Breakdown of a skimpy CyberLink CWT-320ATX 320W PSU (warning: inline images)

[goodpsusearch]

Re: Breakdown of a skimpy CyberLink CWT-320ATX 320W PSU (warning: inline images)

Best CWT psu ever:
http://obrazki.elektroda.net/54_1221423902.jpg

[momaka]

Re: Breakdown of a skimpy CyberLink CWT-320ATX 320W PSU (warning: inline images)

Are you saying that there's a separate transformer tap for the +3.3V rail in the main transformer, though?

Been a while since I took one apart, but yes I think I do remember there being a separate tap.

And are the filtering capacitors for the +3.3V rail before or after the MOSFET?

Both.
One before and two after if I remember correctly.

Also sort of surprised that they didn't cheap out more on the minimum load resistors in this unit as Deer are known to do (but that is Deer and not CWT).

Deer actually doesn't cheap out on the load resistors - they just use too low of a resistance on most rails and that causes excess heat near the capacitors (and often cooks them, especially when the fan starts to fail or get weak).

Channel Well actually got the minimum load resistors right with this unit - more or less. Only the one on the 3.3V rail is a bit too low, disipating close to 0.8W of heat. In general, I like to see 0.5W or less, and resistors rated for at least twice as much power as they are disipating.


One thing I forgot to review in this unit is the case and wiring.
The case is actually not too bad - not quite the thick stuff that good OEM PSUs use, but not the super cheap thin stuff that feels like candy wrapper foil.
As for the wiring - both the input and output was wired with all 20 AWG wire. Even worse, the input (such as the voltage selector switch) had 300V -rated wire instead of the usual 600V.

Best CWT psu ever:
http://obrazki.elektroda.net/54_1221423902.jpg

Hey, it all depends how it was rated on the label. If they said 50W or less, I wouldn't mind that much. But if they said anything more... then yeah, not good.
That must be a really old PSU, though. Look at the "vent" on those primary capacitors! (If you can even call it that). AFAIK, that's late 70's or early 80's cap vents. Also, the snubber ceramic cap looks tiny.

[Wester547]

Re: Breakdown of a skimpy CyberLink CWT-320ATX 320W PSU (warning: inline images)

Been a while since I took one apart, but yes I think I do remember there being a separate tap.

Well, I have noticed that there is one particular tap visible in the main transformer in most of those power supplies that linear regulate +3.3V, except in some very old half bridge ones where it isn't visible at all without unwinding the transformer (IE taking off the insulation tape also). I don't know if that's for +3.3V or if it's just another tap though. I don't think +3.3V has its own tap - otherwise, I don't think they'd list a +5V and +3.3V combined rating on the label.

Both.
One before and two after if I remember correctly.

Well, it seems to me that those capacitors form a PI filter, with one capacitor before and two after the ferrite coil... unless the ferrite coil isn't actually forming a whole output filter. Some other Hipro units have one capacitor before the ferrite coil and one after, which is why I ask, since the way the capacitors are arranged in that section makes it seem like they form a whole filter rather than a separate filter before the MOSFET and after.

Channel Well actually got the minimum load resistors right with this unit - more or less. Only the one on the 3.3V rail is a bit too low, disipating close to 0.8W of heat. In general, I like to see 0.5W or less, and resistors rated for at least twice as much power as they are disipating.

Huntkey weren't that great about minimum load resistors in their older, overrated half bridge units either. It was very common to see 27 ohms resistors used on +5VSB and 42 ohms resistors used on +5V. No discoloration was present on the PCB but the crappy capacitors used (KSC or FCON) almost always bulged because of it (though I guess they'd eventually fail anyway because they're that bad), and that was with either an 80mm fan directly wired to +12V or a 120mm or 140mm fan spinning fast.

As for the wiring - both the input and output was wired with all 20 AWG wire.

Could be worse (IE 22 - 26 AWG). I think the thickness of the wire is less of a concern so much as exceeding the rating of the individual pins on the molex connector itself (I believe they're rated 6A per pin on 18 gauge wires) as if the connection isn't firm enough it's very easy for the pins to develop high resistance and on a connector like the ATX12V that eventually could lead to charring as seen on some OEM Lite-on and Hipro units even. Also, I've seen some molex connectors go bad on some Hipros as well (when connected to a hard drive it would initially work but then the hard drive would start clicking and the boot process would hang until I swapped connectors... I guess that also could have been a bad solder joint but simply replacing the connector worked like a charm). Power cycling the computer ("peak" currents) really made the problem obvious and more likely to happen. Note that the molex connector's pins were not "loose" at all. Of course, all of this is easy to fix with a simple soldering or crimping job for the latter.

