Announcement

Collapse
No announcement yet.

HiPro HP-P3017F3 rev:03 [Dell H057N] - capacitor repair

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

    HiPro HP-P3017F3 rev:03 [Dell H057N] - capacitor repair

    PSU Recapping time!

    This is the same HiPro model HP-P3017F3 PSU detailed here . Only created this thread to make the information easier to find… and to show the slightly different capacitors used in this second PSU.
    https://www.badcaps.net/forum/attach...1&d=1578889373
    https://www.badcaps.net/forum/attach...1&d=1578889373
    https://www.badcaps.net/forum/attach...1&d=1578889373
    https://www.badcaps.net/forum/attach...1&d=1578889373
    https://www.badcaps.net/forum/attach...1&d=1578889373
    https://www.badcaps.net/forum/attach...1&d=1578889373
    https://www.badcaps.net/forum/attach...1&d=1578889373

    And the daughterboard showing the Weltrend WT7525 controller IC;
    https://www.badcaps.net/forum/attach...1&d=1578889373
    https://www.badcaps.net/forum/attach...1&d=1578889373

    Cap diagram of what rail each cap is connected to on the output side.

    Everything should be self-explanatory on that picture. Only thing I need to note is the yellow arrow, which points to resistor R422. This resistor is installed as a minimum load on the 12V rail, but isn’t really needed if the PSU will be used on a 12-based PC. I usually remove it and replace it with a higher value one (anything ~250 Ohms or more is a good idea.) Reason why is because it is rated 75 Ohms and 3 Watts. Being connected on the 12V rail, this means it dissipates close to 2 Watts of heat – enough to darken the PCB over time and expedite the demise of the nearby output caps (especially the one right next to it for the 3.3V rail.) For the PSU above, I removed R422 and installed three (3) 1-KOhm 1/4W SMD resistors on the bottom of the PCB between the 12V rail and ground (the parallel equivalent of one 333-Ohm resistor, which @ 12V will dissipate less than ½ Watt of heat.) You can see them annotated here with the red arrows:


    And finally, here is how I recapped my unit:

    It’s not quite a full recap (due to always being short on caps in my inventory), but most of the important caps have been replaced.

    This is what the PSU has on the outputs, in terms of caps:

    * 5VSB
    *** 1x Ltec LZP, 10V, 2200 uF, 10x20 mm before PI coil
    *** 1x Teapo SC, 10V, 470 uF, 8x13 mm after PI coil (note: space for this cap has 5 mm lead spacing and can accommodate a 10 mm diameter cap, if needed)

    * 3.3V Rail
    *** 1x Ltec LZP, 10V, 2200 uF, 10x20 mm before linear regulator
    *** 1x Ltec LZP, 10V, 2200 uF, 10x20 mm after linear regulator
    *** 1x Teapo SC, 10V, 1000 uF, 10x15 mm after linear regulator

    * 5V Rail
    *** 2x Teapo SC, 10V, 4700 uF, 12.5x25 mm with PI coil in between

    * 12V Rail
    *** 2x Ltec LZP, 16V, 2200 uF, 10x30 mm

    * -12V Rail
    *** 1x Teapo SC, 35V, 470 uF, 10x20 mm before 7912 linear regulator
    *** 1x Ltec LZP, 16V/25V, 100 uF (?), 5x11 mm after 7912 linear
    .
    .
    And here is what I recapped on this power supply:

    * 5VSB
    *** 1x Ltec LZP, 10V, 2200 uF, 10x20 mm ---> Nichicon HN, 16V, 1500 uF, 10x20 mm

    * 3.3V Rail
    *** 2x Ltec LZP, 10V, 2200 uF, 10x20 mm ---> Rubycon ZLH, 6.3V, 2200 uF, 10x20 mm

    * 5V Rail
    *** no change (left the big 4700 uF Teapo SC caps alone, as they measured good.)

