Announcement

Collapse
No announcement yet.

Power supply build quality pictorial. part 2

Collapse
This is a sticky topic.
X
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

    Re: Power supply build quality pictorial. part 2

    Originally posted by Stefan Payne View Post
    Try a double forward with VEGA. Or x080ti from nVidia. That is aweful.
    It works but not something you'd want.
    Why wouldn't I want it? Does it do damage over long term use? Cause instability?
    I doubt it... and in most cases, it probably is so... thus, not something I'd personally care about. If it works and is not damaging to the system or causing instability, I couldn't care less.
    The only PSUs that I'd really avoid for a new build is gutless wonders and those old half-bridge group designs with barely any protection. Anything above that is fair game to me.

    Originally posted by Stefan Payne View Post
    The Power Saving mechanisms are just that brutal with modern Hardware...
    I think calling them "brutal" is a bit of an overstatement. Yes, they are quite large... but if the PSU can handle it without tripping, who cares.

    Most PCs today still die because of crappy manufacturing with RoHS materials and components running too hot (GPU, CPU, MOSFETs - usually right at edge of their limits, just to squeeze all performance possible to get top $$ to the manufacturer) and not because of not using a "newest top-design" PSU.

    Originally posted by Stefan Payne View Post
    Öhm, no, not really.
    You can't have a group to regulate if you don't have a group, can you? ;-)
    Those PSU are usually single voltage.
    Eh. It depends. I've seen both. Low and mid-range OEM PCs (with PSUs typically less than 400W) still get a beaten-to-death group design... though lately, I'm seeing more and more of those "external power adapter" "laptop in a desktop PC case" machines... which typically do output a single voltage - so you're right about those, at least.

    Originally posted by Stefan Payne View Post
    Maybe, maybe not. I know that old first Generation HEC units back in the day had a pretty brutal Voltage drop.
    Meh.
    This hardly matters, though. As long as voltage ripple and noise are in spec, nothing should get damaged or degraded over time.

    I have plenty of older PCs that have drops on the 12V rail from ~11.7x Volts under load to ~12V idle. PSUs recapped, so pretty sure noise is in spec. Never had an HDD die because of that, even though we all know that's not too good for them. The real HDD killers are high-noise gutless PSUs (or PSUs with failed caps).

    Originally posted by Stefan Payne View Post
    But why risk it when you can get lower end DC-DC units for 40€ and every mid range from pretty reputable Manufacturers have DC-DC anyway...
    Because it's not really a risk.

    ... though technically you are right in a way: if you are building a high-end expensive PC, then it makes sense to spend some extra money on a better-designed PSU. But for any plain PC build? -Just put any decent PSU that can keep the PC working reliably, close your eyes, and press the power button. Ain't going to stress over what topology the PSU is, again, if it's a decent brand PSU (with decent caps), it works, and the system is stable. Too much voltage sag? Watt-ever!
    Last edited by momaka; 04-03-2019, 07:46 PM.

    Comment


      Re: Power supply build quality pictorial. part 2

      I had to replace a failing fan on my Raidmax RX-500XT.

      Here's a few shots. I recapped the unit as soon as I got it, even if the original caps (AsiaX) were only 4 years old. New caps are UCC (don't remember the series right now) and Panasonic FL.

      Also completed the AC filtering while at it. Man, the components were totally put in haywire. I had to move the fuse, the thermistor and AC wires so I could add the common choke. While at it, also installed an EMI board from a Deer I dismantled a long time ago.

      Currently powers my main i3-3220 PC.
      Attached Files
      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


        Re: Power supply build quality pictorial. part 2

        ^ Nice work with the recap there!
        PSU looks good overall and has fairly clean layout for an older forward topology like that. Main traffo looks almost like 39 or 40 mm wide core. I wonder if they just slapped a 35 sticker on a bigger one or if it's just the pictures that make it look bigger. In any case, it's still probably a 350 Watter... maybe 400W peak. Hard to say, though. I see two ICs on the primary, so the second one is likely a PWM controller... which means this is either a single or double forward (or active clamp??). If it's the latter (DTF), it may well be capable of 400-450 Watts then.

        The only thing I don't like about this PSU is how much space they left in the upper-right corner by the main caps - they could have marked the board for more EMI/RFI components there and jumpered through them. Then you could have just installed all EMI/RFI filtering there rather than an add-on receptacle board.

        Other than that... not bad at all. Should do nicely for your i3 rig. Not sure if it will handle an HD6950 at full load, though - that's a 200W TDP card, and all of that is from the 12V rail. So your 12V rail better have some beefy rectifier that can handle the load (along with whatever the CPU pulls... which isn't much for an i3 really - like 60-70W tops.)