As for the power supply goodpsusearch posted, I wouldn't trust that to even do 50W with the ripple in spec and the voltages tight enough. And I see 1997 datecodes on the transformer so it's not quite that old. :P

[momaka]

Re: Breakdown of a skimpy CyberLink CWT-320ATX 320W PSU (warning: inline images)

The 390 Ohms and the 1n1418 sereis connection does not look right in thsi circuit compared to the Antec diagram.

I opened the PSU again today to double-check on this item, and then forgot to check it . I really should get some legal documents to show that I am retarded .

Anyways, that aside, I did find some other mistakes in my diagram.
- Emitter resistor for the 2SC5027 transistor is NOT 0.51 Ohms, but 1.5 Ohms (brown, green, gold, gold). I think when I was looking at the PSU upside down, my mind also partially flipped the color codes .
- Filter capacitor for the secondary side auxiliary rail (labeled Aux. Sec.) is 22 uF, not 47 uF.
- There is a 22 Ohm resistor before the capacitor just mentioned above.

I'm re-attaching the diagram again with all of the corrections. I also added power ratings for some of the resistors. They may not be very accurate and possibly one size higher than what is on the board (just to be on the safe side).

[Pentium4]

Re: Breakdown of a skimpy CyberLink CWT-320ATX 320W PSU (warning: inline images)

That must be a really old PSU, though. Look at the "vent" on those primary capacitors! (If you can even call it that). AFAIK, that's late 70's or early 80's cap vents. Also, the snubber ceramic cap looks tiny.

Looks like 43rd week of '97 judging by the transformer.

Look at how they installed the C3 capacitor

[socketa]

Re: Breakdown of a skimpy CyberLink CWT-320ATX 320W PSU (warning: inline images)

What do the 4 little dots, at the transformer windings, mean?

[ben7]

Re: Breakdown of a skimpy CyberLink CWT-320ATX 320W PSU (warning: inline images)

What do the 4 little dots, at the transformer windings, mean?

That shows which direction they are wound on the transformer.

[socketa]

Re: Breakdown of a skimpy CyberLink CWT-320ATX 320W PSU (warning: inline images)

So one is wound in a different direction than the other 3.
why is that so?

[Heihachi_73]

Re: Breakdown of a skimpy CyberLink CWT-320ATX 320W PSU (warning: inline images)

Looks like 43rd week of '97 judging by the transformer.

Look at how they installed the C3 capacitor

Look at how they installed the bridge rectifier... oh wait.

[socketa]

Re: Breakdown of a skimpy CyberLink CWT-320ATX 320W PSU (warning: inline images)

When the optoisolator switches on, what path does the initial current take?
Looking at the diagram it looks like it would liven up the path from the transformer winding (the one that is wound in a different direction) down through the 1N4148 diode > optoisolator > C945 transistor.
Then 320V would travel through C945.
What is the function of the C945 transistor?

[PeteS in CA]

Re: Breakdown of a skimpy CyberLink CWT-320ATX 320W PSU (warning: inline images)

The function of the 2SC945 is to shut off the 2SC5027 by shorting its base to ground. and the collector resistor for the 2SC945 is 470K, so the 2SC945 will be in saturation, with a V(CE) of .1-.2V.

On another topic, I believe that transformer is drawn as it should for a forward converter, but I believe that circuit is a self-oscillating discontinuous flyback. So I believe the dots for the secondaries should connect to ground for the +5VSB winding, and to the anode of the +5VSB rectifier for the Aux. Sec. winding.

The clues that it's a flyback is the 220uF filter cap before the +5VSB inductor, the lack of an Aux. Sec. inductor, and the probably low value of the +5VSB inductor. In a flyback, the transformer primary is the "output inductor"; that +5VSB inductor is just there to kill spike noise.

[budm]

Re: Breakdown of a skimpy CyberLink CWT-320ATX 320W PSU (warning: inline images)

The C945 is operating in linear mode to control how hard the C5027 is biased on, how much the C945 is turned on depends on the how hard the opto and the 431 (the output is sampled through resistors for feeding the 431) is turned on which is part of the feedback to regulate the output.
The AUX winding is used for Positive feedback for self osc of C5027 so the dot in the drawing has to be connected correctly for it to Osc.

[PeteS in CA]

Re: Breakdown of a skimpy CyberLink CWT-320ATX 320W PSU (warning: inline images)

So, here's how the circuit works.

At power on, current through the 470K resistor flows through the base of the 2SC5027, starting to turn it on. This puts a voltage across the winding in the collector circuit of the 2SC5027, with the dotted end being positive. This induces a voltage of the same polarity in the winding that connects to the base of the 2SC5027, turning the 2SC5027 on fully.

