the gutless, bloated, and fried power supply hall of shame

[PeteS in CA]

It might do 200W continuous without launching. Efficiency will probably be in the 70% range, maybe 75%. If those output inductor cores get hot they will degrade pretty fast.

[RukyCon]

Amazing to think it has been over a year since the last post here, so in an attempt to end the dry spell, here's an overview of a GameMax GM-400M "400W" SFX PSU. One of over a dozen I recovered from the trash. As to why, well, that's something you'll have to figure out.

First and foremost, the label:

Not a lot to say here, other than it being mostly in Chinese and the input voltage only being 200-240AVC, despite the PSU having a voltage selector switch.

Board Overview:


Now we're starting to see some of the shortcomings of this power supply, which I'll be going into in a bit. The soldering looks pretty decent for the most part, and the lack of SMD parts for such a modern unit (manufactured in July 2018) is intriguing.

Primary Side:


Input filtering consists of two 1nf 1Kv ceramic caps between line/neutral and ground, a tiny 5A fast blow fuse, and a 5D-9 NTC thermistor, with positions left for an X2 capacitor and common mode choke, both of which have been omitted, with no room left for any PFC or surge protection components.
The bridge rectifier is a GBU406 4A 600v rectifier, which feeds right into a pair of Cheng KM 560uf, 200v capacitors, both of which test right around 330uf.
The main switchers are a pair of D304X (400v, 12A) transistors in a half-bridge configuration, with one end of the primary coil tied between the two capacitors, and the other end being switched between the positive and negative rails.
The 5VSB is switched by a HS606P, which appears to be an all-in-one switch mode controller IC, but I couldn't find any info about it.
The main transformer is 35mm wide by 30mm tall, and the 5VSB transformer is 19mm wide by 16mm tall,

Secondary Side:


The filtering for each rail is as follows (and every capacitor is a Samwei SRL unless stated otherwise):
5vSB: SB360 (3A, 60v) schottky diode, 2x 1000uf 10v with PI coil between them.
3.3v: MBR2045 (20A, 45v) schottky bridge, 2x 1000uf 10v with PI coil between them (~18AWG, 6 turns, 5mm core).
5v: MBR2045 schottky bridge, 2x 1000uf 10v with no PI coil.
12v: MBR20100 (20A, 100v) schottky bridge, 2x Chengx GR 1000uf 16v with no PI coil.
-12v: 3x? FR104 (1A, 400v) fast recovery diodes, Chengx GR 220uf 16v with no PI coil.

As noted above, the only rails with PI coils are the 5vSB and 3.3v rails.
A weird thing I noticed about the Chengx capacitors, is that, despite all being the same series, almost no sizes have matching color sleeves (browns are all 10uf, blue is 2.2uf, and green-gold is 1000uf, only exception is black-gold), which makes no sense to me.
Both toroids are both -26 type, with the main one being roughly 1" diameter, and the 3.3v one roughly 3/4" diameter.
The controller chip is a HS8018B.
The cables on this thing consist of 1x 20+4 pin motherboard connector (20AWG for all wires except for PG, PS-ON, and 3V3S, which are all 22AWG). 1x 4-pin 12v motherboard cable (20AWG), and 2x Molex/SATA cables (1x Molex & 1x SATA per cable, 20AWG).
One weird thing to note about this unit, is that the fan controller's PTC is located right next to the main toroid rather than next to/in the secondary heatsink, I guess they're expecting that toroid to run hot, though I'm not sure how well silicone glue conducts heat.

Speaking of which, here's the fan:

Don't know the make (gotta love obscure manufacturers who only use their logo on their products), but judging by the low current draw, I'm guessing this isn't a powerful fan at all.

[Dan81]

Re: the gutless, bloated, and fried power supply hall of shame

I managed to recap it. I'm ashamed but I'll admit, I had to use 680uF Seacons from a Sun Pro, as I had nothing else on hand for the primaries. All bulged caps were Koshins, aside from two 10v caps (a 2200 and a 3300) which were Jun Fu and F-Yu respectively.