        Comment


          Re: Power supply build quality pictorial. part 2

          Originally posted by momaka View Post
          ^ Nice work with the recap there!
          PSU looks good overall and has fairly clean layout for an older forward topology like that. Main traffo looks almost like 39 or 40 mm wide core. I wonder if they just slapped a 35 sticker on a bigger one or if it's just the pictures that make it look bigger. In any case, it's still probably a 350 Watter... maybe 400W peak. Hard to say, though. I see two ICs on the primary, so the second one is likely a PWM controller... which means this is either a single or double forward (or active clamp??). If it's the latter (DTF), it may well be capable of 400-450 Watts then.

          The only thing I don't like about this PSU is how much space they left in the upper-right corner by the main caps - they could have marked the board for more EMI/RFI components there and jumpered through them. Then you could have just installed all EMI/RFI filtering there rather than an add-on receptacle board.

          Other than that... not bad at all. Should do nicely for your i3 rig. Not sure if it will handle an HD6950 at full load, though - that's a 200W TDP card, and all of that is from the 12V rail. So your 12V rail better have some beefy rectifier that can handle the load (along with whatever the CPU pulls... which isn't much for an i3 really - like 60-70W tops.)
          Okay, took a second look over it and here's the summary:

          - no idea what topology it is - it's powered by 2x GPT16N50D devices and a ViPER22A for the 5vSB

          -there's also a UC3845B and a lone A928A transistor.

          -main traffo is true 35 size, not 39 or 40. I could fit in a bigger Andyson transformer from a dead 525W Hiper (the one I removed APFC on - concluded the UC3842 is dud and I am not going to chase one just for it, and it lacks a lot of wires that the Raidmax has.) but so far I guess it's okay as it is.

          -bridge rectifier is a GBU606

          -I had to redo the main filtering on the board as the components were placed by a bunch of monkeys

          -main filtering now consists of: 2x safety rated Y caps (previous ones were generic disc caps), EE-25 transformer out of an long defunct Codegen, 8A fuse (original to the PSU), and a 1.5Ohm NTC thermistor.

          -PWM chip is a Weltrend WT7502A

          -secondary devices are: S30D45CS for both the 3.3v and 5v and 2x MBR20100CT in parallel

          -AC receptacle was replaced by myself from a Deer, but not before replacing the tiny X cap it had and yellow generic disc caps for a safety rated parts from a junked 315W FSP (it had dead switchers and lacked any SATA wires.)

          -fan was originally some WAM (I'm waiting on your Mary puns ) fan that had a black body with white blades - I replaced this with a transparent Deepcool LED fan (red) that I bought last year as the WAM was making a ugly squeal.

          So this is the rundown of this PSU. I'm worried if replacing the 5v part with a 20A part is going to affect this PSU's performance - replacing the 12v ones wouldn't be much of a problem - I do have enough PSUs to gut enough to replace the 20A parts with 30A parts

          Anyways, I checked the Chemicon caps, and they're KZE and KZH series.

          As for the empty space - very likely a spot for a PFC choke - take a look at the Tacens Radix ECO II 600W, which uses the same PCB as my Raidmax RX-500XT, and you'll see the choke I'm talking about.

          http://www.pcgameshardware.de/Netzte...lerie/2286693/
          Last edited by Dan81; 04-20-2019, 12:43 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


            Re: Power supply build quality pictorial. part 2

            Originally posted by Dan81 View Post
            ...
            -there's also a UC3845B and a lone A928A transistor.
            ...
            -PWM chip is a Weltrend WT7502A
            ...
            The UC3845B is a current-mode PWM IC. The WT7502A is a supervisor IC.
            PeteS in CA

            Power Supplies should be boring: No loud noises, no bright flashes, and no bad smells.
            ****************************
            To kill personal responsibility, initiative or success, punish it by taxing it. To encourage irresponsibility, improvidence, dependence and failure, reward it by subsidizing it.
            ****************************

            Comment


              Re: Power supply build quality pictorial. part 2

              Originally posted by Dan81 View Post
              - no idea what topology it is - it's powered by 2x GPT16N50D devices and a ViPER22A for the 5vSB

              -there's also a UC3845B and a lone A928A transistor.
              Two MOSFETs and a UC3842/3843/3845 IC means it's probably a double-forward design (could also be an active clamp, though... but hard to say without knowing how the two MOSFETs are connected.)

              Originally posted by Dan81 View Post
              -main traffo is true 35 size, not 39 or 40. I could fit in a bigger Andyson transformer from a dead 525W Hiper (the one I removed APFC on - concluded the UC3842 is dud and I am not going to chase one just for it, and it lacks a lot of wires that the Raidmax has.) but so far I guess it's okay as it is.
              Awww. You should really keep some UC3842 chips on spare. Even if the PSU doesn't have a whole lot of cables, it's still a more modern design than any of the Deer/Allied units you usually get. Like PeteS mentioned, UC3842/3843/3845 is a current-mode PWM IC, so you get OPP if the PSU is designed properly (i.e. won't blow up like a Deer or any other old half-bridge design if you try to overload it.)

