The Bestec Mystery Solved

[everell]

The purpose of this thread is to attempt to resolve the remaining mystery of why the Bestec ATX-250 12E turned into such a disaster and became known as the motherboard killer.

From a previous thread:

Toasty: The cap failure may be a result of, or perhaps only a contributing factor of the meltdown, not the root cause………… There is evidence of other components failing (resistor and diode) that may in fact be closer to the actual cause.

The FIRST root cause has been attributed to failure of capacitor C1 which is the filter on the feedback winding of the 5vsb transformer. Tests done by Severach and myself indicate that a series resistance can simulate esr in this circuit and produce an increase of about one volt for each ten ohms of increase in esr. Further testing has shown that the filter capacitor in this location will produce a similiar problem in other brand power supplies using the two transistor design. Therefore, if this capacitor goes bad, either because of cheap poor quality, heat damage over time, or simply electrolyte drying out, the capacitor will develop higher and higher esr problems.

This problem has been discussed and documented on several previous threads, so I will not beat it to death - this problem which I shall label as Root Cause #1 is already well known.

[everell]

Re: The Bestec Mystery Solved

Root Cause #2 is the result of a different design on the Bestec "12E" On the feedback winding of the 5vsb transformer there are TWO voltage sources. One feeds the two transistor circuit and is part of the voltage regulation. It is rectified, then filtered by the problem causing capacitor C1. It reaches a value of about 12 volts. The other part of the winding is rectified and filtered by capacitor C50. This voltage source is about 20 volts. It goes thru a voltage regulator to bring it down to a regulated 17.3 volts to power the pwm chip and its circuitry.

The problem that comes up is that these two voltage sources are coming from the same transformer winding. Therefore, as the voltage on the winding with the C1 capacitor goes up, the voltage on the other part of the winding with capacitor C50 goes up proportionately. Once the voltage across capacitor C50 reaches its rated voltage, it goes BOOM.

Root cause #2 is that voltage applied to capacitor C50 is going beyond its rated voltage. As the voltage across C1 goes up, the 5vsb voltage goes up and the voltage across C50 goes up. Tests run by Severach indicate that when the 5vsb voltage increases from 5 volts to about 12 volts, the voltage across capacitor C50 goes up to somewhere around 40 volts. In my testing I was able to see it go up to 55 volts before the BOOM.

[everell]

Re: The Bestec Mystery Solved

Root Cause #3 is due to a unique design in the Bestec "12E". When the computer is on, the 20 volt source voltage on the 5vsb winding is providing power to the pwm circuit. But when the computer is turned off, two things happen. First, the load on the 20 volt source is no longer present. Secondly, the fan shuts off. Bestec solved the loading problem with a circuit which I shall call a dynamic load. Its two basic components are transistor Q17 and resistor R102 a two watt 1000 ohm resistor.

This is how the dynamic load circuit works. When the computer is on, Q17 turns off and R102 is not in the circuit. When the computer turns off, Q17 turns on and puts R102 across the 20 volt source as a load resistance. With no esr problems in capacitors C1 and C50, the circuit works great, but will generate a little bit of heat. Again, the fan is off. So the fan is not cooling things down. Any heat from R102 will contribute to "cooking" the nearby capacitors.

As the esr of capacitor C1 begins increasing, the 5vsb voltage increases, and the voltage across R102 increases, causing current to increase, and more heating from the resistor. This cooks the capacitors at a faster rate, and helps to increase the esr even further, which causes the resistor to create even more heat. As you can see, this becomes a vicious circle.

The Third Root Cause is "HEAT" from the 5vsb circuit (still running) and resistor R102 while the fan is not running.

At this point, I believe that heat from these 5vsb two transistor circuits while the fan is not running is a significant problem for all of these power supplies, regardless of brand. The Bestec was the worst power supply only because it had the extra heat from resistor R102.

[everell]

Re: The Bestec Mystery Solved

For a solution to Root Cause #1 and Root Cause #2, I removed the two transistor 5vsb circuit components, and wired in a DM311 IC circuit. I won't bore you with the details because they are in a previous post along with pictures.