Because the secondary windings, dotted as I described above, reverse bias the O/P rectifiers, this charges the primary as if it were an inductor. One of two things happens to shut off the 2SC5027. Either the voltage across the 1.5 ohm resistor gets high enough to turn on the 2SC945, which then shuts off the 2SC5027, or the +5VSB reaches regulation, causing the TL431 to turn on the 2SC945, etc.. The 2SC945 should only be turned on by the 1.5 ohm resistor during the first few cycles of turn on, or if there is an overload.

When the 2SC5027 is turned off, the voltages across the windings flip polarity: the 2SC5027 is off; the two 1N4148 diodes are reverse biased; the two O/P rectifiers are forward biased, so the charged "inductor" discharges through those two rectifiers into the O/P capacitors and load. When the "inductor" is discharged, the 2SC5027 is again turned on, either through 470K resistor or by residual charge on the inductor that could not discharge into the secondaries.

Thus, the circuit is self-oscillatory, and when the +5VSB is in regulation the "on" time of the 2SC5027 is controlled by the TL431. Thus the switch frequency varies with load. With good design and good components this circuit can be very reliable (though not super efficient). The Apple II P/Ss used this topology, and it was very popular in the 1980s for under 150W (low component count = inexpensive).

[PeteS in CA]

Re: Breakdown of a skimpy CyberLink CWT-320ATX 320W PSU (warning: inline images)

Since the TL431 is the regulator, I should add that it is used as a comparator with one input connected to an internal reference, and the other comparator terminal connected to the divided down sample (by the two 4.7K resistors) of the output voltage. Thus, the opto is either off (which allows the 2SC5027 to charge the transformer primary) or on, which turns on the 2SC945, shutting off the 2SC5027.

The AUX winding to which budm referred, which provides regenerative feedback to turn the 2SC5027, is the lower left of the four windings in momaka's drawing (i.e. not the Aux. Sec. winding).

While I'm at it, if momaka drew that schematic from staring at a physical P/S, I tip my hat to him. That's a lot of work, and takes a lot of patience! BTDTGTTS.

[budm]

Re: Breakdown of a skimpy CyberLink CWT-320ATX 320W PSU (warning: inline images)

The 431 is not used as on or off for driving the LED inside the opto (using current transfer function of the opto. That is the classic design using 431, you can see that in so many SMPS regulate feedback control. You can easily monitor the current flowing through the LED which will vary.
BTW, you can change the output voltage by changing the resistor sensing 4.7K/4,7K ratio (within the certain range of course)

[Behemot]

Re: Breakdown of a skimpy CyberLink CWT-320ATX 320W PSU (warning: inline images)

I think a lot of us do, I know I do!

I rather prefer good units

[socketa]

Re: Breakdown of a skimpy CyberLink CWT-320ATX 320W PSU (warning: inline images)

Thankyou petes, for taking the time to explain the operation.

A few of questions:

1. The first paragraph seemed to mention two different windings, but looking at the diagram it seems that they have to be the same winding.
Yes?

2. "O/P rectifiers" - Is that the two diodes on top right hand side of the diagram.
If so, is it acceptable to refer to a diode a rectifier?

3. Is the diode to the 12v rail drawn the wrong way around?

4. When i hear the word induce, electrically, induction usually comes to mind - like a transformer; but can i assume that in this case, induce does not mean that?

5. What is that component that looks like a zenner diode with a tap off the side of it?
Have spent some time searching and can't find anything on it (probably because i don't know what it is called)

[momaka]

Re: Breakdown of a skimpy CyberLink CWT-320ATX 320W PSU (warning: inline images)

While I'm at it, if momaka drew that schematic from staring at a physical P/S, I tip my hat to him. That's a lot of work, and takes a lot of patience! BTDTGTTS.

Yup, drawn simply by looking at the PSU's 5VSB circuit. Doesn't take that much time (any more) - at least not after doing a few. In fact, I drew this one by copying another I did for a different PSU, and then just changed some paths and components.

That said, if anyone thinks there are errors anywhere, that could well be the case as I didn't take this from a schematic or anything. In particular, I drew the dots on the transformers that I "felt" was the right direction.

Fron now on, whenever I open a PSU with a 2-transistor design, I always draw the 5VSB circuit. Seems like there are many designs out there, and some don't have a critical 5VSB cap.

2. "O/P rectifiers" - Is that the two diodes on top right hand side of the diagram.
If so, is it acceptable to refer to a diode a rectifier?

Diode and rectifier are the same thing. However, diode usually implies that the current is small and that the diode is likely used in a circuit that deals with signals. Rectifiers, on the other hand, implies (relatively) high power (i.e. not a signal circuit).