              Originally posted by Dan81 View Post
              -secondary devices are: S30D45CS for both the 3.3v and 5v and 2x MBR20100CT in parallel
              Not bad. Should be able to do about 300W continuous on the 12V rail easily. 350W max, maybe.

              Originally posted by Dan81 View Post
              I'm worried if replacing the 5v part with a 20A part is going to affect this PSU's performance - replacing the 12v ones wouldn't be much of a problem - I do have enough PSUs to gut enough to replace the 20A parts with 30A parts
              Do NOT downgrade the 5V rail rectifiers. This being a more modern design, I expect the 12V rail is most likely just a "7V rail" coupled on top of the rectified 5V rail. This lowers the efficiency a little bit, but makes the PSU A LOT more stable when it comes to cross-loading. (In fact, it's almost impossible to make a PSU above 300-350W with group-regulated design that doesn't cross-load badly.) So downgrading the 5V rectifier will also downgrade the 12V rail current rating.

              Originally posted by Dan81 View Post
              Anyways, I checked the Chemicon caps, and they're KZE and KZH series.
              Perfect! These are well-suited for forward-converter designs.

              Originally posted by Dan81 View Post
              As for the empty space - very likely a spot for a PFC choke - take a look at the Tacens Radix ECO II 600W, which uses the same PCB as my Raidmax RX-500XT, and you'll see the choke I'm talking about.
              Ah, you're right, I didn't think of that. Usually, above 350W, manufacturers use APFC instead of PPFC. (And in general, hardly anyone makes PPFC PSUs anymore, due to costs associated with the large linear coil - Copper is expensive, MOSFETs are cheap. )

              Comment


                Re: Power supply build quality pictorial. part 2

                Ah well. All in all, it seems pretty well suited for my i3 3220, and there's room to spare for a good GPU. I'll probably hunt something in the Radeon R7 260x/GTX 750 range. These cards are pretty low power yet pack a good punch when it comes to gaming (even if in 720p - doesn't matter as long as it upscales very good on a 24" LED TV) which would help my i3 3220 a lot.

                It's nice to find that Raidmax has advanced a lot from the days when they used Sun Pro/Leadman. Okay, maybe their units aren't that bright, but they surely are a step up from the gutless units they made back in the days.

                Not sure but I think they own Recom too - I do have a unit - PowerEngine II 450. That one isn't as good as this RX-500XT, but not the bottom of the barrel either. Half bridge unit tho, and had bad golden/black AsiaX craps. That one, I don't know what caps I used... could either have been Teapo pulls from a defunct CWT ISO or some Rubies I had left from dead motherboards. I'll be sure to check.


                Meanwhile, I do have some recased units that I guess I could post up here. I have a lot of casings left from dead PSUs so I swapped various parts into them after fixing whatever defects they had. Imagine seeing a Deer only to open up and find out it's a CWT ISO

                Then there's my modded RPC 500LBL unit in which I crammed a pretty hefty nJoy unit on which I had to fix a blown to smithereens primary side, and which also recieved parts from a dead 300W Enermax EG301P-VE.

                I mostly reused the PCE-TUR caps on it, along with the original ChengX caps it came with (the board was silkscreened for D10 caps, and I've been able to cram in only one D12 PCE-TUR and even then it was a pain in the arse.) If memory serves me right, that one DOES sport a true ERL-39 traffo.
                Last edited by Dan81; 04-23-2019, 02:18 PM.
                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


                  Re: Power supply build quality pictorial. part 2

                  Originally posted by Dan81 View Post
                  Imagine seeing a Deer only to open up and find out it's a CWT ISO
                  In my case (ha, the pun!) a HiPro 250W in a JNC/Meico PSU:
                  https://www.badcaps.net/forum/showpo...0&postcount=80

                  I also have another HiPro board like that in a RaidMax RX-380K shell/case. Not sure if I posted here or not, though. Either way, the original PSU bord in that RaidMax was from an extremely gutless Sun Pro KY-480ATX PSU.

                  Comment


                    Re: Power supply build quality pictorial. part 2

                    Originally posted by momaka View Post
                    In my case (ha, the pun!) a HiPro 250W in a JNC/Meico PSU:
                    https://www.badcaps.net/forum/showpo...0&postcount=80

                    I also have another HiPro board like that in a RaidMax RX-380K shell/case. Not sure if I posted here or not, though. Either way, the original PSU bord in that RaidMax was from an extremely gutless Sun Pro KY-480ATX PSU.
                    I think I know that PCB, have a bunch of them through my HUGE PSU stash. Some come with good heatsinks though.

