Bestec revisited

[i4004]

Re: Bestec revisited

>If it only has one 8 pin integrated circuit chip between the two heat sinks, AND has two transistors on the heat sink nearest the two big input capacitors, it is a mobo killer.

this one
https://www.badcaps.net/forum/showth...hlight=fortron
has 0 8-pin ics between the two heatsinks, has two transistors on the heat sink near the big caps...
yet it's not unsafe at all.
it has 5vstby design with ic, only that ic is on the same side of the heat sinks as thos big caps...
those two big transistors drive the primary of the main transformer...

what i'm wondering is what is pwm ic doing on primary side?

anyhow, the way to check it is to see what ic(or ics) is (are) on the primary side...
then find its datasheet on the net...

or to see sizes of transistors...those biggest transistor(s) are not connected to 5vstby...but if you have few(or one) more (smaller) transistors on primary, those are 5vstby transistors, which just shouldn't be there...
that circuit should have ic and no transistors...

interesting to see seasonic was making such designs too....
do you have esr-meter?
what's esr of those caps in that circuit?

enermax was also putting 2 transistors in their design
https://www.badcaps.net/forum/showpo...47&postcount=6
guess that was normal 7-8 years back..huh

with the data you provided we can distinguish 3 kinds of 5vstby circuits:

1-self oscillating design(transistors) that doesn't even have feedback circuit...

2-oscillator(transistors) with feedback(as the bestec you described here
https://www.badcaps.net/forum/showthread.php?t=6680 )

3-design with ic, not transistors...

because 2 obviosuly can give too high voltage on 5vstby both 1 and 2 are potential mobo killers(this seasonic included) and 3 shouldn't be...

[Toasty]

Re: Bestec revisited

@i4004 - I'm fairly certain he was speaking only of the Bestec branded units, not every one else's units.

@Everell - According to its UL approval, that Sea Sonic is to have PFC. Does it?

[severach]

Re: Bestec revisited

None of them show a bloated cap, so these are all non-obvious fails.

Those failed in service and were replaced with other supplies. I already fixed these discards before the pictures were taken. Each had a bad output capacitor or two which is easily detected by the supervisor chip which proves nothing about the design safety because even the serial killer ATX-250-12E Rev P6 Bestec have a supervisor chip and shouldn't destroy a motherboard with a bad output cap.

So the conclusion is any 12E series, regardless of revision number is a killer.

I have an ATX-300-12E Rev D in the mix which has the same potentially safe PWM 5vsb design as the 12Z supplies.

Here's a Bestec ATX-250-12E Rev P6 that destroyed an eMachines motherboard. The lone Fuhjyyu output cap among the Jamicon says that this is one of my quick recaps. Had I discarded the Bestec the board might still be alive.

[severach]

Re: Bestec revisited

Here are the pictures. I found an ATX-250-12E Rev P7 which appears to be identical to the ATX-250-12E Rev P6. The board revisions are both P5.6. The P7 unit had not failed and shows no heat or cap damage. I can't find any obvious components with revised values near the heat damaged area.

[Toasty]

Re: Bestec revisited

>>...supervisor chip and shouldn't destroy a motherboard with a bad output cap. <<
Does it monitor the 5vsb though? The Bestec killers were the ones with the 5vsb going to 12v or more. Yes?

>>Each had a bad output capacitor or two which is easily detected by the supervisor chip...<<
How does it detect a bad capacitor? If the voltage goes wonky, okay. But suppose it doesn't? Exactly what is that chip monitoring/checking/doing?

Toast

[Wizard]

Re: Bestec revisited

Like I explained many time, toasty, The design that does not have protections try to keep voltage up and blow up if current is excessive and also if the capacitor go bad, voltage goes out of control.

A good IC that have overvoltage and over current protection will shut down in few cycles with no damage till repair is done.

Cheers, Wizard

[Toasty]

Re: Bestec revisited

@Wizard - Not understanding my questions here.

How does it detect a bad capacitor? - Only if voltage goes out of control. Many units are leaking and blown and still running. We all have them and have seen them.

Exactly what is that chip monitoring/checking/doing? - In the 12E, it's NOT monitoring the 5VSB, -obviously-. Else it would shut it down and we would have no blown mobo's. So, the question is not what DOES the chip do (which I know), rather what IS *that* chip doing in the 12E? What line(s) is it monitoring?

>>have a supervisor chip and shouldn't destroy a motherboard with a bad output cap<<
The 5VSB cap that fails in these is *not* an output cap. At least 22uF@50V doesn't sound like one...

