Announcement

Collapse
No announcement yet.

KingWin ABT-650MM 650 Watt ATX PSU [PCB: SW8835REV:A]

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

    KingWin ABT-650MM 650 Watt ATX PSU [PCB: SW8835REV:A]

    For today's thread, I have a KingWin ABT-650MM ATX PSU. Official web page with the PSU specs is still up and can be found here:
    http://www.rmac.kingwin.com/products.../abt_650MM.asp

    I just can't help it when I see a broken / for parts PSU on eBay that's listed for cheap and looks at least somewhat interesting. Well, OK, I won't lie – on the outside, this PSU looks anything but interesting. It's a plain gray box with colorful “spaghetti cables”.




    What mostly drew me into buying it is its 2x PCI-E power connectors and the voltage selector switch… with the latter suggesting this PSU does *not* have an APFC circuit, which I consider a plus for primary-side capacitor reliability. And of course with the PSU listed for only $0.99 (plus ~$10 S&H), I figured at least the 2x PCI-E power connectors could come in handy (probably among other components), if nothing else.

    Now here is where it gets interesting (to me): the label claims the 12V rail can provide up to 46 Amps (552 Watts) and the entire main output rated for 633 Watts. Does this sound too good to be true?


    I certainly think so! Judging by the weight of the PSU, it didn't feel to be very “packed” with components. Time to see what's inside.

    Oh, but wait, teh warranty!


    Ah, whatever.


    Hmmm… Does this really look capable of 650 Watts?
    I'm pretty sure everyone will agree with this word here: unlikely.
    I mean, some of the components and the build quality itself don't look too bad, save for the bulged crap caps on the output (which I'm sure most of you noticed probably right away.) But here are two more shots for further analysis:




    As can be seen in the first shot above, there are three transformers and two large silicon parts on the primary heatsink, along with a large polypropylene cap. These, along with the KA7500c IC (PWM controller) on the secondary side are a dead giveaway the PSU uses the old, beaten-to-death, half-bridge topology. And for such topology to deliver over 500 Watts of power (at least in the standard voltages for an ATX PSU), everything will need to be a lot more beefy. Even the heatsinks (which IMO are quite respectable in size) won't do here for that kind of power. Perhaps if the secondary side consisted of just a single 12V output (with 5V and 3.3V generated from it via efficient buck regulators) *and* if that 12V output was done with synchronous rectification, I could see it possible to squeeze more power out of such an old design. But sadly, that's not the case here – secondary side has standard group-regulation setup, as evident from the two inductors on the output. The biggest issue, though (IMO), is the main transformer: it's labeled “SF-EE35 Power-1”, suggesting an ERL/EE35 –size. However, it only measures 33 mm on the top and is quite short too. . So it's more likely only an ERL/EE33 part. For the relatively slow-switching speed of a standard KA7500/TL494 PWM controller, the realistic limit for such would be more around the 300-350 Watts mark. Moreover, it's visible from the silkscreen on the PCB that there is space for a much larger transformer in there. So it looks like the main transformer would be the biggest limiting factor in terms of power output for this PSU.

    That aside, at least I like how easy it was to get the PCB out of the power supply: just remove two spade terminals that connect the input wires to the PCB of the PSU, and that's it.


    And with this done, here's more of the primary side:




    Again, we see more proof the PSU's power rating on the label is bogus: 2x 820 uF 200V primary caps (nice… but not good enough for more than 400-450 Watts, usually), and only 2x 0.1 uF X2-class caps with a single common-mode choke for the EMI/RFI filtering (these might be adequate for a 250W PSU.) Even the GBU606 bridge rectifier won't do for that kind of power without a heatsink of some sort. At least the NTC thermistor is nice and large (SCK 0512?? Looks like SCK 200512 might be the correct part number), and the input fuse is heatshrinked. Also, the 5VSB has its own 8-pin PWM controller instead of being a “dumb” 2-transistor self-oscillating circuit. Interestingly enough, this PSU may actually have OPP on the primary side… or at least that's what I think the small toroidal inductor tacked onto the middle transformer is for. I've seen old FSP/Sparkle PSUs with such design. The only question here is how high the OPP was set and if it would work before the primary releases the magic smoke.