                    By the way, the next case swap is going to be a "Techsolo" TP-450. Maybe I can sneak in a Delta DPS-400WB A? Wonder if it's standard ATX though as it's a Fujitsu-Siemens OEM PSU and the colors are different than standard ATX.
                    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


                      FSP Fortron/Source FSP400-60PFN REV:1

                      A while back I posted about a Sparkle Power FPS400-60GN PSU that I got for recapping for someone. This is the second PSU that the same owner gave me to fix: an FSP Fortron/Source FSP400-60PFN
                      https://www.badcaps.net/forum/attach...1&d=1556410929
                      https://www.badcaps.net/forum/attach...1&d=1556410929
                      https://www.badcaps.net/forum/attach...1&d=1556410929
                      It's another old 5V-heavy unit, as evident by that 40-Amp –rated 5V rail and 235W combined power for the 3.3V and 5V rails. Output connectors suggest the same, too, as the unit has a 20-pin ATX and 3.3V AUX connectors. Unlike the Sparkle unit, though, this one also has one or two SATA connectors. As for the wires themselves, main connector has 16 AWG for 3.3V, 5V, and ground. Rest is all 18 AWG. Also note the case has a “W/ Noise Killer” sticker, so this unit should have a fan controller. But let's open it up and see:
                      https://www.badcaps.net/forum/attach...1&d=1556410929
                      Uhm… we can't see much there, can we? This PSU is so packed that you can hardly tell anything about it after removing its top cover. All I see is two very big heatsinks… which is not a bad sign, I suppose.

                      Perhaps looking from the sides will offer more insight…
                      https://www.badcaps.net/forum/attach...1&d=1556410929
                      https://www.badcaps.net/forum/attach...1&d=1556410929
                      -Nope! It's just as packed on the sides. Makes one wonder how air even moves through it. We do see one big primary cap though.

                      In any case, this PSU wasn't functioning at all when I got it (it made only slight screeching and appeared completely dead otherwise), so I had to take it apart to find the problem anyways. After battling with it for more than a few minutes, I finally managed to take the PCB out of the case:
                      https://www.badcaps.net/forum/attach...1&d=1556410929
                      https://www.badcaps.net/forum/attach...1&d=1556410929
                      OK, now we can see some features. Going by the single 450V, 330 uF Hitachi HP3 main cap, I'm pretty sure this is an active PFC (APFC) PSU. The lack of a voltage switch and presence of a “FULL RANGE” sticker on that first picture of the case above should have been a good hint too, of course. We can also see a total of three inductors on the secondary side: one big, one medium, and one small, revealing that this is a group-regulated PSU (heh, what else would it be from that time period? ) with a mag-amp –regulated 3.3V rail. The output caps are all Teapo and CapXon. What are the chances some of them are bad, you think?

                      On that note, I did find the problem. To get to it, however, I had to desolder the secondary heatsink. Given the mass on those heatsinks, I didn't think it was going to be easy to do. Fortunately, the PCB was only a simple single-layer fiberglass board, so desoldering the secondary rectifiers and pulling the HS out wasn't hard. Speaking of which, here is a shot of the soldering before I had to dig in it:
                      https://www.badcaps.net/forum/attach...1&d=1556410929
                      MEH. It's not looking very nice… but not terrible either. It only had a few leads that were slightly-longer than necessary. Overall, it's acceptable. The two daughterboards looked better, though:
                      https://www.badcaps.net/forum/attach...1&d=1556411005
                      https://www.badcaps.net/forum/attach...1&d=1556411005
                      Nice and shiny with leaded solder. Only thing that I didn't like too much on the PWM+APFC control board was the slight heating in one area near transistor SQ3… and the tan organic glue that darkens and goes conductive eventually, of course. It wasn't excessively over-applied, though, so I was able to clean a good chuck of where it could cause problems (but by no means was I able to remove all of it.) Also, speaking of the fan controller earlier, the secondary-side daughterboard contains the circuit for it. Worth noting here is that the fan driver is only a TO-92 NPN transistor (S8050 / H8050). Because of this and the fan's power rating (0.22 Amps), the transistor was running a little hotter than I'd like to see. I'll tell you how I alleviated that a little later below.

                      Anyways, like I said, desoldering the secondary heatsink didn't turn out to be that hard at all – managed to do it with my 40W Radio Shack bulb desoldering iron. Here are the parts attached to the secondary heatsink:
                      https://www.badcaps.net/forum/attach...1&d=1556411005
                      From left to right, we have: MBR4045PT (40-Amp rectifier for the 3.3V rail), E83-004 (30-Amp for 5V rail), BYV32E200 (20-Amp, 200V ultrafast rectifier for 12V), and MBR760 (7.5-Amp Schottky for 5VSB). Overall, this is good arrangement. However, I would have liked to see a stronger rectifier for the 5V rail, as a 30-Amp part on a forward-converter topology will likely max-out near 22-25 Amps. This is quite far from the 40 Amp rating promised on the label and also quite a change from the Sparkle FPS400-60GN, which uses a single 60-Amp part for the 5V rail *and* is a half-bridge design (which means one could theoretically pull the full 60 Amps from it.)