The 5vsb side got a 1500uF 6.3v Panny FL and a 1000uF 16v Samxon GF.
The 3.3v got a 3300uF 6.3v Sanyo WF and a 1500uF 16v Ruby MBZ.
The 5v got a 2200uF 16v cap (forgot the name but it wasn't a F-Yu, rather something else close to it in design) and another Rubycon 1500uF 16v MBz.
12v got a single 2200uF 16v Samxon KM cap as I had nothing else better on hand.

[Dan81]

Re: the gutless, bloated, and fried power supply hall of shame

Popped off the covers and they're bulged. Guess I'll have to figure out how to disable the PFC (it seems their way of PPFC was to use a voltage doubler, which cooks primary caps like crazy) and then move on to the rest of the unit.

As for output... that little? The transformer says (and I have no reason not to believe it, considering it looks unusually big) it's a 39 size, not 35 (and I can tell if it would have been a fake 39)

Main BJTs also look to be of the TO-3P size (couldn't figure out what they where, tho.). Didn't check what they were (as I didn't start removing caps yet, just a quick dusting session and lubing the YateLoon fan) but I suspect anything between 13009, D209L or 2SC2625 at the very least, putting it more close to the 350W output power IMO.

[TechGeek]

Re: the gutless, bloated, and fried power supply hall of shame

Press on the covers. If they give way and pop back up, the primaries are fine. If they don't give, bulging caps. Looks like a solid 250-300W supply after you put good caps in it, lube the fan, and dust it out.

[Dan81]

Re: the gutless, bloated, and fried power supply hall of shame

Another classic I snagged. CWT ISO-500PP, rated for 420W.

Primaries are also bulged (unless it's just the plastic covers that tend to make them look bulgy?), and I think that voltage doubler (I assume) has to do with their demise....

The secondary side doesn't even surprise me anymore.

I figure it'd be worth saving, despite being a '07 unit. ISOs are getting rather scarce - so far I've only seen them bundled with Foxconn cases, of all things. (though at least in two cases they broke the formula - one was a rather gutless FSP unit, the other was branded... Premier, the L&C kind.)

[Dan81]

Re: the gutless, bloated, and fried power supply hall of shame

Celeron would be the most plausible answer - it's dated 2009, according to the PCB underside.

Funny thing, it has a detachable 8 pin PCI-E plug - the most useless thing on the whole unit by a long shot.

Just for fun, I'll look into TELVM's Tooq experience - I remember him making a separate PCB to add PI coils - I also have a few parts from various scrapped Deers of roughly the same design (the 2005 design, not the inverted side 2003 design) so if anything, it'd be interesting to see how can I upgrade it ever so slightly.

[dmill89]

Re: the gutless, bloated, and fried power supply hall of shame

Another shocker of the ancient classics. Alpha LC-8500BTX.

- Fuse is rated 6.3A... no way in the 9th circle of hell.
- Heatsinks are EVEN MORE anorexic than the older Deer designs
- one cap per rail. EW.
- no space for any kind of filtering

I wonder what was it powering that low to not turn the PCB into coal. It barely looks to have been used, dust aside.

Looks good for up to about 100W, and that's sketchy at best. The model number has an 85 in it, so maybe it's an 85W supply that's been overrated to the moon and back twice and a half. Besides, it's automatically a parts unit, and junk parts at that. No room for improving that anorexic design and I bet that I could turn the case into a metallic taco with minimal effort.

^This, I imagine it was probably in a low-end build with a lower-TDP CPU (maybe a Celeron/Pentium or dual-core I3) and integrated graphics. Such a system would probably only draw 20-30W at idle and 60W or so under load, so even a PSU like that could handle it without cooking itself (though I wouldn't be surprised if the ripple was still out of spec).

[Dan81]

Re: the gutless, bloated, and fried power supply hall of shame

Honestly, I'll probably keep the case in the event a low-profile heatsinked CWT ISO might turn up. I have a few AC filtering parts proprietary to Deer that could help so not entirely a loss here - I only paid about $2 and it was specifically to gut it for its case, lol.

As for wattage, as typical of me - 0 watts, as usual.

[TechGeek]

Re: the gutless, bloated, and fried power supply hall of shame

Another shocker of the ancient classics. Alpha LC-8500BTX.

- Fuse is rated 6.3A... no way in the 9th circle of hell.
- Heatsinks are EVEN MORE anorexic than the older Deer designs
- one cap per rail. EW.
- no space for any kind of filtering

I wonder what was it powering that low to not turn the PCB into coal. It barely looks to have been used, dust aside.