    Next, let's look at the secondary side:


    Oh my, oh my! I wonder why the PSU was listed for parts or repair.
    - Yeah, bulging/leaking CapXon GF at their finest! However, there is actually a twist here: while we all know how unreliable CapXon tend to be, I must also note that the PSU's fan was not working at all when I got it. But I'll get to that in more detail later.
    Another interesting detail from the above shot reveals who's the designer/maker of this PSU: SuperFlower. Yes, SuperFlower made this PSU, as is evident by the two lines of text on the PCB:
    SW8835REV:A
    SF-250W-500W

    The second line suggests the PCB is good for 250-500 Watt PSUs. Therefore the 650-Watt rating on the label is clearly a false claim. And personally, I think SF intended this particular build to be a 250-300 Watt PSU, but KingWin likely just shamelessly slapped their over-blown label on it. In any case, the PSU doesn't actually perform bad at all (once I recapped it, of course.) What surprised me the most is that both the 12V and 5V rails were quite impervious to cross-loading tests I ran on the unit. Then again, I suspect the large MicroMetals T130-26 output common-mode toroid for the 5V/12V is probably the reason why. Next to it on the right is the 3.3V rail's output toroid. I forgot to measure this one, but it looks like 0.8” (20 mm) diameter. Frankly, that one is a bit on the small side… but at least it is a -52 core mix (so possibly MicroMetals T80-52.) I don't think it can do 24 Amps like the label claims, though. Also, while not visible in the picture… and I myself actually couldn't read the part numbers… but the 12V rail rectifier is only a single TO-247 part. So at best, that's probably something along the lines of a 30 or 40 Amp Schottky rectifier. If the former, then Kingwin's label is truly shameful, as that would be nowhere near the 46 Amps claimed on it. But if it's the latter, I guess one can say “kind-a close enough”. Whichever the case, there's no way this PSU will do 46 Amps continuous on the 12V rail. What I find even more scary is that apparently this PSU was bundled with Radeon HD7950 video cards on NewEgg at some point (at least according to reviews there), and those are 200+ Watt monstrosities. Luckily, the two PCI-E power connectors won't allow for a crazy GPU setup. Therefore, even with a high-end CPU, it wouldn't be too likely that the PSU gets paired to a system that can over-draw power on its 12V rail. An HD7950 and a 150W max CPU will top around 350-360 Watts… which comes out to 30 Amps on the 12V rail. So even with a 30 Amp rectifier, there's a chance the PSU could pull it off… barely. But I personally think that would be a very sketchy setup and don't recommend anyone draw more than 300 Watts total from this particular PSU, mainly because of the ERL/EE-33 transformer size.

    Finally, I have a shot of the solder side. It looks kind of messy due to the left-over flux that wasn't cleaned up at the factory. But aside from this, SuperFlower did a pretty OK job with the soldering itself. It's not very pretty, but it's functional and sound.


    Now for what I think is the most interesting part: the fan.


    Actually, it's just a boring Globe Fan, model S1202L.
    What's interesting about it (IMO) is the failure.
    Notice how the sticker is ever so slightly darker in the middle. You can also see that in the first shot of the case of the PSU. It looks like something might have overheated in there. Not only that, but as I was taking the PSU apart, I noticed that the fan could turn fairly easily, but only for a few degrees. When I forced it a little further past those points, it sounded like sand being grinded and the fan felt rough while turning.

    So I took it apart and found this:




    Rust! But that's not water or moisture-induced rust. The only time I've seen that on a fan is when its stator windings had shorted and/or burned, causing the metal to oxidize from the high temperature… or at least I think that's how it happens. Perhaps it was the other way around? In other words, maybe the rust buildup caused the fan to slow down and eventually the windings to burn out? But then what caused the rust in the first place? Moreover, I've seen it on various different fan sizes and brands (most often on small 50 mm ARX CeraDyna fans on XFX video cards that run very hot.) So my hunch here is this was not done by water or moisture.

