Announcement

Collapse
No announcement yet.

EZ-Cool ATX-650JSP ATX PSU recap.

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

    EZ-Cool ATX-650JSP ATX PSU recap.

    I have been given this EZ-Cool ATX-650 JSP. I don't believe that it is actually capable of 650W but it seems reasonably well built with a decently sized transformer and heatsinks.
    Please see photos before reading further.
    I have removed all the output caps and labeled where they go and some other things in the photos.

    The 5V and 3.3V output caps were all 2200uF 10V fukyuu's and were visibly bulging. I intend to replace these with 10mm 2200uF 10V panasonic FRs.

    It has 2 12V rails each with its own pi coil and output capacitor, which are 10mm 2200uF 16V Gi LOWRSR 105C 07B. They are not bulging and test OK on my ESR meter with 0.04ohms ESR. I have never heard of Gi capacitors before. Are they any good? Should I replace them anyway. The only 2200uF 16V 10mm caps that i have found are Panasonic FKs but these are over £1 each. I think I could manage to fit the better 12.5mm 2200uF FRs or FMs here if I raised them up above the pi coils with long legs but would that be ok?
    The 12V output is rectified by two MOSPEC 16A ultrafast To-220 rectifiers in parallel. Should I replace these with Schottky's or will the existing ones be good enough.

    The 5VSB circuit uses the 2 transistor design but doesn't have the critical electrolytic cap that causes the output voltage to go high. There has not been any excessive overheating.
    The output caps were 2 1000uF 10V fukyoos, which were not bulging but I'm going to replace them anyway with 1000uF Panasonic FRs.

    There is an unpopulated area on the PCB for an active PFC circuit. I would like to add this back if possible.
    I would need the values for the resistors, caps, zeners and the part number for U1. In this place is a passive PFC choke that is screwed to the case. Does anyone have this PSU with the active PFC circuitry populated and could tell me the component values?

    Should I also replace the small 10uF and 2.2uF caps?

    The primary switchers are 2 D209Ls in half bridge configuration. I think equivalents would be MJE13009's?
    Attached Files
    Last edited by technicalspecialist; 07-11-2018, 11:07 PM.
    If it's broken, fix it. If it's not broken, improve it.

    #2
    Re: EZ-Cool ATX-650JSP ATX PSU recap.

    12v sense is optional in ATX spec, so it's ok to leave it out.
    Measure the voltage drop to get an idea if it would help to add it, might be more parts missing.

    On 12.5mm caps on extended legs: don't do it: the ESR will be affected.
    As you have already measured the current caps to be ok just leave them.

    As for the small caps always replace them in a power supply, otherwise it's better to do nothing with it.
    In most cases they are either critical for the 5VSB section and/or critical for the driver chip.

    On the APFC why would you add it?
    The only sensible case for an end customer is if you use a UPS, then it will present a lower VA load.
    Otherwise it will just be a downgrade because private citizens don't pay for apparent power, only real power.
    "The one who says it cannot be done should never interrupt the one who is doing it."

    Comment


      #3
      Re: EZ-Cool ATX-650JSP ATX PSU recap.

      Mr. Hanson covered pretty much all of the questions above, so I only have a few small additions.

      Originally posted by technicalspecialist View Post
      It has 2 12V rails each with its own pi coil and output capacitor, which are 10mm 2200uF 16V Gi LOWRSR 105C 07B. They are not bulging and test OK on my ESR meter with 0.04ohms ESR. I have never heard of Gi capacitors before. Are they any good? Should I replace them anyway. The only 2200uF 16V 10mm caps that i have found are Panasonic FKs but these are over £1 each.
      Just like the Fuhjyyu caps, they are certainly not going to be high quality reliable parts. As to whether to replace them or not... that depends on what you intend to use the power supply for and how much you value it overall. If the PSU will be used just for the occasional motherboard test, you probably can leave those Gi caps until they go bad (they will eventually, but it might be a while before they do). However, if you intend to give this to a customer or someone who needs their PC working reliably, then it's better to replace them.

      Other options for 16V, 2200 uF caps in 10 mm diameter are United Chemicon KYB, IIRC.

      Otherwise I agree with what Per posted: don't install caps with long leads, because that will affect the circuit ESR. That said, there is a trick I sometimes use when installing 12.5 mm caps in spots for 10 mm ones: if the space is tight but not too tight, you can install the 12.5 mm caps with slightly longer leads so that they are just 2-3 mm above the PCB. Then, bend the leads in such a way that allows the 12.5 mm cap to fit in. Again, obviously this won't work if the space is too tight. But if you have very close components on one side of the cap and a little bit more space on the other side of the cap, you can use the above trick to fit 12.5 mm caps.

      Originally posted by technicalspecialist View Post
      The 12V output is rectified by two MOSPEC 16A ultrafast To-220 rectifiers in parallel. Should I replace these with Schottky's or will the existing ones be good enough.
      In general, Schottky rectifiers will give you better efficiency. They will also raise the 12V rail's voltage a bit, which may be helpful when the PC is used in a system that draws power primarily from the 12V rail (i.e. pretty much all computers since 2004-2005).

      That said, because this is a half-bridge PSU... you may also encounter problems with the Schottkies on the 12V rail (generally due to slower recovery time.) Also, if you happened to use this PSU on an old PC that draws power for the CPU from 5V (like a socket 462 motherboard or some socket 478 boards), the 12V rail with the Schottky diodes could go too high (out of spec).

