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

[Wester547]

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

Yes, as far as I remember, HiPro (at least those 250W units) have a separate rectifier for the 3.3V rail, hence a separate winding on the output toroid, and separete filter capacitors. This then goes through the MOSFET to regulate down to 3.3V. I'm not sure what the input voltage is exactly. Could be very close to 3.3V, or could be as high as 5V.

Pretty sure every Hipro that linear regulates the +3.3V rail works that way. Are you saying that there's a separate transformer tap for the +3.3V rail in the main transformer, though? I thought it was taken from +5V's tap as they both share the same winding in most older power supplies that use the magamp circuit for +3.3V at least (unless you mean the voltage of the winding itself in the magamp). And are the filtering capacitors for the +3.3V rail before or after the MOSFET? It looks to me like they're after and that they're there to further smooth and regulate the voltage output. 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).

[momaka]

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

Speaking of Hipro and linear regulation, wouldn't the main toroid still be used for that, in Hipro units that linear regulate the +3.3V rail

Yes, as far as I remember, HiPro (at least those 250W units) have a separate rectifier for the 3.3V rail, hence a separate winding on the output toroid, and separete filter capacitors. This then goes through the MOSFET to regulate down to 3.3V. I'm not sure what the input voltage is exactly. Could be very close to 3.3V, or could be as high as 5V.

On most cheap PSUs with linear 3.3V rail, it's the rectified 5V rail is used for the generation of the 3.3V rail.

Here you just get 3.3V straight out of a tap on the transformer, so as you might have seen in c_hegge's review, the 3.3V regulation can be quite horrible depending on how you load the PSU.

Not that much worse compared to what my father bought a while ago (and it died soon after) : https://www.badcaps.net/forum/showthr...561#post400561

Yeah, that one was quite a crappy PSU as well. If I had to pick... ah, mixed emotions here ... maybe run them in a ghetto parallel setup . Then you get the best (and worst) of both PSUs .
And the possibility of double fireworks.

By the way, I just got another computer with one of these CWT PSUs. It's still sitting in the trunk of my car, but I'll go tear it apart in a little bit. May be a while before I post pictures of it, though. I still have that other TurboLink PSU that I need to post. It's funny too, though nowhere as bad as this CyberLink.

[mariushm]

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

Not that much worse compared to what my father bought a while ago (and it died soon after) : https://www.badcaps.net/forum/showth...561#post400561

At least this psu has room to add capacitors and inductors for filtering, while my model had none... but it had a dedicated 5v SB chip.

[Wester547]

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

Wow, that is one nasty power supply. I can't think of anything positive to say of it maybe besides the fact that it doesn't have finless heatsinks, but even then I'm not sure the fins aid the heat dissipation of the rectifiers and bipolar transistors all that much given the fact that the base is so thin (probably not the thinnest ever, but thin enough for sure).

Speaking of Hipro and linear regulation, wouldn't the main toroid still be used for that, in Hipro units that linear regulate the +3.3V rail, unless the linear regulated +3.3V rail has its own transformer tap? I say that because before the voltages in the transformer windings are rectified (by the rectifiers) and stepped down by the toroid to much lower voltages they are often quite high (though much lower in forward compared to half bridge), so the idea of having the +3.3V rectifier directly connected to the MOSFET that regulates it down would probably squander an inane amount of heat even at very low loads; I would think that there's either a seperate +5V output in the toroid that it's connected to or there's a separate +5V transformer winding altogether, unless both rectifiers really are used and shared between the +5V and +3.3V rail (thus directly stepping down +3.3V from the +5V winding in the toroid), but it doesn't seem to work that way in those PSUs.

