Gutless wonders (pics and discussion)

[Behemot]

Bought few of these wonders for about 5 bucks on discount to check. Fortunatelly I've tested them right as they arrived on my loading tester. Guess what I found They can manage up to 3 A with voltage already under 12 V. At 3.5 A it started to heavily oscillate and than it began restarting itself. I asked 40 % back as these are 3A bricks at best. How close I've been to the truth

LX1205_0.jpg (139.63 KiB) Viewed 34028 times

So I cracked it open - which was very easy to do actually - to find most likely the worst carnage I've ever seen. Missing components? Forget about that, this is better Well, is it?

LX1205_1.jpg (142.82 KiB) Viewed 34028 times

Something missing? Forget about missing components, there is half of the board missing

LX1205_2.jpg (149.62 KiB) Viewed 34028 times

Oh yeah grounding/earth reference my ass!

LX1205_3.jpg (86.32 KiB) Viewed 34028 times

Where did the cables go?? The rectifier is actually capable of delivering the power, I think its some 10 or 15 A part. Not sure what the switchers is right now, can check later, but as long as it provides at least 1 A, it should also be OK (on 230 V). But the transformer is too small…

LX1205_4.jpg (90.71 KiB) Viewed 34028 times

Oh, look, they are even using quality caps. But wait! What's the silicone glue doing on the top of it? Using? Maybe re-using aren't we? I mean 31st week of 2008?

After this experience I resigned on even trying chinese makers ever again. If you approach them directly, the prices are not really nice already, if you'll want quality upgrade, you'll pay even more and with all those stupid import costs & taxes you end almost on the price of really quality stuff…yeah, you make 2 bucks more but have to hold the warrant all on yourself. Nah, not doing that. I've started looking after quality bricks from european distributors instead. As for the 5A ones, I think Seasonic will be the choice, already got one, just have to find the time to crack that one open…

[ArchOlson]

Hi..i am a new user here. Its true what you said. You should always go for the better one instead of cheaper Chinese products. Ofcourse the quality matters in the power supply. First thing is the transformer and the capacitors also plays the main role on the quality output. So transformer should be of good quality as well as the capacitors used for filteration.

[Behemot]

The problem is, absolute most of the market wants cheap crap for 10 bucks. Kinda the same situation as with ATX PSUs some 10 years ago. The power brick market has to evolve as the ATX PSUs did…

[Behemot]

I was to have a quick look at this chinese wonder, R-Senda SD-1600EPS. As usually, there is no word about this model anywhere on their site or even on the Internet so who knows what the heck this is. Also no 80 PLUS certification papers available from Ecova so it is likely bogus, though it does provide slightly over 80 % peak efficiency (close to bronze actually). Close to maximum power it was barely 80 %. The maximum output is about 1300 W than it shuts down. I wonder what is actually the rated power, 1800 W, 1600 W, 1600 W peak or what?

Notice the complete lack of input filtering, there is not even place for it, with the exception of the three Y ceramics (only two of them actually populated). The ripple on two rails I was looking at (+5 V, +12 V) was high over 200 mV Uses ancient CM6802.

[LongRunner]

Well, for a gutless wonder it's fairly impressive.

But I'm not sure what the point of making it is; the manufacturing cost of it is obviously much higher than a typical A-Power or Powmax, so anyone willing to pay that much really ought to just get a decent unit of whatever wattage they really need. And it's the first unit I've seen with a relay to bypass the NTC thermistor, but without input filtering.

[Behemot]

Been told it costs 1/3 of EVGA G2 1300 W.

[Behemot]

Not that I am that surprised, but just noticed while cutting of some SATA cabling from dead crappy PSUs for making molex-to-dual-SATA adapters, they are not satisfied with using small wire gauge, they even fake it on top of that.

Had one very thin set which made me curious if that's even gonna be enough for two enterprise drives (I'll rather use it just for SSDs I guess) but it states 20AWG on the insulation. Than another one also 20AWG but by eye-meter at least one third thicker, if not twice the gauge, this is surely genuine 20AWG.