    * 12V Rail
    *** 1x Ltec LZP, 10V, 2200 uF, 10x20 mm ---> United Chemicon KYB, 16V, 2200 uF

    * -12V Rail
    *** 1x Ltec LZP, 16V or 25V, 100 uF (?), 5x11 ---> Rubycon YXJ, 25V, 220 uF, 6.3x12

    I also replaced many of the small caps related to the control circuits. Here is an almost complete list of these small caps:
    2x Teapo SEK, 25V, 100 uF, 6.3x11 mm
    3x Ltec TK, 50V, 2.2 uF, 5x11 mm
    1x Ltec TK, 16V, 47 uF, 5x11 mm
    There are three more small caps, but I could not identify/read their values.
    But really, replacing the output filter caps and the small caps listed above should pretty much make this PSU bullet-proof. I use PSUs like this all the time in my basic builds, so they always come in handy to have.
    Attached Files

    #2
    Re: HiPro HP-P3017F3 rev:03 [Dell H057N] - capacitor repair

    Nice heavy metal! 300W might be conservative, or it might have a really quiet fan.
    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


      #3
      Re: HiPro HP-P3017F3 rev:03 [Dell H057N] - capacitor repair

      Originally posted by PeteS in CA View Post
      Nice heavy metal! 300W might be conservative, or it might have a really quiet fan.
      Probably a little bit of both.
      I think the 350W version of this PSU is very similar, if not identical on the primary side. And I've never heard the fan ramp up on these PSUs, even when I run them in a hot 30C/86F room in the summer at 1/2 load.

      Speaking of the 350W version, I need to post that too.

      Comment


        #4
        Re: HiPro HP-P3017F3 rev:03 [Dell H057N] - capacitor repair

        The capacitor after the filtering coil on the +5V rail looks like a 3300uF 10V Teapo SC, not 4700uF 10V (looking closely at the fourth and fifth attachment), if the capacitance and voltage rating is printed after the polarity stripe. Unless they're used interchangeably in these PSUs? Shame on Hipro for putting such a hot load resistor right next to crap caps. This is what I expect to see in PSUs manufactured by Deer or FSP, not Hipro.

        Comment


          #5
          Re: HiPro HP-P3017F3 rev:03 [Dell H057N] - capacitor repair

          Originally posted by Wester547 View Post
          The capacitor after the filtering coil on the +5V rail looks like a 3300uF 10V Teapo SC, not 4700uF 10V (looking closely at the fourth and fifth attachment), if the capacitance and voltage rating is printed after the polarity stripe.
          You are 100% right!
          Can't believe I missed that. I think the fact that both caps have the same size is what threw me off. I saw 4700 uF on one of them and assumed both to have the same specs, being on the 5V rail.

          Either way, I guess it can't hurt to replace them with 2x 4700 uF caps, for whoever is recapping this PSU (I know I have two more awaiting this procedure, and they both actually have their 5V rail caps failed.)

          Originally posted by Wester547 View Post
          Shame on Hipro for putting such a hot load resistor right next to crap caps. This is what I expect to see in PSUs manufactured by Deer or FSP, not Hipro.
          Yeah, I'm not sure why they went with such a hot load resistor there, considering these PSUs were made after the AthlonXP/Penitum 3 era and are geared towards 12V-based PCs. So the minimum load on the 12V rail is not necessary, IMO. The PSU runs absolutely fine without it, even when not connected to a motherboard.

          As for FSP PSU - I wouldn't put them in the same class as Deer. Sure some of them do have minor goofy design flaws, but they are decent to good for the most part. Using a 400W unit in one of my PCs right now, and it's one of the more efficient PSUs I have in my fleet - above 80%, despite being a half-bridge topology.
          Last edited by momaka; 01-19-2020, 06:53 PM.

          Comment


            #6
            Re: HiPro HP-P3017F3 rev:03 [Dell H057N] - capacitor repair

            Originally posted by momaka View Post
            It’s not quite a full recap (due to always being short on caps in my inventory), but most of the important caps have been replaced.
            Argh, this came back to bite me in the end. I pulled this PSU from the shelf the other day for use in a PC. Took a peek inside through the vents, as usual, and one of the caps on the 5V rail looked a bit iffy in the way light reflected off its vent. Opened the PSU for a quick inspection and found the first cap in the 5V rail output filter (a 4700 uF Teapo SC) bulging.