Looks good for up to about 100W, and that's sketchy at best. The model number has an 85 in it, so maybe it's an 85W supply that's been overrated to the moon and back twice and a half. Besides, it's automatically a parts unit, and junk parts at that. No room for improving that anorexic design and I bet that I could turn the case into a metallic taco with minimal effort.

[Dan81]

Re: the gutless, bloated, and fried power supply hall of shame

Another shocker of the ancient classics. Alpha LC-8500BTX.

- Fuse is rated 6.3A... no way in the 9th circle of hell.
- Heatsinks are EVEN MORE anorexic than the older Deer designs
- one cap per rail. EW.
- no space for any kind of filtering

I wonder what was it powering that low to not turn the PCB into coal. It barely looks to have been used, dust aside.

[Dan81]

Re: the gutless, bloated, and fried power supply hall of shame

The ones used there aren't the extended kind of T, if anything. Oh, and I can definitely attest they're thick. The decent Sun Pro units usually don't skimp on that, from what I've managed to see over the years - it's only the truly rock bottom units that have heatsinks as thin as paper.

And since I forgot mentioning in the original post - thank goodness the unit I have at home DOES NOT have that sideways-mounted AC receptacle filtering. That's just asking for trouble six ways to Sunday. Instead, it's the classic receptacle mounting style as seen on the Husk unit (yes, it's about the same size of the one found in the Husk unit, actually, except the guys at RPC went a step further and, while still kinda anorexic, they used proper safety-rated Y caps on it)

As for what happened to the two new units I've recently posted? The Husk unit is still alive and used for testing since it's rather quiet (surprising, especially since it looks to be older than the Spacer), while the Spacer got gutted. I saved the Spacer case since it looked pretty unused (2018 datecode - all I got to say is WTAF??) and scrapped the PCB to harvest parts that would "maybe" help me fix the Raidmax which as of this post, is still not turning on whatsoever, albeit 5vsb being there. Maybe one of the other two rectifiers are gone? That wouldn't make much sense, especially since it came with a molex2PCI-E adapter... which is supposed to eat power from 12v.... right?

[PeteS in CA]

Re: the gutless, bloated, and fried power supply hall of shame

I think I'd expect heavier stamped & bent heatsinks to do a bit better that the thin Tees. If you think of the metal as a heat conductor with thermal resistance, the Tee will have a higher resistance - being less thick and the longer distance to the section crossing the Tee (I'm assuming a single-piece extrusion, not two extrusions that slide together to form a Tee ........ = .

[Dan81]

Re: the gutless, bloated, and fried power supply hall of shame

Honestly, I'll pass the challenge and let uN1Qu3 do it. I plan on saving my Husk unit for some low power machine w/ a decent low power GPU (thinking of a 1150 w/ GT630 or so - the 630 has kept me entertained for as long as mid 2020, when I sold it with a 775+DDR3 PC I had to make room for my actual i5 3470 + 650) and if for whatever reason the actual toroid isn't enough, I either can go and swap the torroid from a Delux unit, or just go ahead and do a complete overhaul on the ATX harness of the 380W Spire (or even 450, who knows - again, I got that one without its original case, but a generic RPC 55P00P casing was used to house it, and it's already slowly crumbling... shows how good RPC casings are, smh >_< ) I have.


This is what mine roughly looks. I couldn't get a better picture (I'm currently in Arad w/ college exams, and the unit itself is at my home, 800km away...) but to roughly put it, these would be the differences:

- mine already came with the primary caps recapped - instead of whatever craptacular caps Sun Pro used (e.g Seacon in the Raidmax I have to fix), mine already came with Panasonic CE primaries (I think they were 680uF or so... not sure but they were pretty big)

- also has a amber 5vSB LED - that's something rather unique (no pun intended) for a Sun Pro

- IIRC, main sillicon is 2SC2625 on mine - the reason for this is that I had to steal the original D209Ls for another unit (that may be scrapped altogether - Rhino/Raidmax RX-500K, which already got a lower end Sun Pro unit until I can get some transformers for that one too - it's that awry kind of Sun Pro without center tap - the original unit does have center tap, unfortunately...) and without much options left, I had to reuse those 2625s from a scrapped Wintech unit.