The new circuit does not contain an electrolytic capacitor in the feedback circuit. Therefore the 5vsb voltage will not go up over time. As long as the IC doesn't fail, it will hold a nice steady 5 volts. This solves Root Cause #1. If the 5vsb does not go up, then the voltage across C50 will not go up. This solves Root Cause #2.

The new circuit does contain an electrolytic capacitor in the startup circuit of the IC. If it has a problem, the chip will simply not start up the 5vsb.

This should solve the mother board killer part of the problem.

[everell]

Re: The Bestec Mystery Solved

The solution to Root Cause #3 was more difficult to figure out. First, to remove the dynamic load circuit with R102, the heat generator.

I tried several circuits, but all required cutting pc board traces and changing components. Lots of work!

Then there is the matter of the fan being off when the computer is turned off. I was trying to develop a new circuit which would run a fan when the computer is off as well as when on.

Then the light turned on....Eureka, I found the solution.......and so much easier than expected.

To solve the problem, remove resistor R102. Then solder the leads of a fan into the two holes where R102 was. Now the fan serves as the load. Transistor Q17 turns the fan off when the computer is on, and turns the fan on when the computer is off. The fan serves as the needed load, and the fan also cools the circuitry.

Now just cut a hole in the power supply case on the opposite side of where the other fan is. Mount the fan on the outside of the case (no room on the inside). Have this fan blow into the power supply while off and the other fan pulls air out when the computer is on.

This is my solution to Root Cause #3.

Someone else might find another root cause, and hopefully a solution. I think this covers the major issues. I believe this converted Bestec will now be a good power supply.

[Per Hansson]

Re: The Bestec Mystery Solved

Could you not simply connect it to the original fan with a diode in-line, so when the computer turns off the fan just gets it's power from R102?
Would look a bit more slick
And perhaps a small resistor to slow the fan down too if noise concerns you

[everell]

Re: The Bestec Mystery Solved

Using a common fan with a diode network becomes very complicated because the existing fan uses the 12 volt rail on the secondary side and case ground also on the secondary side. The R102 location used a voltage source on the primary side and common (or neutral) on the primary side. Connecting something with power from both sides is inviting a big BOOM.

Slowing down the fan with a series resistor is an option. The added heat from the resistor gets back to the same heating problem, but at least you have some cooling action from the fan. Another idea I am using is to install a LED in series with the fan. Then you can see when the fan is on. The LED was TOO bright, so I added a shunt resistor across the LED to take some of the current. I used two 33 ohm resistors in parallel. Once I added the LED circuit, the fan noise wasn't all that objectionable. Remember that when this fan is on, you have turned off the computer and are busy doing something else. So who will be paying that much attention to it? Also, before turning the computer on, if you don't hear that fan running you know you have a problem.

The real advantage of the fan idea is that while the computer is off and you are doing something else, the power supply is staying cool. It is not busy cooking those cheap Chinese capacitors. I think this would be an advantage for any power supply. Turns out that the killer Bestec already has a circuit that lends itself to this concept by simply changing R102 to a fan circuit. Now you don't have to throw away that Bestec, you can experiment with the "fan on while power off" idea.

[jpdoe]

Re: The Bestec Mystery Solved

Thanks for your work on this issue, everell.
Your posts are very enlightening.

[i4004]

Re: The Bestec Mystery Solved

everell, the root cause to all issues here are badcaps (c1 and c50 for those psus), and issues just differ in what those badcap(s) do to circuit, it seems to me, if i got everything you wrote.

ie it all starts with cap's esr increasing.