Also, in these units, the fan is directly wired to +12V, so the fact that there is still discoloration on the PCB shows how much of a travesty these units really are. Okay, that might have something to do with the 3A-diodes-on-brackets-treatment, but still. These supplies also appear to be quite old - over a decade old, actually, seeing as how yours has 2002 datecodes and c_hegge's 2003 datecodes. Not sure if no ferrite coil for the +12V rail is the worst thing in the world - c_hegge reviewed a 305W Hipro that has no ferrite coil for the +12V rail at all, just two 10x30 2200uF 16V Teapo SCs in parallel (in the review that is, it's been recapped since) and the ripple voltage never touched 17mV even at around 26A load.

[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'll double-check it.

That is a nice schematic, but I think you might want a better eraser.

Yeah that eraser has started to become soft and smudges a little bit. It doesn't help that I put too much pressure on the pencil either, though.

the 5V rail doesn't have any protection (I discovered that when a component in a CD-ROM drive shorted 5V to ground - it burned a trace on the drive's PCB instead of shutting down),

I've had the same thing happen with an L&C PSU. May not be lack of protection. Could be that the resistance of the wires and connectors is too high and so the short circuit doesn't trip.

[lti]

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

Ha! I remember reading that review a while back. I'm still amazed that it made it to 200W of "RAW" power. I wonder how much further the primary could have went if you didn't overload the 12V "rectifier" .

He said the 5V rectifier failed, which was surprising. I think everyone expected the little 3A diodes on the 12V rail to fail.

Well, what do you expect? - I do have almost 7 years worth of college classes after all (and still no degree ). I keep doing the homework right but then screw up on the tests .
(and on bad days/weeks/months, I screw up both)

I can't even do the homework, and I still get good grades. I feel like people who actually know what they're doing get lower grades than I do.

That is a nice schematic, but I think you might want a better eraser.

Well, that's a HiPro - no wonder it is so under-rated. With HiPro, you do the exact opposite as you do with cheap PSUs - you take the printed wattage on the label and multiply it by two .

That one isn't that good. It runs hot, the fan is loud (it's temperature controlled, but it receives 12V after the computer has been running for a few minutes), the 5V rail doesn't have any protection (I discovered that when a component in a CD-ROM drive shorted 5V to ground - it burned a trace on the drive's PCB instead of shutting down), and the one in the second link is full of the bad tan glue. That second one is also a year newer than mine (you can see the date code on the PCB - 9840 on mine, 9933 on the one with the bad glue), and I'm surprised that Compaq was still making that computer that late in 1999.

[momaka]

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

As you say, 1uf doesnt look like it would be too critical.
So from that respect, it looks like it's designed better than many other budget psu's, unless, of course, if the designer's were trying to save money by using a 1uf when they should have really used a higher value cap.

Could be that too. The other CWT unit I have has a 3.3 uF cap.
I might try experimenting with them to see what will happen when I increase and decrease the capacitance.

What is the actual job of that capacitor?

Smooths out the primary auxiliary rail, which is used for controling the regulation of the 5VSB via that optocoupler. budm pretty much explains it in his post. It's a closed-loop system.
With 5VSB too low, the 431 shunt does not conduct, and thus voltage on the optocoupler photodiode is low. With that, the output side of the optocoupler is not conducting as well. This leaves the voltage bias on the 945 transistor's base low, so the 945 is not conducting much either. Therefore, the voltage bias on the 5027 transistor's base rises (due to that 470 KOhm resistor on the 320 V DC line and the 1N4148 diode on the primary auxiliary winding) so the 5027 stays on for a longer amount of time. If either the current through the 5027's Emitter resistor goes too high or the 431 shunt begins to conduct more (due to 5VSB rising and going a slight bit above 5V), the 5027 transistor will turn off and/or stay on for a shorted amount of time.

It's amazing that you can get 200W out of this thing when this is 75W:
https://www.badcaps.net/forum/showpo...&postcount=310
https://www.badcaps.net/forum/showpo...4&postcount=62

Well, that's a HiPro - no wonder it is so under-rated. With HiPro, you do the exact opposite as you do with cheap PSUs - you take the printed wattage on the label and multiply it by two .