            *Sigh*
            I need to order some 6.3/10V, 3300 uF and 4700 uF caps soon. Got a lot of HiPro PSUs now waiting for these.

            Also as a side note, this PSU and it's 350 Watt bigger brother do not seem to ramp up their fans too much, even when they start running hot. I noticed this when I checked the 350W version of the PSU that I have installed in my parents' media/TV streaming PC. Essentially the PSU vents on the back (after the PC running for several hours in a very warm room) felt quite hot. The fan didn't seem to be moving that much air, though - and yes, I checked that it's not becoming seized. Rather, the stock fan controller is designed to run the fan around 3.6V in idle, which is quite low.

            So looks like these PSUs should get a full recap for more reliable operation due to running pretty warm. And if you are the experimental type, maybe mod the fan controller too. I'll probably be doing that once I have it figured out how it works.

            Comment


              #7
              Re: HiPro HP-P3017F3 rev:03 [Dell H057N] - capacitor repair

              Originally posted by momaka View Post
              ] And if you are the experimental type, maybe mod the fan controller too. I'll probably be doing that once I have it figured out how it works.
              Is the fan controller a software-based controller within the PCB somewhere? It still blows my mind how complex these systems can appear, but really it's just cleverly manipulating the flow of electricity.

              Comment


                #8
                Re: HiPro HP-P3017F3 rev:03 [Dell H057N] - capacitor repair

                Originally posted by upgradestations View Post
                Is the fan controller a software-based controller within the PCB somewhere? It still blows my mind how complex these systems can appear, but really it's just cleverly manipulating the flow of electricity.
                In server-grade PSUs fan speed is often controlled by a microcontroller that also controls the output regulation and supervisory functions such as over-current and UVP/OVP protection.

                In the case of desktop PSUs such as the one discussed in this thread, fan speed is usually adjusted using a thermistor and opamp to control the DC voltage to the fan.
                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


                  #9
                  Re: HiPro HP-P3017F3 rev:03 [Dell H057N] - capacitor repair

                  Originally posted by PeteS in CA View Post
                  In the case of desktop PSUs such as the one discussed in this thread, fan speed is usually adjusted using a thermistor and opamp to control the DC voltage to the fan.
                  Correct.
                  This HiPro uses a "simple" thermistor + transistor circuit as a fan controller. I haven't began to trace it yet, but I did find the fan "driver" transistor... and it's (sigh) the usual TO-92 package. Probably going to run very hot when the fan is set to run between 5-7V. At least the fan is only rated for ~160 mA @ 12V. The more powerful fans tend to cook TO-92 transistors when they are used as a series linear regulator.

                  Comment


                    #10
                    HiPro HP-P3017F3 rev:03 [Dell H057N] - recap v2 & 3LF models

                    Originally posted by momaka View Post
                    Argh, this came back to bite me in the end. I pulled this PSU from the shelf the other day for use in a PC. Took a peek inside through the vents, as usual, and one of the caps on the 5V rail looked a bit iffy in the way light reflected off its vent. Opened the PSU for a quick inspection and found the first cap in the 5V rail output filter (a 4700 uF Teapo SC) bulging.
                    And here's that:

                    Right... so the recapped PSU needed a recap.
                    The 4700 uF cap on the 5V rail was bulging and reading low on capacitance and high on ESR.
                    The 3300 uF cap on the 5V rail was not bulging yet, though it's capacitance read around 3800 uF... which is still barely within +20% spec (but IMO suspiciously high and will likely go bad.) ESR OK, otherwise. While at it, I also pulled the 470 uF cap on the -12V rail, and the 1000 uF cap on the 5VSB (originally, I wrote above that it's a 10V, 470 uF cap, but it looks like HiPro uses either 470 or 1000 uF in these units - I guess whatever is available.) Also pulled the output cap on the -12V rail after the regulator (16V, 100 uF) - it read normal. Same with a few more small caps on the daughterboard on the secondary side.

