

I have never seen a switcher with the PWM on the low side.
Where are schematics or model numbers to see this design?

Your numbers for the Newton sound familiar. That Newton
(250 w) had two ICs - DNA1001 and DNA1002. It also had an op
amp (LM358?), numerous transistors, and many small signal
diodes (1N4148 etc). It used no LM339 since that Power Good
monitor function appeared to be part of those DNA100x ICs.

Meanwhile, returning to the original post, OVP is one of
many functions required in minimally acceptable supplies. As
the OP posted:

Luckily so far none of my other components have fried.

Of course. Any minimally acceptable supply must not 'fry'
computer components - today as was also true 30 years ago.
However some bought supplies that sold on price; therefore
have had computer component damage.

OVP and other essential functions are necessary in any
minimally acceptable supply. If functions are not
specifically claimed on its spec sheets, those functions have
probably been forgotten to cut costs and increase profits

 

[larry moe 'n curly]

I have never seen a switcher with the PWM on the low side.

Your stern, long-winded lectures implied that you knew everything.

Where are schematics or model numbers to see this design?

www.sg.com.tw/semiGP/data/6105/6105-datasheet.pdf

A variation of this can be found in this schematic of a 300W InWin PSU:

http://spblan.narod.ru/bp/shema/iwp300a2.gif

Notice that the SG6105 controller chip includes over/under voltage
protection in the same package as the PWM regulator.

Lots of photos and descriptions of other ATX PSUs here, including
details about what controller chips are used:

http://terasan.info/dengen/

The ATX PSUs with low side PWM regulators have two small transformers
instead of one, in addition to the large main transformer(s).