- secondary caps used to indeed be GoldLink caps if I'm not mistaken (this one predates the Sapcon/SC caps era by quite a bit I think...), all I know is that I recapped it with mostly Panasonic FL for the 5vSB section, Rubycon MBZ for anything up to 5v, and 2008 datecode Nichicon HD caps for the 12v side

- fan is still the original "Spire" fan (which, being a Sun Pro, I doubt they used anything else than the "TE BAO PLASTIC METAL" fan that we all know - they're honestly not bad, and that thing is actually having a proper thermal control, after I extended the leads on the thermistor to reach the 12v side.

-for 12v, not sure on this either (it's been some time since I opened it) but I definitely do remember it actually having TWO 12v rectifiers in parallel, about 20A each. A few of the 2005B iterations (namely the TB1.3 variant which I also own) made during late '07 - early '08 don't have the PCB marked for more than one 12v rectifier. The TP3.0 inside my Raidmax (which uses a newer peculiar chip named AT2008B) does have those spaces, at the expense of a ridiculously small fuse, while everything else is absolutely unusually oversized for a Sun Pro.

[Th3_uN1Qu3]

Re: the gutless, bloated, and fried power supply hall of shame

I guess they mean to say it's an underpowered POT (piece of trash) for your PC.

And with a name like that... just overload and send it into space!

That is what you should do with anything with the name Spacer on it. They can't even give you a keyboard where there isn't at least one key sticking down, from the second you crack the seal on the box.

Oh, I challenge you to build up that unit and pull 250 Watts through it. Heck, even 200 Watts continuous, if you dare.
If you replaced absolutely everything that's in it right now (BJTs, traffos, heatsinks, rectifiers, output toroid, and output caps), then yes, MAYBE it will. But at that point, you've almost built a PSU from scratch, lol.

Idk about Dan but I accept the challenge. More so, I am going to go a step further and claim I'm going to pull 250 watts through it using the stock transformer and primary power stage.

Secondary side, output wires, output filtering (including toroid) and input line filtering are all game though. Anyone else in?

I'm going to toss in a prize too - my prize is a Deer/Allied/L&C of my choice in working condition. I still have a few that actually run and haven't been taken apart more than to clean the fan or replace 1 or 2 caps.

[Dan81]

Re: the gutless, bloated, and fried power supply hall of shame

The Husk is on my "to work on" list anyways. The other one would be the Raidmax posted a while back - I still haven't figured out what's awry with that one that I only get 5vSB and nothing else. It's having Samxon KMs now, and a 12A part on 12v for the time being (it will be replaced once I get it running - PCB does have space for two 12v and two 5v rectifiers, so it's probably a sign Raidmax did indeed slowly step up their game from the criminally gutless RX-380K you had. That, and my unit is probably the first and only to ever sport an AT2008B chip instead of the 2005B that most Sun Pros boast) and yet it won't turn on besides 5vSB.

[momaka]

Re: the gutless, bloated, and fried power supply hall of shame

For the PFC coil, unfortunately I won't crack it open.

And you don't have to. Just use a transistor tester on it if you have one. It should give you the inductance and resistance... and from that you may infer at least some numerical data vs. just looks fake or not fake.

All things considered - the Husk unit would make a safer 250W unit than the Spacer. It's truly sad when a usual overrated unit manages to outperform something that was actually rated to do 250W.

Oh, I challenge you to build up that unit and pull 250 Watts through it. Heck, even 200 Watts continuous, if you dare.
If you replaced absolutely everything that's in it right now (BJTs, traffos, heatsinks, rectifiers, output toroid, and output caps), then yes, MAYBE it will. But at that point, you've almost built a PSU from scratch, lol.

[Dan81]

Re: the gutless, bloated, and fried power supply hall of shame

Nah, the Spacer one is really right about itself - No user serviceable parts inside. Well, maybe the fan, lol.

Again, compared to the Spacer, the Husk unit does at least have spaces for a few PI coils (jumpered, of course - but hey, they exist at the very least!!), and does use a IC for the 5vSB, so this would be a redeeming quality. Oh, and unlike the Spacer oe LC-Power, this one does allow for a TO-3P 13009/D209x part to be used on it - ironically, Stinkworlds should have these plenty.