>Toasty: The cap failure may be a result of, or perhaps only a contributing factor of the meltdown, not the root cause………… There is evidence of other components failing (resistor and diode) that may in fact be closer to the actual cause.

yes, i remember this...seemed like a crappy response then (and i explained why in that thread), seems like it now too, esp., after this problem dissection by you, which proved main cause are caps of these crappy psus which nobody should use anyway.
<wink>

>Now you don't have to throw away that Bestec, you can experiment with the "fan on while power off" idea.

i must say that i find that idea quite horrible, because my aim is to have silent pc even when it's working, let alone when it's in stby.
so no, i wouldn't make any concessions to crappy psu.
better buy better one.

still, i applaude to your efforts to investigate this issue fully(and offcourse for ways to solve the issue for those who want to continue using those psus).
well done.

to everybody else i would just say
"better pick fortron".
(i'm also seeing some cheap psus with decent 5vstby circuits and big fans etc. but i see asiax caps and that kinda spoils it...it won't fry the mobo, but i doubt it'll work for long...)

[ratdude747]

Re: The Bestec Mystery Solved

i can confirm root3. i had a bestec 12e out of a problematic emachines. mobo was fine but psu destroyed. i no longer have the psu but one cap near a pink LARGE resistor was burnt and charred. another cap was bloated. the burnt one had leaked form the base and had burnt so bad that the back powdwery ash was all over the insides of the comuter and was concentated inside the case of the psu.

my conclusion was the resistor overheated, or root cause 3. i still have the bare board to the psu i will check if it was r102 that went up.

[ratdude747]

Re: The Bestec Mystery Solved

mine was r203. everell, would r203 have been affected? thats what overheated in my 12e.

[370forlife]

Re: The Bestec Mystery Solved

"better pick fortron".
(i'm also seeing some cheap psus with decent 5vstby circuits and big fans etc. but i see asiax caps and that kinda spoils it...it won't fry the mobo, but i doubt it'll work for long...)

Ya, pick FSP if you want out-of spec ripple and mediocre voltage regulation....FSP is not a recommended brand anymore, at least not at jonnyguru (about the same as sirtec/sirfa built OCZ's and coolmax's)

Whats funny is that the bestec atx-250-12e on jonnyguru actually had better ripple suppression than some fsp's, and it had some bad caps.

I would not say that bestecs are crappy, they are actually quite well built. They are very similar to Hipro units. They can run their ratings, within spec ripple, its just the abysmal 5vsb on the 12e models, the 12z models are fine.

[ratdude747]

Re: The Bestec Mystery Solved

hell, deer wouldnt kill mobos. just make then no post and fry hard drives.

[i4004]

Re: The Bestec Mystery Solved

>FSP is not a recommended brand anymore, at least not at jonnyguru

was it ever?

seems to me lately they prefer 1kw(and near 1kw) psus and that's rather uninteresting...
they "grew" to become outlet for kids that wanna play games...hehe...

>Ya, pick FSP if you want out-of spec ripple and mediocre voltage regulation...

what 300-500w models would you be talking about here?

>I would not say that bestecs are crappy, they are actually quite well built.

quite well, but with a mystery here and there...hehe...
and some of those mysteries can blow the mobos up...

no, that's not a good company if you ask me....


bottom point: there are good psus that are cheap too, so why bother with these?
it's good for experiments and learning, but you won't really be putting these(even after mods) into customers machines, right?

--------------------------
>my conclusion was the resistor overheated, or root cause 3.

and what made that resistor overheat?

[momaka]

Re: The Bestec Mystery Solved

hell, deer wouldnt kill mobos.

An old J&C with a no-feedback 5vsb will quickly change your perspective (those designs are even worse than the Bestec ATX-250 12E).

Allied/Deer/L&C use the same 5vsb design as the Bestec ATX-250 12E. Perhaps what's keeping them from killing mobos is the fact their critical capacitor is in a different place and maybe it isn't overheating. But that still doesn't eliminate root cause #1 for the Allied/Deer/L&C units. This actually goes out to all PSUs with a 2-transistor design and all have potential to more or less become mobo killers if the conditions are right. The ATX-250 12E just happened to be the PSU with those magical talents.
everell and Severach did a spectacular job documenting these flaws. Without them doing these tests, all Bestecs would still be considered mobo killers. But I would have to agree with 370for life that Bestec PSUs (models other than the ATX-250 12E) are indeed built quite well. Actually even the ATX-250 12E is well built, it's just the crappy 5vsb design.