I hope that is clearer.

severach >>The lone Fuhjyyu output cap among the Jamicon says that this is one of my quick recaps.<<
Which was used in the quick recap? The Fuhjyyu or the Jamicon?

Toast

[severach]

Re: Bestec revisited

From the data sheet it looks like the TPS3510 isn't responsible for turning off an out of control 5vsb. That would be in the circuitry surrounding the IRIS-A6351 switcher for the models that have it. We hope Bestec didn't forget like they forgot for the transistor design.

How does it detect a bad capacitor? - Only if voltage goes out of control. Many units are leaking and blown and still running.

A bad output cap should show as an over voltage from the ripple and the inability of the switcher to maintain a steady output voltage. Many badly damaged caps still have low enough ESR and high enough micro farads to still function.

what IS *that* chip doing in the 12E? What line(s) is it monitoring?

TPS3510 datasheet:
Overvoltage Protection and Lockout for 12 V, 5 V, 3.3 V
Undervoltage Protection and Lockout for 5 V and 3.3 V
Fault Protection Output With Open-Drain Output Stage
Open-Drain Power Good Output Signal for Power Good Input, 3.3 V and 5 V

The 5VSB cap that fails in these is *not* an output cap. At least 22uF@50V doesn't sound like one...

Definitely not an output cap and the next time I get a disposable motherboard I can test the PWM safety by putting the bad caps from my serial killer 12E into one of these supposedly good Bestec and see if the sparks fly.

Which was used in the quick recap? The Fuhjyyu or the Jamicon?

The supply was full of Jamicon that I didn't replace. The Fuhjyyu is the recap. When I don't replace all caps I tend to use similar quality caps as the rest. A bunch of Jamicon do not deserve the company of a Rubycon.

[Wizard]

Re: Bestec revisited

These SMPS REQUIRES feedback to regulate voltage but poorer designs simply regulate voltage as much as it liked to do so as long as components are good then all bets are off if any of components degrades. AKA spirals out of control as SMPS tries to keep voltage but not knowing if it is shorted or bad capacitor.
I made lot of money on fixing CTC203 chassis due to this design. SMPS blow up if the HOT die (either bad flyback particuarly if 32-36" size or all of sizes with bad joint on the L14401 inductor for the base of HOT was running too long and HOT dies of bad signal on base, caused by hot glue flowing into solder and breaking the bond).

Intelligently designed SMPS uses good IC that has internal secondary overvoltage and over current monitoring if any of the conditions like short, feedback failure etc happens. Detecting shorts works because the IC is designed to detect in a way it see out of bounds condition, similarily with the overvoltage directly or indirectly. IC simply knew and instantly stop when it didn't expect feedback within bounds.

Replace the Jamicons also.

Cheers, Wizard

[Toasty]

Re: Bestec revisited

>>The Fuhjyyu is the recap.<<
I'm sending the cat to see you...

>>A bunch of Jamicon do not deserve the company of a Rubycon.<<
But they certainly didn't deserve a Fuhjyyu...

[everell]

Re: Bestec revisited

The 5vsb circuit DOES have some voltage feedback control. Kc8adu posted a photo showing the capacitor which causes the problem. See the attached drawing of the Bestec ATX250 12E two transistor 5vsb circuit. The capacitor in his photo is labeled C1 on the pc board and in my drawing. Notice that this is the capacitor which provides filtering for the optocoupler PC2 which controls the voltage regulation of the 5vsb circuit. If this capacitor goes bad, you don't have voltage regulation!

The SeaSonic has a two transistor 5vsb circuit almost identical to the Bestec. It is labeled C24 and has a value of 47uF/50 volts. My blue esr meter indicated that this capacitor is open. I replaced it and the 5vsb is now holding steady on 5 volts and the power voltages (3.3, 5, and 12) come up properly. So this is a CRITICAL capacitor.

Now for the question - why does it fail so often? All boards that I have worked on with this problem showed burn discoloration on the pc board, and some components possibly burnt. Is heat causing the capacitor to fail, or is the component failure causing the heat? Also, is it possible that voltage spikes are causing damage to the capacitor over time? I could be wrong, but I think it a possibility that when the power is first applied to the 5vsb circuit, the first inductive kick from the power transistor feeds back a spike in the feedback coil. I would think it would take a cycle or two for the voltage regulation to kick in. If so, this capacitor is taking the spike.