3. Is the diode to the 12v rail drawn the wrong way around?

On my diagram? -No.
When the PC is in standby or turned OFF (soft OFF), the only thing working is the 5VSB circuit inside the PSU, which supplies 5V standby power to the motherboard and about 12V to the PWM chip of the PSU's PWM controller and BJT drive circuitry on the secondary side (not drawn on that diagram above).

Once the PC is turned ON (and hence the main PS inside the PSU), the 12V rail ofthe PSU feeds into that point labeled "12V rail". So the Aux. Sec. rail becomes the 12V rail and the PWM controller of the PSU is now powered by the 12V rail instead.

5. What is that component that looks like a zenner diode with a tap off the side of it?
Have spent some time searching and can't find anything on it (probably because i don't know what it is called)

It's a standard "431" shunt regulator. Try searching for TL431, KA431, AZ431(just to name a few). The two letters on the front, in thise case, are just manufacturer-specific, but don't really mean anything. All of these 431 shunts are equivalent in function.

[socketa]

Re: Breakdown of a skimpy CyberLink CWT-320ATX 320W PSU (warning: inline images)

Thanks momaka
Makes quite a lot more sense now.
Interesting that the 5vsb also supplies 12v - so, really its a "5V/12v(PWM & 12v secondary BJT circuitry)SB"

Would still like to get a clearer picture on it's operation
Seems that there is a language barrier present when i try to understand the following comment from post #33 - the last sentence is undecipherable to me:

"At power on, current through the 470K resistor flows through the base of the 2SC5027, starting to turn it on. This puts a voltage across the winding in the collector circuit of the 2SC5027, with the dotted end being positive. This induces a voltage of the same polarity in the winding that connects to the base of the 2SC5027, turning the 2SC5027 on fully."

Thanks for the name of the shunt regular - am going to check it out

Spent some time looking at your diagram, and used it to redraw mine (the ISO 420W that blew up.)
It was very tricky to redraw and although it starts off looking the same as yours, it eventually looks very different.
Reasonably sure that i have traced it out correctly and checked it a couple of times - it took many hours/days (see attached thumbnail)
Does it look OK?, and why is it so different when compared to your schematic?

Would like to understand how your 5vsb works step-by-step, so then i can have a go at understanding how mine worked.

[momaka]

Re: Breakdown of a skimpy CyberLink CWT-320ATX 320W PSU (warning: inline images)

Interesting that the 5vsb also supplies 12v - so, really its a "5V/12v(PWM & 12v secondary BJT circuitry)SB"

For this power supply, YES. In general, this applies to most (if not all) PSU with half-bridge topology for the main power supply. (Half-bridge = either 494, 7500, or SG6105 PWM controller). With single-transistor forward for the main power supply, the 5VSB circuit may or may not have that additional 12V Aux. Sec. supply. Depends on the supervisory chip on the secondary.

Would still like to get a clearer picture on it's operation
Seems that there is a language barrier present when i try to understand the following comment from post #33 - the last sentence is undecipherable to me:

"At power on, current through the 470K resistor flows through the base of the 2SC5027, starting to turn it on. This puts a voltage across the winding in the collector circuit of the 2SC5027, with the dotted end being positive. This induces a voltage of the same polarity in the winding that connects to the base of the 2SC5027, turning the 2SC5027 on fully."

What he means is that once the 2SC5027 begins to turn on, a voltage is induced in the top left winding (primary side) of the transformer. This makes a voltage to become induced in the lower left winding (also primary side) as well. The induced voltage in the lower left winding has a positive voltage at the dot. The voltage follows the path through a 180 Ohm resistor, then through the 1N4148 diode, and into the Base of the 2SC5027. This makes the 2SC5027 turn on "even more".

Spent some time looking at your diagram, and used it to redraw mine (the ISO 420W that blew up.)
It was very tricky to redraw and although it starts off looking the same as yours, it eventually looks very different.
Reasonably sure that i have traced it out correctly and checked it a couple of times - it took many hours/days (see attached thumbnail)
Does it look OK?, and why is it so different when compared to your schematic?

Probably different because CWT improved (or cheapened) the circuit.

Yours looks almost exactly like the one on the Turbolink I posted in this thread. Here it is:

I think you got it right, except maybe that diode on the lower-right side of the image (it's by the 431 shunt, but it's not the 431 shunt's diode). I think the direction of that diode should be the other way around, but double-check. Actually, on the Turbolink, that diode is a Zener diode, Z1. Seems to match what you have, just flipped around (and the entire circuit is arranged slightly differently).

I also have another Channel Well PSU (Rev: L, from 2002) that I haven't posted on badcaps yet. The 5VSB circuit for that looks slightly different from both the Turbolink and the Cyberlink. I'll try to post that soon, so as to complete my CWT PSU collection uploads .