                      With the secondary heatsink out of the way, now we can see the output side… and also the problem that made this PSU appear “dead”.
                      https://www.badcaps.net/forum/attach...1&d=1556411005
                      Indeed we have bad caps, as I suspected above. And they are on the 5VSB output, which goes inline with the symptoms: screeching PSU and no 5VSB rail. That's all it took to bring an otherwise decent PSU down. On that picture above, you can also see how packed the secondary side is. So in their defense, those caps probably didn't get much ventilation. Nevertheless, even if they did, I'm still pretty sure they would have failed anyways. It's Teapo and CapXon, after all! Is anything else to be expected of them?

                      Also, in case it's not clear from the pictures above (and it isn't, I can see) – the output caps for the 3.3V and 5V rails are pretty hefty: 2x 6800 uF (CapXon KM) units for the 5V rail and 6800 uF (CapXon KM) + 3300 uF (Teapo SC) for the 3.3V rail. Meanwhile, the 12V rail gets *only* two 16V and 2200 uF caps (those blue Teapo SD.) That's plenty of filtering capacity if you ask me.

                      Finally a shot of the fan label:
                      https://www.badcaps.net/forum/attach...1&d=1556411005
                      It's just a Nidec Beta SL D08T-12PU (2 ball bearing) fan – good stuff. Since this one was on a fan controller, its bearings were still OK (unlike the one in the Sparkle FPS400-60GN, which had failing/loose bearings and was very loud.) I fixed the Sparkle PSU by using the fan from this FPS PSU (and I connected it straight to 7V to reduce noise, bearing wear, and dust accumulation). Meanwhile, this FSP PSU received a brand new San Ace 9A0812S402 fan, rated for 2.16 Watts @ 12V (that is, 0.18 Amps). The reason I made this change was to reduce the power dissipated by the fan driving transistor (that S8050 NPN BJT I mentioned above), as otherwise it was getting quite hot with the original 0.22 Amp Nidec fan. Of course, I did this change only after I looked into the datasheet for the San Ace fan above and verified that the fan would have similar airflow specs as the Nidec fan at the voltage that the fan controller was providing.

                      Anyways, that's all for this Fortron PSU. Below is a summary of the parts in it.

                      Primary Side
                      EMI/RFI filtering: 3x 1 uF + 1x 0.47 uF X2-class caps, 2x 3.3 nF Y2-class caps, 1x CM choke + 2x SM chokes
                      Protection: T8A fuse and SCK 085 NTC inrush thermistor
                      GBU805 bridge rectifier
                      Primary bulk filter cap: Hitachi HP3, 450V, 330 uF, 30 x 45 mm (ø x h)
                      APFC: 2x 11N60 MOSFETs + HFA15T60 diode
                      Main PS: 2x 11N60 MOSFETs, 2-transistor forward design (2TF)
                      Main PS Snubber: 1.5 Amp diode + (190 kOhm 2W resistor || 0.47 nF ceramic cap)
                      Transformers: 41 mm core (main), 19 mm core (5VSB)

                      ICs
                      FAN4803 (APFC + main PS PWM, primary side daughterboard)
                      some off-line switch IC for 5VSB
                      3x ET1109 optocouplers
                      Welltrend WT7510 (supervisor, secondary side daughterboard)

                      Secondary Side
                      5V Rail:
                      * 1x E83-004 (30-Amp, 45V, TO-3P) rectifier
                      * 2x CapXon KM, 10V, 6800 uF, 12.5 x 35 mm (capacitors C29 and C30)
                      * 6 mm core, 5-turn, 14 AWG(?) PI coil between caps
                      * 33-Ohm, 2-Watt resistor for minimum load (over ¾ Watt of power wasted as heat! )

                      12V Rail:
                      * 1x BYV32E200 (20-Amp, 200V ultrafast, TO-220) rectifier
                      * 2x Teapo SD, 16V, 2200 uF, 10 x 20 mm (capacitors C23 and C24)
                      * 6 mm core, 5-turn, 16 AWG(?) PI coil between caps
                      * 150-Ohm, 1W minimum load resistor connected to 5V rail

                      3.3V Rail:
                      * 1x MBR4045PT (40-Amp, 45V, TO-247) rectifier
                      * 1x CapXon KM, 10V, 6800 uF, 12.5 x 35 mm (capacitors C34)
                      * 1x Teapo SC, 10V, 3300 uF, 12.5 x 25 mm (capacitors C35)
                      * 6 mm core, 5-turn, 14 AWG(?) PI coil between caps
                      * 1-KOhm, 1-Watt (?) minimum load resistor

                      -12V Rail:
                      * 2x UF303 (3-Amp, 300V) diodes feeding into 7912 regulator
                      * 1x OST RLG, 35V, 470 uF, 10 x 20 mm (C27, before 7912 regulator)
                      * 1x Teapo SEK, 16V, 100 uF (C28, after 7912 regulator)