[Toasty]

Re: Bestec revisited

I've seen the burns too, but what component is causing that? My experience is typically near the 1 or 2 watt power resistor. Others?

>>why does it fail so often?<<
Scope pin 4 on the Iso with/without cap & with/without load and see what the p-p is. Perhaps a clue? Maybe cap underrated voltage?

Toast

[everell]

Re: Bestec revisited

I would love to look at that signal on pin 4 of the isolator, BUT --- Monday I had catarac surgery on my eye and I am resting this week. Will have to wait another week or two when I get back to work. Perhaps someone else would like to do this?

I have drawn some of the schematic for three different PSU transistor type 5vsb - all of which follow the same basic circuit. They are:

Bestec ATX-250 12E
SeaSonic SS-250 FS
Powerman IW-P250-A2-0

The Bestec ran me in circles until I replaced the transistors with a PWM chip (see my thread Bestec ATX-250 12E up and running) The SeaSonic had an open cap which I replaced. The Powerman blew up after I attempted a fix. The feedback winding was open. I don't know why that feedback winding opened - or if it is even related to this problem. Then again ...... it might be.

All three PSU had burn discolored pc board in the 5vsb area. Only the Bestec has the big power resistor. The worst burned area looked to me like it was around the diode.

Toasty - this "older" SeaSonic SS-250 FS does NOT have active PFC, that is - it does not have any chip for PFC but uses same chips as Bestec. If it has passive PFC, I have not identified it yet. (I'm still just a newbie with a new blue esr meter)

[everell]

Re: Bestec revisited

Scope photos of Bestec ATX-250 12E at isolator pin 4 showing:
Normal operation with good capacitor - about 11 volts DC
Capacitor removed to simulate open capacitor - about 1 volt peak oscillations
Resistor 100 ohms to simulate leaky open capacitor - about 14 volts peak oscillations

[Scenic]

Re: Bestec revisited

ouch..

[Toasty]

Re: Bestec revisited

ouch..

Huh?

[Wizard]

Re: Bestec revisited

That's good, can you repeat this with scope on the standby 5V output and voltage to make it more sensible to us?

Toasty, yes it is OUCH. Typically the optisolator is usually quiet and little ripple.

Cheers, Wizard

[severach]

Re: Bestec revisited

Isn't the best test to put a 10 or 100 ohm resistor in series with a good cap?

Here's another not yet failed ATX-250-12E Rev P7 I proactively replaced today. The shading around the standing up diode D7 is not bad lighting or glue. I think its where the heat damage is starting. Notice how the leads on D7 are mottled and gray and the glue is baked. The brown lead section nearest the diode still has some glue. The light gray part is where I chipped away the glue with my fingernail. The gray remainder going down to the board never had glue. The Jumper J45 is even a bit dark. The leads on D6 to the right and D33 behind are still shiny. No heat damage on the leads of D7 on the back.

This Bestec Rev P7 C1 10uf 50v measures 16 ohms.

In another not failed Rev P7 C1 measures 3.5 ohms.

In another not failed Rev P7 C1 measures 15 ohms.

Random brands of 10uf caps in my cap box average 2.2 ohms. Looks like I caught today's serial killer supply just in time. I'm replacing them with the "new and improved" Bestec.

[Toasty]

Re: Bestec revisited

Hmm... I thought the "ouch noob" was a reference to a mistake that Scenic thought Everell made when testing. I couldn't figure out what was wrong with the readings as they looked to be what I expected from the cap failing. Although, I really hoped to see some really high peaks to help explain the failure.

Understand what it is supposed to be, but the test here was to see what peaks were showing up there to discover if the cap was under-volt-rated, and if that was what causing the failures. Since it doesn't seem to exceed the caps voltage rating, are we left with just a crappy cap or heat or poor placement issue or a combo of the three?

At least a quick scoping of the opto now will tell if the cap is bad.

@Severach - I wonder if that diode is overloaded, or failing because it isn't fast enough (wrong spec) for the circuit?

Going by the schematic from Everell, that diode sits between ground and the transformer winding that feeds (thru center tap) the cap that fails with the other end going to D33.

[everell]

Re: Bestec revisited

Scope pictures of Bestec ATX-250 12E in a more orderly fashion
Good Cap C1
No cap C1 to simulate open
100 ohm resistor to simulate open capacitor with some leakage
47 ohm resistor to simulate open capacitor with leakage
6 ohm resistor to simulate open capacitor with lots of leakage