Finally, sleeved main cable - the Spacer isn't even sleeved. And did I mention the Spacer has even thinner wire harness than any unit I have? It feels like it's nearing 30AWG - free Kynar knockoff I guess?

As for the Spacer, it truly has no redeeming qualities whatsoever - for Christ's sake, the output torroid is as big enough as the mag-amp torroid of my Raidmax RX-500! You're expecting anywhere past 50W with that? No way. Attaining 100-120W as you suggested - impossible.

For the PFC coil, unfortunately I won't crack it open. What I can say though, is they don't seem much different than the ones I'd find on Deers (and mind you - Deer never used fake PFC coils - if anything, it would either be installed or completely absent - unlike Kissquiet and their fake PFC coil that was just a jumper wire.), and my easiest giveaway that it is indeed real - I am fairly sure that if they were to fake coils, they wouldn't go as far as adding polypropylene caps (which this one does have) to the jumper wire - so, it surprises me that the chinese finally started doing real PFC - even if it probably sucks bigtime in terms of efficiency.

All things considered - the Husk unit would make a safer 250W unit than the Spacer. It's truly sad when a usual overrated unit manages to outperform something that was actually rated to do 250W.

[momaka]

Re: the gutless, bloated, and fried power supply hall of shame

Spacer PSU that's sold in Auchan Romania. Claims "250W for 500W Desktop PC"... whatever they wanted to say with that.

I guess they mean to say it's an underpowered POT (piece of trash) for your PC.

And with a name like that... just overload and send it into space!

I barely doubt it does a single watt in spec without self immolating on power on.

Single output cap per rail. Yeah...

Anytime I see those cheap thin "T" style heatsinks on a unit, I know it's not worth my time to repair/upgrade for PC use. I still might use it for a bench PSU - particularly for experiments where I don't necessarily care about blowing up the load or where the load just doesn't care about ripple (DC motors, heater coils, large Nickel-based batteries, and etc.)

I find it ironic that the manufacturer bothered to put good line filtering on this unit, though. I guess that makes it this better for harvesting parts from it.

The fuse isn't blown - probably my photo is a bit unclear but the fuse is intact, actually.

Needs fixing then?

Again, I have high doubts this will even attain 150W without self immolating. Scratch that, anything past 50W looks like a incendiary event for it.

Oh, it will do 50W... but even with that, the ripple output will be terrible.
Personally, I'd be scared to put it past 100-120 Watts. Up to that, it might just barely hold onto dear life. Those thin T heatsinks are really too wimpy to transfer heat from the parts on them to their larger surface area on top.

Speaking of this...
I wonder what's better/worse : thicker stamped heatsinks, like on L&C/Deer/Solytech PSUs, or thinner "T" style heatsinks like this.

After all, we know surface area is king when it comes to cooling. And I think those T heatsinks might actually have more surface area than the stamped heatsinks. On the other hand, that surface area won't amount to shit if heat can't be "moved" quick enough to the rest of the heatsink's surface area. So IDK. Seems almost like a tie. Though overall, I think the thicker heatsink will do better, as at least it has more thermal mass and can take on short overload / high-power transients better.
Surely would be an interesting experiment to conduct one day, if I ever get the time for that. Currently working on testing some fake eBay 13009 transistors. You can sort of guess how that went... but can you, really?

250W.. never ever.

Well, also, everyone always seems to talk about the "overall" power output "Wattage". Yet, I bet you these PSUs will run absolutely terribly with anything more than 120-150 Watts if the load is either 5V-heavy or 12V-heavy... which is how it would be, depending if you used this in an old Athlon XP / Pentium 4 (without 4-pin CPU connection) vs. anything more modern. In my experiments, at least, I find the output toroid always takes the most toll when you start getting into high cross-loads, as it tries to transfer/balance some of the energy from the 5V rail onto the 12V rail or vice versa, depending on the cross-load. The smaller the core, the easier it is to overheat it. Though with these cheap units with very few windings, I guess that does away with some of the overloading on the output toroid at the expense of more ripple on the output. So there's no free lunch anywhere, really.

The Husk "450W" unit I also got along this one isn't much different, except for a PFC coil (which to my surprise, is actually real) and that one does allow for some improvements.