                      -5V Rail:
                      * no rectifier – uses a 7905 regulator with the -12V rail being the input
                      * 1x Teapo SEK, 50V, 10 uF (C26, after 7905 regulator)

                      5VSB Rail:
                      * 1x MBR760 (7.5-Amp, 60V Schottky, TO-220AC) rectifier
                      * 2x Teapo SC, 10V, 1000 uF, 8 x 15 mm (capacitors C37 and C38)
                      * 3-4 mm core, 12-turn PI coil between caps
                      Attached Files

                      Comment


                        FSP Fortron/Source FSP400-60PFN REV:1 - recap info

                        And now for some recap info and pictures on the FSP400-60PFN PSU above…

                        Basically, I did a full recap, despite only the 5VSB caps being bad. To my surprise, those large 6800 uF CapXon KM units were all still in spec. But one or two of them had started to drift towards the high-side of their capacity spec (i.e. close to +20%). This means the capacitors are probably starting to become (electrically) leaky. So again, rather than taking a chance with them, I did a full recap. After all, I already needed to purchase 6800 uF caps of the same size for the Sparkle PSU. Adding a few more to the cart just made sense (actually, I got a fairly decent discount, as I ordered 10 of them.)

                        Anyways, here is a picture of the finished recap:
                        https://www.badcaps.net/forum/attach...1&d=1556412020
                        https://www.badcaps.net/forum/attach...1&d=1556412020
                        It's mostly a forest of Chemicons (KY and KYB series) in there, mixed with a few Nichicon caps. I also ended up recapping the primary-side daughter board and most of the secondary-side daughterboard (some caps just couldn't be removed without desoldering the entire daughterboard, so I skipped a few really small ones.)
                        You can also see in the second picture above that I soldered a short piece of 14 AWG copper wire to a Zener diode. This was actually the component behind the darkened PCB near transistor SQ3 on the primary-side daughterboard. Ugly looks aside, that should help keep it just a tiny bit cooler, I think.

                        After all of these procedures, of course, I tested the PSU on the same PC setup as the Sparkle PSU: my trusty ECS P4VXASD2+ motherboard that delivers power for its 2.66 GHz Pentium 4 CPU from the 5V rail (a good 90W load on the 5V rail alone.) The FSP400-60PFN PSU had no problems whatsoever powering that PC, an all voltage levels were in spec (note: I only checked these with a multimeter and didn't measure noise + ripple.) Unsurprisingly, the FSP400-60PFN also had no problems powering that PC. I forgot to write down the values of the voltages, but the 5V rail was pretty much dead-on 5V and didn't move much regardless of whether the CPU load was high or low.

                        As for some details about the recap… here is a full list of the electrolytic capacitors found in this PSU (minus the big input-side primaries):

                        * 3.3V rail:
                        ** [capacitor C34]: CapXon KM, 10V, 6800 uF, 12.5 x 35 mm
                        ** [capacitor C35]: Teapo SC, 10V, 3300 uF, 12.5 x 25 mm

                        * 5V rail [capacitors C29 & C30]: CapXon KM, 10V, 6800 uF, 12.5 x 35 mm

                        * 12V rail [capacitors C23 & C24]: 2x Teapo SD, 16V, 2200 uF, 10 x 20 mm

                        * 5VSB rail [capacitors C37 and C38]: 2x Teapo SC, 10V, 1000 uF, 8 x 15 mm

                        * -12V rail…
                        ** [capacitor C27]: OST RLG, 35V, 470 uF, 10 x 20 mm (before 7912 regulator)
                        ** [capacitor C28]: Teapo SEK, 16V, 100 uF, 5 x 11 mm (after 7912 regulator)

                        * -5V rail [capacitor C26]: Teapo SEK, 50V, 10 uF (after 7905 regulator)

                        * PWM daughterboard:
                        ** [capacitor SC18]: OST RLG, 50V, 100 uF, 8 x 13 mm
                        ** [capacitor SC11]: Teapo SEK, 35V, 33 uF, 5 x 11 mm

                        * Secondary-side daughterboard:
                        ** [capacitor C15]: Teapo SC, 50V, 1 uF, 5 x 11 mm
                        ** [capacitor C16]: Jamicon TK, 25V, 47 uF, 5 x 11 mm
                        ** 3x Teapo SS, 50V, 1 uF, 4.5 x 8 mm
                        ** 1x Teapo SC, 50V, 1 uF, 5 x 11 mm
                        ** 1x Teapo SS, 35V, 10 uF, 5 x 5 mm
                        ** 1x 50V, 4.7 uF, 5 x 11 mm
                        ** 1x 50V, 10 uF, 5 x 11 mm