    I suppose we may never know. In any case, that's another shorted fan for my dead pile. Or is it? 0.o

    Anyways, let's finish this PSU with the parts summary before I hit the character limit (now expanded to 15k – thanks TC! )

    Primary Side:
    * EMI/RFI filtering: two 222M Y2-class caps; two 0.1 uF X2-class caps; one CM choke
    * Protection: heatshrinked glass fuse (can't read info, though); SCK 200512 NTC thermistor; *no* MOVs
    * Other: GBU606 bridge rectifier (not heatsinked)
    * Caps:
    *** 2x CapXon LP, 200V, 820 uF, 22x46 mm
    *** 3x CapXon KM, 50V, 10 uF, 5x11 mm: 2x for BJT drive and 1x for 5VSB startup/run
    *** 1x 2.0 uF 250V P.P. for main PS transformer coupling
    * Main PS (H-bridge topology): 2x 2SC2625 NPN BJTs (TO-3P); 2x 100-Ohm 3-Watt resistors in parallel for snubber
    * 5VSB switch: UTC 2N60L MOSFET (TO-220FP)
    * Transformers: “SF-EE35 Power-1” (33 mm) for main PS; “SF-EEL19 Drive-2” (19 mm) for BJT drive; “SF-EEL19 SB-1” (19 mm) for 5VSB

    ICs:
    *LD7550bbn (PDIP-8) for 5VSB PWM; KA7500c (PDIP-14) for main PS PWM; WT7510 (PDIP-8) for supervision, PS-ON, and PG.

    Secondary Side:
    *Output Inductors: T130-26 common-mode toroid for 5V/12/-12V; T80-52 (20 mm dia.) for 3.3V rail

    * 5VSB
    *** 1x CapXon GF, 16V, 1000 uF, 10x20 mm cap before PI coil
    *** 1x CapXon KM, 16V, 470 uF, 8x16 mm cap after PI coil
    *** SB540 (5 Amp, 40V) Schottky rectifier
    *** PI coil: 13.5-turn, 22 AWG, 2.5 mm dia. rod core
    *** Load Resistor: 47-Ohm, 1 Watt

    * 3.3V Rail
    *** 2x CapXon GF, 10V, 2200 uF, 10x20 mm caps with PI coil in between
    *** 1x MOSPEC s30d40c (30 Amp, 40V, TO-247) Schottky rectifier
    *** PI coil: 2.5-turn, 16 AWG, 6 mm dia. rod core
    *** Load Resistor: 15-Ohm, 3 Watt

    * 5V Rail
    *** same exact arrangement for the caps and PI coil, except 1st filter cap is in a 12.5 mm spot
    *** 2x MOSPEC s30d40c (30 Amp, 40V, TO-247) Schottky rectifiers in parallel
    *** Load Resistor: 2x in parallel 51-Ohm, 1 Watt

    * 12V Rail
    *** 1x CapXon GF, 16V, 1000 uF, 8x20 mm cap (in a 10 mm spot) before PI coil
    *** 1x CapXon KM, 16V, 2200 uF, 10x30 mm cap after PI coil
    *** 30 or 40 Amp (?), 60V (?) Schottky (?) rectifier in TO-247 case
    *** PI coil: 2.5-turn, 16 AWG, 6 mm dia. rod core
    *** Load Resistor: none

    * -12V Rail
    *** 1x CapXon KM, 16V, 220 uF, 6.3x11 mm cap after PI coil
    *** 2x 1.5 Amp (?) diodes
    *** PI coil: 3.5-4 mm dia. rod core
    *** Load Resistor: 2x in parallel 620-Ohm, 1/2 Watt

    *Other caps:
    *** 1x CapXon KM, 35V, 47 uF, 6.3x11 mm for secondary aux. rail filter
    *** 3x CapXon KM, 50V, 10 uF, 5x11 mm
    *** 1x CapXon KM, 50V, 2.2 uF, 5x11 mm
    Attached Files

    #2
    KingWin ABT-650MM 650 Watt ATX PSU – recapping

    Up next: repairs (recap) and a few modifications. As usual, a diagram is attached below of the output side to show what caps go where for those who are new to PSU recapping.