[budm]

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

It is the similar set up as the one in this thread:
https://www.badcaps.net/forum/showth...?t=6999&page=2 which it uses 47uF in stead of 1uf.
Ref to SCH on this thread as shown in post #1, this is a one transistor (2SC5027) self oscillating circuit, the bottom winding on the left side provides the positive feedback for the circuit (180 Ohms, 22nF cap, 1N4148, Base of 2SC5027) to oscillate, it is also used to generated DC Voltage generated by 1N41418 DIODE and filter by 1uF cap. That DC voltage is fed into the OPTO, the more the LED in the OTPO is turn on the more DC BIAS is applied to the 2SC945 which is use to control the BIAS current feeding the Base of 2SC5027 through 470K bias resistor. When 2SC945 receive more bias, the bias current for the 2sc5027 will be routed to ground (through C and E), the amount of current through 2sc945 is determined by how much the OPTO is turn on (just think of 2sc945 as a variable resistor and as part of the bias circuit for 2sc5027), if the output voltage tries to go upm the opto will be turned on more as so the the 2sc945 so 2sc5027 will be turned on less so the output voltage will be reduced. If there are too much current through 2sc5027, then there will be more voltage drops across the current sensing 0.5 Ohms resistor which will alo turn on 2sc945 more to reduse the bias current for the 2sc5027. The 390 Ohms and the 1n1418 sereis connection does not look right in thsi circuit compared to the Antec diagram.
If the filter cap 1uF is not present (lost its capacitance) then the drive for the 2sc945 will be driven by high frequency AC which will make it swithing on and off at high frequency which will then there will be no control feedback to keep the circuit in check.

[lti]

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

It's amazing that you can get 200W out of this thing when this is 75W:
https://www.badcaps.net/forum/showpo...&postcount=310
https://www.badcaps.net/forum/showpo...4&postcount=62

Fortunately, mine doesn't have the brown glue. That PSU uses 2SC4242 switching transistors. It's hard to see the part numbers on the rectifiers, but I would guess that they used a 10A rectifier for +12V and a 30A rectifier for +5V. I have no idea how that 20mA 5VSB rail is generated.

[socketa]

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

Nice diagram

If that 1uf cap fails (looses it capacitance) does that mean that excess current will go through the optoisolator (when it is on), instead of charging the capacitor,
and hence, excess current through the transistors, which results in a greater voltage potential (since I is proportional to V) at the 5vsb pin?

(of course, i'm assuming that leakage current through the cap is negligible, compared to the effects of decreased capacitance - so the combined effect is reduced current flow into the charging cap)

And is the reason why the feedack circuit does not correct this, because it does not regulate voltage, but rather the duty cycle?

As you say, 1uf doesnt look like it would be too critical.
So from that respect, it looks like it's designed better than many other budget psu's, unless, of course, if the designer's were trying to save money by using a 1uf when they should have really used a higher value cap.

What is the actual job of that capacitor?

[c_hegge]

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

Ha! I remember reading that review a while back. I'm still amazed that it made it to 200W of "RAW" power. I wonder how much further the primary could have went if you didn't overload the 12V "rectifier"

13007s usually let the magic smoke out at about 300W load. The switchers in this PSU are rated for 1A lower. So, I would say that they could come pretty close. I recon 250W might have been OK for them if the secondary side had been up to the job.

[momaka]

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

It looks very similar to the Okia 420ATX in my original el-cheapo PSU roundup - http://hardwareinsights.com/wp/the-e...-round-up-2/9/. Needless to say, it can't deliver anything in spec.

Ha! I remember reading that review a while back. I'm still amazed that it made it to 200W of "RAW" power. I wonder how much further the primary could have went if you didn't overload the 12V "rectifier" .

No this is not a very similar PSU - it's the same exact thing. Besides OKIA, I think Athena also had a few units on New Egg before that appeared with that same fan vent and QC sticker.