                        And the capacitors I used were as follows:
                        3.3V rail: United Chemicon KYB, same voltage and capacitance for both caps (but ended up with a 10 mm diameter capacitor for the 3300 uF Teapo SC.)
                        5 rail: United Chemicon KYB, same voltage and capacitance for both caps.
                        12 rail: United Chemicon KYB, same voltage and capacitance for both caps.
                        5VSB rail: Nichicon HE, 6.3V, same capacitance for both caps.
                        -12V rail: Rubycon YXJ, 25V, 1000 uF for the OST RLG and Rubycon YXA for the Teapo SEK (same voltage and capacitance).
                        -5V rail: Rubycon YXA, 16V, 100 uF, 5 x 11 mm.
                        Capacitor SC18: Nichicon PW, same voltage and capacitance.
                        Capacitor SC11: Rubycon YXJ, same voltage and capacitance.
                        All the other small caps: a combination of Nichicon PW and Rubycon YXJ, same voltage and capacity for each.

                        I didn't make a picture of how the rails are connected this time. But hopefully the capacitor marks should give a clue to whoever might be trying to recap this PSU. If not, just ask. That's all for this PSU recap.
                        Attached Files

                        Comment


                          Re: Power supply build quality pictorial. part 2

                          I think I saw a late 2004 date code (043#), so it's nearly 15 years old. The 2A 5VSB is a little low for that era, perhaps. OTOH, the 12V current is pretty hefty. Love the heavy metal , and a fan that isn't WhoIsThat?! brand.
                          PeteS in CA

                          Power Supplies should be boring: No loud noises, no bright flashes, and no bad smells.
                          ****************************
                          To kill personal responsibility, initiative or success, punish it by taxing it. To encourage irresponsibility, improvidence, dependence and failure, reward it by subsidizing it.
                          ****************************

                          Comment


                            Re: Power supply build quality pictorial. part 2

                            Ablecom SP-450RP
                            NOTE: This has almost the same part number as a prior entry. This is a YM-1062PK 450W compared to a YM-1062A 420W. A couple of changes include reducing 5V output to 26A vice 40A and increasing 12V output to 20A from 15A. Combined +5V and +3.3V max power reduced to 160W from 220W. This one also has dual fans, led and reset button.

                            This came out of an OLD dual Xeon system that was in a SuperMicro tower case.

                            First off the specs
                            +3.3V 30A
                            +5V 26A
                            +12V 30A
                            -12V 1A
                            -5V 0.5A
                            +5Vsb 2A
                            +3.3V and +5V max combined 160W
                            +3.3V, +5V and +12V max combined 450W

                            First take a look at plugt and plugb, besides the common mode choke that is built in, there is another choke on the bottom along with a CLC PI filter.

                            Then a Hitachi HP3 270uF 450V along with active PFC provide by a Fairchild ML4800CP.
                            Then we have a full house for input filtering. MOV, Thermister, Xcap, Ycap, 2 big chokes with a HUGE blue cap in the middle. The bridge rectifier has a heat sink so I can't see the model. There are no less than 6 semiconductors attached to the input heat sink.
                            Next up we have xfm01 and xfm02 which are two of the transformers. There are also 3 optocouplers.

                            On the output side it looks like another 6+ semiconductors attached to heat sinks. Look at output 1 and 2 for coils and caps. Output 3 is showing a little bit of the side board. The one regulator I can see is a 7912 for the -12V rail.
                            As far as output caps go, I can see 2 4700uF 10V, 2 3300uF 16V, 2 470uF 16V, 1 330uF 16V and 2 other caps the same size as the 4700uF 10V.
                            There is also small board attached to back that controls the 2 fans via temperature, led, reset button and other stuff I can't figure out.

                            Workman ship is high grade overall.
                            Attached Files
                            Cthulhu for President!
                            Why vote for the lessor of two evils?

                            Comment


                              Re: Power supply build quality pictorial. part 2

                              You have more pics on the PSU, less detail pics, more overhead, where you see the whole beauty of that unit?

                              Comment


                                Re: Power supply build quality pictorial. part 2

                                Ok, I had to take them from my phone so they are not the worlds best. That computer had sat out in the shed(Where I keep random machines running) for about a year and the PS case was starting to rust(I guess that swamp cooler was just too much) so I'm going to re-case it, move the fan control board over, the fans everything else. I'm going to make that into a bench power supply using a plug in board.

                                Please note, I had bought that system from ASU salvage years ago and used it as a FreeBSD server until I moved that system over to a AMD quad about 3 months ago. It ran 24/7 with no problems. There is not a bulge in sight... I'm going to guess that the two fans helped just a little bit.
                                GB













                                Attached Files
                                Last edited by ae7oo; 05-04-2019, 03:38 PM.
                                Cthulhu for President!
                                Why vote for the lessor of two evils?

                                Comment


                                  Re: Power supply build quality pictorial. part 2

                                  Going by date codes on parts, that was built in July or August of 2003. All that heavy metal probably helped with its longevity.