    Before going any further, I'd like to note that the 1st cap spot in the 12V rail can accommodate up to a 10 mm diameter cap. Likewise, the 1st cap spot for the 5V rail can accommodate (although it may not be easy, due to the main output toroid nearby) a 12.5 mm diameter cap.

    And what did I do? – Exactly as mentioned above: I shoved 10 mm and 12.5 mm caps for these two spots. Sadly, it was only a partial PSU recap, because (as usual) I was low on stock for many caps. (But that's what happens when *oneself* constantly keeps hoarding all kinds of PSUs and wants to fix each and every one of them. )


    For the 1st cap in the 12V rail, I used a Nichicon HD, 16V, 1200 uF, 10 mm cap in place of the original CapXon GF 16V, 1000 uF, 8 mm cap.
    For the 1st cap in the 5V rail, I used a United Chemicon LXV, 6.3V, 3900 uF, 12.5 mm cap This is quite an upgrade over the original 2200 uF CapXon GF. I only did it because the few caps I have more in stock are typically the “fatter” ones that not many PSUs have space for.
    The 3.3V rail 1st cap received a 6.3V 2200 uF Rubycon ZLH. The 5VSB rail 1st filter cap received a Nichicon HN, 16V, 1500 uF (because it's the 5VSB, which typically is not problematic with ultra low-ESR caps… and since I have a lot of these.)
    Lastly, the -12V rail received a 16V, 470 uF Nichicon PJ in 8 mm diameter (I slightly upped the diameter here as well, since space allowed it.)
    The secondary side aux. rail and 5VSB startup/run caps were also changed with Rubycons (YXJ, if I remember.) And the 3.3V rail's 2nd bulged cap received an “OK”-looking CapXon GL cap that was originally on the 5VSB 1st filter spot (16V, 1000 uF.)

    Again, if I had more caps, I'd do the entire PS properly. But that's all I had caps for. And surprisingly, the few caps I didn't change showed proper ESR and capacitance on my GM328 tester, despite the seized fan. So they may be running on borrowed time… but at least for testing purposes to confirm the PSU is working well, my recap did the trick. In fact, this PSU performs really well in terms of voltage stability. The secondary heatsink also stayed relatively cool – and I wasn't even using the PSU's original fan here, but rather just an 80 mm fan placed loosely on top. I didn't record efficiencies this time, due to the temporary setup, but I do recall doing a few quick calculations showed the efficiency was better than most other half-bridge PSUs. Also, I tested all of the rails for short-circuit and had no problem making the PSU shut down and stay off when it detected the fault, which is good. The KDMPower I fixed/tested a while back wouldn't always shut down on the 3.3V rail with an overload, particularly if the input AC was low.

    Next, I did a few modifications to the PSU.


    Basically, the first thing I did was to remove the 15-Ohm load resistor on the 3.3V rail, because let's be honest: no one wants 0.7+ Watts of heat right between two output caps that are surrounded by hot components (heatsink and output toroid) and probably don't get much airflow anyways. To not leave the 3.3V rail without a minimum load resistor, I took one of the 51-Ohm load resistors from the 5V rail and connected it to the 3.3V rail – but not in place of the old 15-Ohm resistor. No, instead I found that the 3.3V rail runs to one of the end pins on that unpopulated connector on the PCB (for OCP board?), and next to it is also a ground pin. So I placed the resistor there – far away from the output caps to allow them to stay cooler.

    Next, the fan connector also got uprooted and connected with long wires in the air. Why? – Because I wanted to add a series resistor to drop the fan voltage (thus speed) a little, since the fan on this PSU runs quite loud otherwise (and there's actually more to it than that, but to keep it short, I'll post that elsewhere.) The long wires simply allowed me to install an 8.2 Ohm resistor in series with the fan.

    Lastly, I changed the label on the PSU accordingly to what I felt it could do comfortably.