[ratdude747]

Re: The Bestec Mystery Solved

oh, didnt know. sorry, my bad.

[everell]

Re: The Bestec Mystery Solved

I think some people are misunderstanding my intentions. Late last year I asked the question - why do so many computer power supplies fail, and why are they simply tossed in the garbage instead of being repaired. At that time I was not particularly interested in opening the "box" with all that crowded complicated stuff inside. I simply trusted the label on my computer with a PowMax 400 watt power supply.

I have opened the box. I was exploring inside. And I found this website. Just a coincidence, the first computer power supply I started looking at was the Bestec ATX-250 12E.

My initial opinion - the two transistor 5vsb circuit was terrible - and I experimented with a workable design which replaced the original two transistor circuit. I put this experimental circuit in the Bestec and two other power supplies. I do NOT do this for other people or as a business venture. It was simply a challenge to me personally. As Severach put it, you can buy a pallat of HiPro power supplies cheaper than you can retrofit Bestecs. From a business prospective, I agree completely. But for myself, I feel that all three power supplies are now SAFER.

As for the root cause of the Bestec failures - IN MY OPINION - it is a heat related problem. Yes, the capacitors C1 and C50 failed. But with the amount of heat generated by the power resistors near these capacitors AND the poor circuit design, I believe these capacitors are doomed to failure over time. The burned areas under capacitors AND resistors AND diodes are sufficient proof that things got hot. With no fan running while the computer is off, the problem is only compounded.

If the original Bestecs had a fan running during power off, would the same problem with failures have occurred? At worst it would have slowed it down. Too many parts in a crowded area and too much heat over time equals a problem. I also now believe that it is a problem area for most of the other brands. As the ATX specification for the 5vsb has more and more current capability - with no fan we can EXPECT more problems and psu failures.

So I disagree with you - i4004 - The root cause was HEAT. The capacitors in the Bestec were no worse than any of the other economy power supplies.

As for the fan noise - sorry it bothers you. I like it - because this project is EXPERIMENTAL. I like being able to hear that fan going - if it is not - then I know my project has been a failure. It is like music to my ears. This psu is NOT powering an important computer - it is running a AZZI KT3-AV which I saved from the trash and fixed by recapping. Anyone know anything about the AZZI KT3-AV (other than it works just fine)?

[370forlife]

Re: The Bestec Mystery Solved

So I disagree with you - i4004 - The root cause was HEAT. The capacitors in the Bestec were no worse than any of the other economy power supplies.

Jamicons really aren't that bad. I would rate them just under teapo/taicon/capxon. Lite-on likes to use them in even their high end server psu's for some of the small caps.

ohh, and i4004, jonnyguru.com has not done any low wattage fsp units, as you have stated because so many people with a intel atom decide they need a 1kw psu to power it...but the lowest wattage fsp psu was the Everest 700W, based on the epsilon platform (well, that is not a relabeled fsp):

Say, have you ever started to climb a really icy hill only to slide back down to where you started with every step you took? I kind of felt like that from the beginning on this unit, for it absolutely refused to start test one. Ten times in a row. I would hit the on button on the front of the 268, the fan would twitch, I would see a brief spurt of activity on my Brand 4-1850 wattmeter, and the unit would play dead. Clearly, protection circuitry was at work. Malfunctioning protection circuitry that is, for test one's parameters are well above minimum load specs for this unit.

It's not unheard of to get the odd turkey in your PSU chicken coop, after all. And so, I waited and waited. Finally, the second sample showed up. I eagerly unboxed it, hooked it to the load tester, and... the second unit failed test one. Same symptoms. Ten times in a row. Only this time, unlike sample #1, I could fire it up at test two where the fan would spin and promptly quit.