So if you see very thin cables next time, check if the spec on it is not faked

[LongRunner]

As we reach the 10th anniversary of c_hegge's first el-cheapo round-up, I decided to have a go myself; I don't have a proper load tester yet, but I have more than enough resistance wire (from broken heaters) to rig up whatever loads I need I briefly mentioned them in my thread regarding the 'proper' (which is to say, usable in an old computer after recapping) Macrons, but anyhow, I dug this out of my shed (hence the corrosion on the screws and coils) to put out of its misery: It has 2SC4242 switchers (7A rated, normally found in half-decent units up to 200W) and the secondary rectifiers are an SBL1640CT for +3.3V, SBL3040PT for +5V, and STPR1020CT for +12V (plus F06C20C for +5VSB, same as in the MPT-301); each just good enough for the label rating.
I actually borrowed the casing (not shown) from an MPT-301, but suffice to say that it's mostly the same (the gutless versions just have some single-insulated AC wires, where they are double-insulated in the MPT series). Even the fan is the same, actually. However, the main PCB has 1kV ceramics instead of Y class; I can't say I trust the transformers either, though the platform would have allowed half-decent (for the era) units to be built (the ATX9912P, and then ATX0103P, just revised a few things from it). Primary capacitors claim to be 470μF, but actually measure 330μF ; other than that, much the same capacitor choices as in the MPT series (just a few different values). Only the +3.3V wires on the main ATX are 18AWG, the rest are 20AWG. In this unit I replaced the 2A input diodes with a PBL405 years ago, but I don't think that changed the results too much Nor should my removal of the −5V cap.
I tested yesterday, but waited a day so I could take the photos in natural light.

At the oscilloscope connection point, I used 10μF 25V X5R ceramic capacitors (with wire leads, made by TDK) which I have to hand, a bit optimistic compared to ATX specification (100nF ceramic + 10μF electrolytic) but if anything perhaps a better match for what modern hardware actually has (though I would still follow the spec for any official review); since I don't have the connectors, I just cut and hard-wired for now. I've also rigged up a 3Ω mains series resistor (inside the casing from a 150W PS/3 PSU, using the original inlet and pass-through outlet and a fan which can be externally powered; perhaps an internal 12V SMPS in the future) to limit fault currents to below what will trip my circuit breaker (C16; to do it properly, I'd set up a dedicated circuit for testing).

I arranged +3.3V 10A, +5V 15A×2 (one half switched) and +12V 8A loads; altogether, slightly under the claimed rating (give or take).

Even so, +12V ripple was well over the limit (although +5V ripple was within spec), though this may be partly down to the age of the caps (even so, they don't measure terribly bad on the Atlas ESR+). It failed quietly after 1 minute 17 seconds - one switcher shorted collector-to-base (you can see that its insulation bushing is melted slightly). Despite that, the other switcher and the fuse remained intact (and +5VSB still worked). But no fireworks, what a pity

An even more-gutless Macron was sold as a Bliss MPT-A250W (other Bliss units from other OEMs can be found among c_hegge's reviews), based on the ATX9806A-P platform (which used a 7805 for +5VSB, 0.8A as was a typical rating then); label claimed +3.3V 14A (SBL1640CT used), +5V 25A (S16C40C used), and +12V 10A (F06C20C used), but I didn't even bother putting it back together as it'd probably burn the rectifiers first.
Anyway, the primary capacitors (330μF "JP", not CEC or JPCON) were bulging…

For comparison, I actually donated an MPT-301 to c_hegge for the 2014 el-cheapo round-up; it (if you can't recover the pictures yourself, I can reupload from the copy I saved for reprocessing) not only delivered the label rating, but even survived an overload to 350W.

So in that respect Casing Macron seems a bit like CWT; capable of making decent units, but not above doing gutless wonders if ordered to.
Though I so far don't know of them having designed a platform exclusively for garbage…