                    Well, let's not goof around anymore this time. Here is the recap, V2:

                    Used a Chemicon KY 10V, 4700 uF cap for the original 4700 uF on the 5V rail.
                    Used a Rubycon ZLQ 6.3V, 3900 uF cap for the original 3300 uF on the 5V rail.
                    Used a Chemicon KZE 25V, 470 uF cap for the original 35V 470 uF on -12V rail.
                    Used a Nichicon PM 16V, 100 uF cap for the original 16V 100 uF on -12V rail.
                    Used a Nichicon PL 10V, 1000 uF cap for the original 1000 uF cap on 5VSB spot #2.
                    Changed 3 more 50V, 22 uF caps on the daughterboard with Rubycon YXJ (IIRC)
                    Replaced a 50V, 1 uF cap near the secondary heatsink and 5VSB #1 cap with a Panasonic FC 50V, 1 uF. The original Ltec was starting to read slightly high capacitance.
                    Replaced a 50V, 10 uF Teapo SEK cap under the secondary heatsink with a Nichicon PS (or PW?) 50V, 10 uF.
                    So that pretty much covers everything.

                    I guess the only possible "bad caps" left now is the single 10V, 1000 uF Teapo SC on the 3.3V rail that is in parallel with the 2200 uF cap before it. I didn't change it because I didn't have any 1000 uF caps in 10 mm dia. in stock. Could have fit a 8 mm 820 uF ZLH but... whatever, the 2200 ZLH in parallel before it should be able to handle the entire load even if that Teapo cap failed.

                    Also, I had two more HiPro PSUs that needed a recap like this. These have model number HP-P3017F3 3LF. They are literally the same exact PSU inside. However, the label is slightly different: it downgrades the 3.3V rail from 17 Amps to 13 Amps (and consequently the combined 3.3V+5V rating to 140 Watts) and the -12V rail from 0.8 Amps to 0.3 Amps. I don't see any changes to the rectifier configuration, though. So here is the label on those 3LF versions:
                    https://www.badcaps.net/forum/attach...1&d=1606803639

                    And here is how they looked inside, in terms of bad caps:
                    https://www.badcaps.net/forum/attach...1&d=1606803639
                    https://www.badcaps.net/forum/attach...1&d=1606803639
                    https://www.badcaps.net/forum/attach...1&d=1606803639

                    Both of them got recapped pretty much the same way as the other HiPro above. On one of them though, I found a few Nichicon PJ caps for the ones on the -12V rail, so that's what I used there. And on the other, that 10V, 1000 uF Teapo SC on the 3.3V rail that I left in the first unit, was bulged in this one. As mentioned, I didn't have any 10V, 1000 uF caps in stock in 10 mm dia., so I used an old-as-dirt Chemicon KME series cap (also 10V, 1000 uF). For those who aren't aware, the KME series was replaced by the KMG series... many many years ago. That KME cap is at least 20 years old. But it read within spec, so whatever - better than a bad Teapo. Being that it will filter the 3.3V rail (which is linearly regulated and has very low ripple output), I don't see it getting stressed much. And it's also for an experiment to see how much longer it will last.
                    https://www.badcaps.net/forum/attach...1&d=1606803639

                    I will also post an updated cap list shortly, as the one mentioned in post #1 is not entirely accurate. Stay tuned...
                    Attached Files
                    Last edited by momaka; 12-01-2020, 12:25 AM.

                    Comment


                      #11
                      HiPro HP-P3017F3 - cap list v2

                      Below I'm re-posting an updated list of the caps found in this power supply.
                      This time, I will omit the brand and series in most cases, since these HiPro PSUs seem to use a mix of Teapo SC, Ltec LZP, Teapo SY, and United Chemicon KZE - all interchangeably.