Though you can't see them at this resolution, I discovered several tiny solder blobs here and there. Some were shorting things out. I cleaned everything up, desoldered the heatsinks for part ID purposes, put the heatsinks back in, and suddenly sample #1 was working properly, although still not starting up at test one.

Did I mention the lack of an insulator below the PCB? Well... it was missing in this unit, just like the 900W. Craptastic.

From above, at first glance nothing seems amiss here. But a closer look reveals the lack of a MOV in the otherwise complete transient filter. Man, they're cutting corners everywhere in here.

Is that a black wire I see there soldered to the 3.3V output? Why yes, I do believe it is. Folks, that right there is the negative wire for the fan. This just gets better and better, doesn't it?

It was around this point I tore into sample #2. Guess what - the fan was wired up the same way. Positive to the 12V based fan controller, negative to the 3.3V lands. While this in itself is not too much a cause for concern, it is an extremely half baked way to go about keeping your fan below 12V in an effort to be silent. I mean, surely there are better ways to drop the full power voltage to the fan than this. Perhaps a resistor or something.

I did some experimenting to discover how high the unit had to be loaded before it would stop playing dead and start working. As is turned out, one rail only had an effect on the problem. Guess which rail it was. Ding ding ding!!! If you said the 3.3V rail, give yourself a prize! As it worked out, the unit would not start up with any load below 4A on the 3.3V rail. The minimum load spec is significantly below that number, at 0.8A.

[i4004]

Re: The Bestec Mystery Solved

>As for the root cause of the Bestec failures - IN MY OPINION - it is a heat related problem. Yes, the capacitors C1 and C50 failed. But with the amount of heat generated by the power resistors near these capacitors AND the poor circuit design, I believe these capacitors are doomed to failure over time. The burned areas under capacitors AND resistors AND diodes are sufficient proof that things got hot. With no fan running while the computer is off, the problem is only compounded.

i wrote about enermax +5vstby section, and also added some images.
its fans are surely not spinning when its in stby.
and what caps?
jpce-tur....huhu...

https://www.badcaps.net/forum/showpo...47&postcount=6

(that was before your work on bestec that has feedback and still can blow the mobo...ie i thought only self-oscilating circuits could do such damage...)

also for example this by kc
https://www.badcaps.net/forum/showpo...24&postcount=4
"i have ben keeping a close watch on it but as i suspected years ago they are fine when good parts are used."

>So I disagree with you - i4004 - The root cause was HEAT. The capacitors in the Bestec were no worse than any of the other economy power supplies.

well ok: overheating the crappy caps because of bad design.
guess it could be condensed like that too.

what you said in first post was ok
"Therefore, if this capacitor goes bad, either because of cheap poor quality, heat damage over time, or simply electrolyte drying out, the capacitor will develop higher and higher esr problems."

it's all of that...and design...
it's just that heat accelerates its demise...much simillar to badcaps on mobos...
if you were to keep them on 0°c they would probably last 10years....heh...
otoh, good caps can do that without the need to keep them cool.

so bestec did something we see all crappy psu makers do: bad design+bad components=boom!

as for what are they doing now with improved designs? well just getting on with the time i would think....it's probably cheaper for them to do it with chip now.
ie i don't see them intentionally making an effort to move away from "mobo killer" badge. i would say you have many simillar companies that started making better +5vstby circuits just because chip is cheaper....heh...
if transistor design was cheaper they would probably stay on it, like they did for so many years after that technology was essentialy obsolete in most better mfrs' models.
(excuse me for not finding good words for bestec, but i lost a config to simillar make ("jnc") and i say they should not be praised but spat on even if they make better models today...one pc lost to bad psu is one pc too many...)

btw. what exactly were those caps that were leaking after being subjected to that heat torture?
jpce-tur?

>From a business prospective, I agree completely. But for myself, I feel that all three power supplies are now SAFER.

i think nobody contested that.