OMG, Your drawing skills are amazing! 5 stars! :P

Thank you.
Well, what do you expect? - I do have almost 7 years worth of college classes after all (and still no degree ). I keep doing the homework right but then screw up on the tests .
(and on bad days/weeks/months, I screw up both)

We share the same passion for smps and especially the badly designed units :P

Yes, it's almost like a fetish - and a weird one, nonetheless

Leadman might take the cake though

Agreed.
Their designs are just crap even when they use decent components. Just look at everell's thread on the LP-6100D. I always though that would make a decent PSU. It blew under or around 200W load. And it has components and heatsinks similarly sized to LiteON.

I think the only explanation can be the fact that the fan is wired straight to the 12V, giving the thing plenty of airflow (Although the dust might void my theory) And the fact that it uses lead based solder. What fan does it use by the way?

I guess.
But that still doesn't explain how the motherboard caps (OST RLX) haven't cooked yet, despite being overloaded with ripple and heat from an inneficient SINGLE-SIDED buck converter (i.e. old and inefficient).
So I'm just going to believe that magic is possible.

The fan is the same as the one in c_hegge's review...
It's a Rulian Science RDM8025SA rated for 12V and 0.13A

Could the higher capacitance caps also explain the very minor drop in efficiency after swapping?

Yes, possibly. Or it could have been the temperature in my room. When the AC is ON, it can drop by as much as 5F.

And also I think you meant 0.356A on the diagram

Good catch. Yes, should be 0.356 A (or 356 mA as post states).
I also had a mistake in the diagram but found out just before submitting the post and corrected it with Photoshop (drew the 5027 BJT as a MOSFET).

If I make a post without a mistake, something must be wrong with me .

[Pentium4]

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

Excellent post! That's probably one of the worst PSU's I've ever seen. Leadman might take the cake though

I suspect these are Fuhjyyu rejects in disguise (since Fuhjyyu does have an “MK” series). But who knows.

Probably, it looks like they have the typical "Y" style Fuhjyyu vent. I've thought the same as with "Fhy" caps. They're usually paired with Asia'X capacitors which are re-sleeved Fuhjyyu. Plus, has anyone actually seen a non-primary Fhy capacitor?

How everything has survived is beyond me, though.

I think the only explanation can be the fact that the fan is wired straight to the 12V, giving the thing plenty of airflow (Although the dust might void my theory) And the fact that it uses lead based solder. What fan does it use by the way?

Yes, I’m sure the higher capacitance of the new cap is helping, but that Fuhjyyu TNR does appear to be developing high ESR

Could the higher capacitance caps also explain the very minor drop in efficiency after swapping? And also I think you meant 0.356A on the diagram

[goodpsusearch]

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

If the OEM spent some dimes using 680uF caps for bridge rectification then the psu maybe can do its 300W.

You should check if their capacity is indeed 680uF though cause many manufacturers like to put fake primary caps in psus.

[eccerr0r]

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

I was just looking at a 300W PSU that recently died on me, looks very similar to this in terms of heatsinks, but it had 2x680uF primaries and the main transformer is a bit larger. I think it needs new output caps...

Speaking of bad PSUs I just measured a Deer PSU, it consumes 6W (for 5VSB) while off and 16W while on, both unloaded - using a Kill-a-Watt to measure. At least it still works... not sure for how long however...

[ben7]

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

We share the same passion for smps and especially the badly designed units :P

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

[goodpsusearch]

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

We share the same passion for smps and especially the badly designed units :P

Great post and analysis!

Well done momaka

[ben7]

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

Heh, I have some "470uF" MK caps. They read about 330uF... go figure! I don't have an ESR meter, so I can't be sure if they are bad or not.

OMG, Your drawing skills are amazing! 5 stars! :P

[c_hegge]

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

It looks very similar to the Okia 420ATX in my original el-cheapo PSU roundup - http://hardwareinsights.com/wp/the-e...-round-up-2/9/. Needless to say, it can't deliver anything in spec.