                                  That Fairchild ML4800 was one of the very first PFC controller ICs, originally designed and manufactured by MicroLinear (probably before Unitrode's UC3852), starting IIRC in the late 1980s.
                                  PeteS in CA

                                  Power Supplies should be boring: No loud noises, no bright flashes, and no bad smells.
                                  ****************************
                                  To kill personal responsibility, initiative or success, punish it by taxing it. To encourage irresponsibility, improvidence, dependence and failure, reward it by subsidizing it.
                                  ****************************

                                  Comment


                                    Re: Power supply build quality pictorial. part 2

                                    Just an update. I've been using that supply as my master supply, this means that I'll take an out, lets say 5V and then use that as an input for when I'm doing linear reg work for audio or precision analog work. Or better yet, I used the 12V as an input to a buck converter(3V and 6 amps) that supplies current to a candle/cigarette(yes, I smoke, so??)/whatever lighter.
                                    I did ramp up voltage slowly(I'm got an old true sine inverter that my son had bought broken(he thought he a use for the big caps, but lucky me, he never did) and turned into a variable AC voltage, I power this with one of my ham power supplies(13.8V at up to 50amps). It lets me go from 15V at 60Hz to over 240V at 50Hz, with enough power that, at least for testing, I can power just about anything(Ok, I will admit that when I plugged a fridge in(for testing), every damn fan through out the system kick on as one and went to full speed, I pulled that plug real quick).

                                    I did replace about 6 caps on the output. I thought the the ESR was a little high(not that I blame them, I know for a fact it has over 6 years at 24/7/365 on it). That is over 50,000 hours, so I would say they have lived up to mfg specs...

                                    AE7OO, aka GB
                                    Cthulhu for President!
                                    Why vote for the lessor of two evils?

                                    Comment


                                      Re: Power supply build quality pictorial. part 2

                                      Here show you a PSU Dell alienware aurora r3, specifically a Newton Power/Delta electronics n875ef-00
                                      The source carries an ADDA fan.
                                      I am surprised by its design, it has hardly any transformer, does anyone know how this type of PSU works? It amazes me how it is built, I have never seen anything to it.
                                      There are LTEC capacitors, but they have 30,000h, they have not given problems, but surely change the capacitors for quality ones.
                                      Attached Files
                                      Last edited by kevin!; 07-12-2019, 05:55 PM.
                                      Gaming pc:
                                      nVidia RTX 3080 TI, Corsair RM750I.
                                      Workshop PC:
                                      Intel core i5 8400, Intel SSD 256GB, nvidia gt1030, asus b365-a.
                                      Server:

                                      Comment


                                        Re: Power supply build quality pictorial. part 2

                                        Originally posted by kevin! View Post
                                        Here show you a PSU Dell alienware aurora r3, specifically a Newton Power/Delta electronics n875ef-00
                                        The source carries an ADDA fan.
                                        I am surprised by its design, it has hardly any transformer, does anyone know how this type of PSU works? It amazes me how it is built, I have never seen anything to it.
                                        Very interesting PSU, indeed.
                                        Probably uses some kind of resonant LC topology, thus allowing it to have a very small transformer. Some PS3 PSUs were done that way, IIRC. Synchronous rectification and DC-DC converters too, if I had to guess on this one. Delta usually really knows what they are doing, so I won't question its label at all.

                                        Originally posted by kevin! View Post
                                        There are LTEC capacitors, but they have 30,000h, they have not given problems, but surely change the capacitors for quality ones.
                                        Definitely a good idea. Ltec are quite crap. Though Delta does know how to design PSUs to make them last in there for a decent amount of time. Still, sooner or later, they will go. So good idea to change them, as you mentioned.

                                        Comment


                                          Re: Power supply build quality pictorial. part 2

                                          Originally posted by momaka View Post
                                          Very interesting PSU, indeed.
                                          Probably uses some kind of resonant LC topology, thus allowing it to have a very small transformer. Some PS3 PSUs were done that way, IIRC. Synchronous rectification and DC-DC converters too, if I had to guess on this one. Delta usually really knows what they are doing, so I won't question its label at all.


                                          Definitely a good idea. Ltec are quite crap. Though Delta does know how to design PSUs to make them last in there for a decent amount of time. Still, sooner or later, they will go. So good idea to change them, as you mentioned.
                                          Thanks Momaka for the data, here I learn new things hehe.
                                          From this source I believe what the label says, it is feeding an i7 2600, an nvidia gtx 760 4GB, several fans, an ssd, and an HDD, and the 80 silver certificate, I also consider it legitimate, the pcb that it has for the PFC circuit is awesome.
                                          Well, I will do that, when I can buy new capacitors and replace them.
                                          Gaming pc:
                                          nVidia RTX 3080 TI, Corsair RM750I.
                                          Workshop PC:
                                          Intel core i5 8400, Intel SSD 256GB, nvidia gt1030, asus b365-a.
                                          Server:

                                          Comment

                                          Working...
                                          X