                      Updated cap list v2:

                      * 5VSB
                      *** 1x 10V, 2200 uF, 10x20 mm before PI coil
                      *** 1x 10V, 470 uF, 8x13 mm *OR* 10V, 1000 uF, 8x15 mm after PI coil (note: space for this cap has 5 mm lead spacing and can accommodate a 10 mm diameter cap, if needed)

                      * 3.3V Rail
                      *** 1x 10V, 2200 uF, 10x20 mm *OR 16V, 2200 uF, 10x25 mm before linear regulator
                      *** 1x 10V, 2200 uF, 10x20 mm after linear regulator
                      *** 1x 10V, 1000 uF, 10x15 mm after linear regulator

                      * 5V Rail
                      *** 1x 10V, 4700 uF, 12.5x25 mm before PI coil
                      *** 1x 10V, 3300 uF, 12.5x25 mm after PI coil

                      * 12V Rail
                      *** 2x 16V, 2200 uF, 10x30 mm

                      * -12V Rail
                      *** 1x 35V, 470 uF, 10x20 mm before 7912 linear regulator
                      *** 1x 16V, 100 uF, 5x11 mm after 7912 linear regulator

                      * "Startup cap" C202 (located on the primary side by the bridge rect.)
                      *** 1x 25V, 100 uF, 6.3x11 mm

                      * daughter board (on the secondary side)
                      *** 3x 50V, 22 uF, 5x11 mm
                      *** 1x 25V, 100V, 6.3x11 mm (filter for fan)
                      *** 1x 50V, 2.2 uF, 5x11 mm

                      * around daughter board (on the secondary side)
                      *** 1x 50V, 22 uF, 5x11 mm
                      *** 1x 16V, 47 uF, 5x11 mm

                      * next to / under secondary side heatsink (near 5VSB output capacitor #1)
                      *** 1x 50V, 1 uF, 5x11 mm
                      *** 1x 50V, 10 uF, 5x11 mm


                      Also, here is an observation I've made about Teapo SC and Ltec LZP capacitors:
                      In their early stage of failure, they start to exhibit high capacitance, probably due to electrolyte breaking down the insulation layer / thinning the foil (which also tends to result in the cap being able to withstand lower voltage than before.) At that point, they may or may not have bulged yet. ESR-wise, they usually still show "in spec". But as time progresses, the electrolyte will break down more and more, eventually causing pressure to build up inside the cap until they bulge and vent/leak. After this, the capacitance will start to go down (typically to a 100th to a 10th of the rated capacitance) and the ESR go up to several Ohms or more. And again, it seems that only Teapo SC and Ltec LZP seem to suffer this fate (so far) in these HiPro PSUs. The Teapo SY 16V, 2200 uF caps that are used on the 3.3V rail first cap appear to be OK (but I guess time will tell.) And as mentioned above, some of these PSUs come stock with a few United Chemicon KZE here and there (I have not seen a HiPro full of KZEs from the factory yet.)

                      Anyways, I think that is all I have for now. I might even post the 350 Watt version of this PSU here. It's almost the same thing inside as well, just with bigger rectifiers and heatsinks. Capacitor-wise, it's the same deal, except for a few of the very small 5x11 caps.

                      Comment


                        #12
                        Re: HiPro HP-P3017F3 rev:03 [Dell H057N] - capacitor repair

                        The old HiPro 250W PPFC units for Dell were rock solid, once their Asiacon and similar caps were recapped. I have a few running fine after ~15-17 years. One key cap is a primary side 47uF or 100uF cap in the +5Vsb flyback converter - I used NCC LXA in some recaps there.

                        I also replaced the OEM ADDA fans with Minebea/NMB ball-bearing fans in a few - not really needed, just additional insurance.

                        Comment


                          #13
                          Re: HiPro HP-P3017F3 rev:03 [Dell H057N] - capacitor repair

                          Originally posted by linuxguru View Post
                          The old HiPro 250W PPFC units for Dell were rock solid, once their Asiacon and similar caps were recapped. I have a few running fine after ~15-17 years.
                          Yeah, those were pretty awesome units too. I own a few myself. Rock-solid performance indeed. I actually haven't recapped mine yet, and somehow the original Asiacon caps are still in spec... or were when last checked a few years ago. Will probably recap them eventually.