Hmmm...
You seem to run into a lot of cheap PSUs with real PPFC coils now, which makes me wonder: did the Chinese figure out that we figured out their scam with the fake PPFC inductors?
And if so, I wonder how all of these "real" PPFC inductors compare to each other and to one from a known reputable PSU manufacturer.

If you have one of those transistor tester thingies, I think these PPFC coils should have enough turns to show a pretty large inductance. But how large? - That would be the interesting part to compare... i.e. see the inductance value of these and of ones made by good PSUs like Delta, HiPro, and etc. It wouldn't surprise me if the cheaper PSU PPFC coils are cutting corners and have less turns in them. Sure, they could still be "real"... but if their inductance is really small, they would hardly make a difference.

Another consideration is what kind of wire thinkness was used inside the PPFC inductor. With the cheapo PSUs, perhaps thinner wire is used inside. If that's the case, they may run quite hot under a high load (or not, if they don't have as many turns, as contemplated above.)

The reason I mention all of these points is that because in my experience with these ultra-cheap PSUs now, it rarely seems like they cut a corner just here and there. Most of the time, it's more like they're driving in a straight line and rarely making a bend here and there to make the PSU just meet some absolute minimum spec where it can still function barely. So IDK, I tend to be very weary of these things now. The repair of the cheap KDMPower MI-X8775CD I did a while back really made me reconsider at how I look at cheap PSUs anymore. It hasn't blown yet, though I also haven't used it much (actually, at all.) Given the fake 13009 transistors I got from eBay recently, I suspect the KDMPower PSU may have been a victim of this itself. And when you start dealing with fake silicon parts, it starts to become really hard to tell what will go first or how far the PSU can be pushed.

EDIT - here's the Husk unit. This one might be a good candidate for some modifications as it feels slightly less cheaper than the Spacer posted earlier - that one is so cheap even the Molex pins come undone from the plastics themselves. I have a feeling XHY did the Husk, but not sure. The transformer markings seem familiar for XHY to me, yet not the rest of the design.

You know, maybe if you take the Spacer PSU and this HUSK, you just might have enough parts to build one slightly more complete PSU.

Going back to the discussion of heatsinks... perhaps you can take the output heatsinks of both units and join them together to make a thicker one - i.e. cut the left side of the "T" on top of one heatsink and the right side of the "T" of the other heatsink, then join them together by their flat side with lots of thermal compound. This should increase the thickness in the center. Probably too much work, of course... but fun to see if that would actually help anything (before and after results with a thermocouple?)

You can also do the following for the primary side, though beware it is pretty risky: cut the heatsink in half, and separate the two pieces by about 5 mm or more (preferably 10 mm), then mount the primary BJTs on each directly with silicone grease and NO isolation pads or spacers. This will give much better heat transfer to the heatsinks, but you have to be careful, because now each heatsink will be at the collector potential of each BJT. So you must be careful to isolate the heatsinks from everything else. - At least that's one thing to do if you want to get more "oomph" out of these cheapo heatsinks. But again, probably too much work for such cheap PSUs, unless you just enjoy doing this for fun. Oh, and the heatsink tied to the Collector of the BJT that's connected to the main transformer - that one will probably spew quite a bit of EMI/RFI everywhere, since it will be constantly jerked back and forth between +DC and -DC primary bus. LOL.

Oh, and I forgot to mention... those Canicons are pure CONs. Every time I find one anywhere, it's always grossly out of spec after a some years - even in spots that didn't run hot or weren't under stress. Total garbage!

[Dan81]

Re: the gutless, bloated, and fried power supply hall of shame

The fuse isn't blown - probably my photo is a bit unclear but the fuse is intact, actually.

Again, I have high doubts this will even attain 150W without self immolating. Scratch that, anything past 50W looks like a incendiary event for it.

250W.. never ever.

The Husk "450W" unit I also got along this one isn't much different, except for a PFC coil (which to my surprise, is actually real) and that one does allow for some improvements.

EDIT - here's the Husk unit. This one might be a good candidate for some modifications as it feels slightly less cheaper than the Spacer posted earlier - that one is so cheap even the Molex pins come undone from the plastics themselves. I have a feeling XHY did the Husk, but not sure. The transformer markings seem familiar for XHY to me, yet not the rest of the design.