    Most notably, I dropped the current on the 12V rail from 46 Amps to a mere 25 Amps Yes, that's quite dramatic, but this is what I think the 12V rail can really do continuously all day without trouble. Also, 25 Amps works out to 300 Watts, which I think is at the upper end of what the main transformer can do continuously.
    After this, I dropped the 3.3V rail current… in half (12 Amps). In retrospect, I think this might have been a bit too conservative. But I honestly don't see how that small output toroid can handle 24 Amps. The windings on it look like 18 AWG… or maybe 16 AWG wire at best.
    And finally, I dropped the entire power rating of the main PS to a much more “sensible” 333 Watts. Why 333 Watts? – Because it felt like a magic number that day. I figured if KingWin could pull numbers out of a hat, then why shouldn't I too? Except, I kind of did it more in the opposite direction. And all in all, my idea was to keep the total output from the main PS under 350 Watts… so 333 Watts seemed like a cooler number (pardon the pun.) I highlighted it in yellow too, lest someone gets a hold of this PSU without my presence

    And that's as far as I got with this KingWin PSU. The seized fan ended up being a show-stopper. I thought about replacing it (and I should have! ) But instead, I did something unthinkable that no sane man (or woman) should do. (Dear fellow readers of the “Worthless/Pointless Repairs” thread, brace yourselves! )

    So for now, let's just say this PSU is still “inoperable” until its fan “situation” is rectified. That should be an easy fix... once I decide which one of my 120 mm fans I want to "sacrifice" for this PSU, as most I have are 3-pin. Ideally, I'd like to get something closer to the original, though.
    Attached Files

    Comment


      #3
      Re: KingWin ABT-650MM 650 Watt ATX PSU [PCB: SW8835REV:A]

      Originally posted by momaka View Post
      Hmmm… Does this really look capable of 650 Watts?
      A wise man once said: every PSU is capable of 650w, once!

      Now that I read all this if anyone asks me today if I read any books I can honestly answer yes
      Good work momaka, please don't disappoint with the fan
      "The one who says it cannot be done should never interrupt the one who is doing it."

      Comment


        #4
        Re: KingWin ABT-650MM 650 Watt ATX PSU [PCB: SW8835REV:A]

        Nice job, as usual.


        If you're up for it, maybe reduce the timing cap for the '7500. Increase the freq & run the output inductor cooler; stay below ~50 kHz with that -26 core.

        Yea, this one's a sleeper. The 3.3 has two magamp inductors feeding its rectifier. You can see them going from the +5 rectifier (& trans winding) to the 3.3 one.
        So it doesn't "half wave" during light 3.3 load & the usual single magamp inductor typically used.

        Edit: if you change any more caps, the two base bias ones for the main switchers should be on your list.


        Good thing you reduced the rating by half; 333 is half of what???
        Last edited by kaboom; 03-30-2022, 03:07 PM.
        "pokemon go... to hell!"

        EOL it...
        Originally posted by shango066
        All style and no substance.
        Originally posted by smashstuff30
        guilty,guilty,guilty,guilty!
        guilty of being cheap-made!

        Comment


          #5
          Re: KingWin ABT-650MM 650 Watt ATX PSU [PCB: SW8835REV:A]

          Surprised to see Kingwin (which sounds anything but familiar.) using CWT (IIRC, wasn't VIKING transformers CWT? I remember some Antecs had those VIKING trafos as well.) as their OEM. Unsurprised to see them using CapXon though.

          And FWIW, EE35 could be an 35 sized transformer crammed into a EI-33 package? As in same windings of an 35 transformer but crammed through the smaller 33 casing.

          Although, it could also be an EI-40. I've seen those in some rather low to mid-end Leadman LP-7700A units. The heatsinks weren't that anorexic (almost on par with Bestec 250W) and I was surprised to find they used screws to hold the heatsinks down. And the EI-40 traffo was definitely real.
          Last edited by Dan81; 03-31-2022, 12:18 AM.
          Main rig:
          Gigabyte B75M-D3H
          Core i5-3470 3.60GHz
          Gigabyte Geforce GTX650 1GB GDDR5
          16GB DDR3-1600
          Samsung SH-224AB DVD-RW
          FSP Bluestorm II 500W (recapped)
          120GB ADATA + 2x Seagate Barracuda ES.2 ST31000340NS 1TB
          Delux MG760 case

          Comment

          Working...
          X