                          I think the primary "startup" cap in those was a 47 uF, 50V. Fixed one a long time ago with a shorted diode fore the primary auxiliary rail and replaced that cap while at it too... actually, here is the post with that PSU, posted 10 years ago!
                          https://www.badcaps.net/forum/showth...656#post128656
                          And yes, it's still working.
                          Last edited by momaka; 12-10-2020, 08:46 PM.

                          Comment


                            #14
                            Re: HiPro HP-P3017F3 rev:03 [Dell H057N] - capacitor repair

                            Originally posted by momaka View Post
                            I think the primary "startup" cap in those was a 47 uF, 50V. Fixed one a long time ago with a shorted diode fore the primary auxiliary rail and replaced that cap while at it too... actually, here is the post with that PSU, posted 10 years ago!
                            https://www.badcaps.net/forum/showth...656#post128656
                            And yes, it's still working.
                            Yup, same PPFC series - mine were probably an intermediate series that used the sawblade-type heatsinks from the 300W units, but are otherwise very similar. The only downside today is that there's no 6-pin PCIe graphics connector, but this was from the AGP era, so I can't really complain. I could try to solder one in, but it's only 250W anyway, so no real point in running a big GPU with it - a 4-pin molex to 6-pin PCIe adapter cable should be fine if needed for some PCIe cards with modest power requirements.

                            Comment


                              #15
                              Re: HiPro HP-P3017F3 rev:03 [Dell H057N] - capacitor repair

                              Originally posted by linuxguru View Post
                              Yup, same PPFC series - mine were probably an intermediate series that used the sawblade-type heatsinks from the 300W units, but are otherwise very similar.
                              Ah, those.
                              Aside from the slightly different heatsinks on the secondary, I think it's pretty much the same PSU. I have a few of them as well. Actually, one is still in service as-is right now, with the original Teapo caps in my Dell OptiPlex 170L PC that I'm posting from right now.

                              The other two were just the PSU's PCB with PPFC coil that someone gave me for parts (not sure why they pulled them out of their shells, but both PSUs were working.)
                              One is currently homed in the shell of what used to be a very crappy JNC PSU. I gutted the JNC PSU for parts many moons ago. The case was actually fairly solid, though. So this is the handywork of that:
                              https://www.badcaps.net/forum/showpo...0&postcount=80

                              The other P2507FWP PCB I put into the shell of a crappy Raidmax RX-380K PSU, as that's all I had at the time. Don't think I have uploaded pictures of it, though. Still, it's a similar ordeal as the PSU above, but just not so horribly botched/put together. This one is also still in service for the GP PC I put together for my parents (mostly used for printing/scanning documents, paying bills, and occasional Word/Excel work, as well as "Guest" use PC.)

                              Originally posted by linuxguru View Post
                              The only downside today is that there's no 6-pin PCIe graphics connector, but this was from the AGP era, so I can't really complain. I could try to solder one in, but it's only 250W anyway, so no real point in running a big GPU with it - a 4-pin molex to 6-pin PCIe adapter cable should be fine if needed for some PCIe cards with modest power requirements.
                              Yup, I did the same with a 4-pin Molex to a 6-pin PCI-E power adapter cable on one of these when I needed a PSU for a test system in a pinch. It managed to power a system with a 67-Watt TDP CPU and Radeon HD3870 (105 Watts TDP) GPU no problems... though the 12V rail did dip down to 11.80V under full load. All things considered, though, that's not too bad at all. 5V was still rock-solid and in-spec.

                              Really, I only wish these PSUs came with at least 1 or 2 SATA power cables so I don't have to constantly try to find adapters for them when I need to put one of these PSUs in a newer PC with SATA drivers. Funny enough, the HP-P3017F3 mentioned in this thread come with all SATA and not a single Molex, so sometimes I also wish the reverse was true for those. In fact, I've been considering doing a cable connector "swap" between these older and newer HiPro PSUs, where I might move 1 SATA connector from the newer HiPro's to the older ones, and take 1 Molex from the older ones and move to the newer HiPros. Will see, though.

                              Comment

                              Working...
                              X