-----------------
370, what i said was
"to everybody else i would just say
"better pick fortron"."

ie like you better pick fortron over bestec and alike.

are you arguing with that, and if so, what proof do you have bestec is a better make than fortron?
in the field, you ecountered more problematic fortrons than bestecs?
come on...

or, do you have pix that show fortron is using worse components(in the same psu class) than bestec?
when was the last time you took fortron apart? what did you see?

overall, what are you saying? that a particular fortron had some ripple on jonnyguru?
so?

for a moment stop and ask yourself this; can i confirm that even this out of spec fortron damaged some pc, and after which time period did this happen, and what exactly was the damage?

today i had a guy asking about what to buy: he brought a list of components dealer gave him and i just wrote on the paper "fortron or chieftec psu, not those bundled with case".
(chieftec is delta..somewhat more expensive than fortron)
i say i was right.
prove me wrong if you can
<wink>

>ohh, and i4004, jonnyguru.com has not done any low wattage fsp units, as you have stated because so many people with a intel atom decide they need a 1kw psu to power it..

i honestly don't know for whom are those tests made....but they all seem to have that tendency...top kw models are tested....
i guess that brings more money for them.
and keeps top kw makers happy as they'll sell at least few pieces of every 1kw model they make...
(heh...ok, a bit over-cynical there...)

but there is a clear tendency to go with higher and higher wattage indeed.
look at the earliest tests...

>but the lowest wattage fsp psu was the Everest 700W

no, not really
http://www.jonnyguru.com/modules.php...=Story&reid=84
"In a market where the latest and greatest computer power supplies are all huge monsters capable of delivering gobs of current at a time, the middle ground of units promising just enough power at affordable prices with good reliability is often overlooked. Today I'm looking at a unit that promises to slide right into that middle ground, do a good job powering the average system without driving you to the poorhouse, and look good doing it: the FSP Blue Storm II."

kinda ironic, isn't it?
hehe...

this is good
http://www.jonnyguru.com/modules.php...Story&reid=154

i like the humor
"Like a life preserver in the ocean, I kept hitting the button and the Bestec kept bobbing right back up and grinning at me without shutting down. It was like it was saying, "I'm gonna fry me a motherboard, and you ain't gonna stop me." But the SunMoon is no motherboard, Mr. Bestec. You can't kill it by overvolting. So, who's grinning now?"

hehe..

i mean they surely are not recommending bestec, are they?

but heck, that's at least one interesting test they did in a long time.
i guess i'm too cheap to look at 1kw models.
what about you?
<wink>

anyho, wolf condensed it well
"What good is above average ripple suppression on the main rails when you have to replace your mainboard every three months, anyway?"

again; i would avoid bestec, but would encourage anbody to tinker with it(in safe environment) to see what made it tick(or better to say "go kaboom"..heh).
i would do it myself fi i had them here.
and i DID investigate what made my JNC go boom and destroy whole system.
so i say i did my share.
i hope i won't be having the need to investigate such accident on my gear ever again.
i moved away from cheapest psus long time ago, so i'll probably be just fine.

[370forlife]

370, what i said was
"to everybody else i would just say
"better pick fortron"."

ie like you better pick fortron over bestec and alike.

I would just stay away from both. They are both good psu's and fine after a recap, but if you want a psu that you can buy and throw in your computer and never have to worry about, stick with seasonic, enermax, delta, antec (now almost completely delta except for the TPN series.)

overall, what are you saying? that a particular fortron had some ripple on jonnyguru?
so?

for a moment stop and ask yourself this; can i confirm that even this out of spec fortron damaged some pc, and after which time period did this happen, and what exactly was the damage?

Well, lets see, it could kill your hard drive, motherboard, graphics card, pci cards, etc... and as for time period? That depends on the aforementioned parts.

today i had a guy asking about what to buy: he brought a list of components dealer gave him and i just wrote on the paper "fortron or chieftec psu, not those bundled with case". (chieftec is delta..somewhat more expensive than fortron) i say i was right. prove me wrong if you can <wink>

Ya, and I know somebody who thinks that coolermasters are the best psu's and won't buy anything else...just mediocre acbel's.