      So all in all, using Schottki diodes on the 12V rail on a half-bridge PSU like this is a bit experimental. Nevertheless if you do, make sure your Schottkies are rated for *at least* 100V reverse voltage. Also, if you go with Schottkies... there's no point in using two smaller ones in parallel. Just use a bigger single one. Something rated along the lines of 30 Amp, 100V should be plenty.

      Originally posted by technicalspecialist View Post
      The 5VSB circuit uses the 2 transistor design but doesn't have the critical electrolytic cap that causes the output voltage to go high. There has not been any excessive overheating.
      The output caps were 2 1000uF 10V fukyoos, which were not bulging but I'm going to replace them anyway with 1000uF Panasonic FRs.
      Always a good idea.

      Originally posted by technicalspecialist View Post
      There is an unpopulated area on the PCB for an active PFC circuit. I would like to add this back if possible.
      Don't bother.

      Your power supply will be much more reliable with the passive PFC coil you have there right now. It also doesn't decrease the efficiency as much as Active PFC does. (Yes, APFC actually decreases efficiency on the customer side).

      The only benefit true benefit from APFC is universal input voltage (typically 90-250 V AC) - this can be helpful if you live in a country with 220/230/240 V mains that has the tendency to go low from time to time.

      The other not-so-important benefits (to the customer) of APFC is low line harmonics, which only matters if you live in a place where there are only a few central step-down transformers in the city/town/village, and most of the power to the houses is directly carried on 220/230/240V lines. APFC also makes the PF close to unity (1), which can matter if you run 100's of these PSUs in one room.

      But like Per said, for a customer's PSU, you don't need APFC. And I can tell you for a fact that PSUs with no PFC or only PPFC are much more reliable in the long run over their APFC counterparts.

      Originally posted by technicalspecialist View Post
      Should I also replace the small 10uF and 2.2uF caps?
      Yes.
      They are also usually pretty cheap, so no reason not to.

      Originally posted by technicalspecialist View Post
      The primary switchers are 2 D209Ls in half bridge configuration. I think equivalents would be MJE13009's?
      More or less.

      On that note...
      Originally posted by technicalspecialist View Post
      I have been given this EZ-Cool ATX-650 JSP. I don't believe that it is actually capable of 650W but it seems reasonably well built with a decently sized transformer and heatsinks.
      Yeah, this is definitely NOT a 650W power supply.
      Probably 400 Watts continuous at best, and maybe 450 W peak. I suggest keeping the load under 350 Watts to be safe. Reason being is the label looks pretty sketchy. First giveaway is the lack of "combined" power draw from all of the rails. Then there's the agency approvals: none of them look legit. Given the trace spacing on the solder-side, I doubt this PSU will pass UL and CE safety standards, let alone S/N/D/FI, which are even more strict.
      Also, apparently your -5V rail is capable of 18 Amps! SWEET!

      But in all seriousness, the biggest show-pooper is probably that 18 Amp rating on the 12V rail, which means you'll have only about 216 Watts of power for a modern PC. So if you have a dedicated GPU in your system, check its TDP before using this PSU. For safety margin, the combined TDP of the GPU and CPU should not exceed 200 Watts. And that's only if the PC has 1-2 HDDs. If you have more HDDs, further drop the combined power for the CPU and GPU by approximately 10W per HDD.
      Last edited by momaka; 07-12-2018, 05:23 PM.

      Comment


        #4
        Re: EZ-Cool ATX-650JSP ATX PSU recap.

        that's one hell of a -5v supply!!
        nice stealth pfc circuit too.

        Comment


          #5
          Re: EZ-Cool ATX-650JSP ATX PSU recap.

          Yeah, I really laughed a lot as well when I saw that 18A -5V rating. Its ridiculous! its obviously a typo.
          No motherboard in history ever drew 18A at -5V. Modern ones don't use -5 at all.
          What did old boards even actually use -5V for? Did some ISA cards need -5V?
          -12V can easily be regulated to -5V with a LM7905 anyway.

          This PSU was taken out of a customers computer and I have already replaced it with a better quality delta PSU, which they have paid me for.

          I will replace the GI caps. I may be able to fit 12.5mm FR caps in without extending the leads too much if I adjust the pi coils and other caps a bit .
          If not I'll have to use Panasonic FKs as Farnell don't stock chemicon KYBs.
          I don't trust chemicon caps anyway as I've replaced bulging ones before on motherboards.
          Never seen a bad Panasonic.

          I have a whole drawer of TO-247 Schottky's taken out of PSUs that were used for the 5V output.
          I'll check the data sheets to make sure they are rated for at least 100V (why?) reverse voltage and 30A .

          I will replace the small caps too. It already has a PPFC choke (not pictured) so I won't bother with APFC.
          As you said - its not necessary and i don't know the component values anyway.

          I'll get the replacement caps ordered today and post photos of them soldered in next week.
          If it's broken, fix it. If it's not broken, improve it.

          Comment


            #6
            Re: EZ-Cool ATX-650JSP ATX PSU recap.

            -5v could be used along with +5v for the audio section and maybe the serial ports.
            -12v works just as well though.

            Comment


              #7
              Re: EZ-Cool ATX-650JSP ATX PSU recap.

              Originally posted by momaka View Post
              Reason being is the label looks pretty sketchy. First giveaway is the lack of "combined" power draw from all of the rails. Then there's the agency approvals: none of them look legit. Given the trace spacing on the solder-side, I doubt this PSU will pass UL and CE safety standards, let alone S/N/D/FI, which are even more strict.
              I can vouch for that, the Scandinavian safety agencies are very strict.
              Unfortunately they have gone to pasture: almost nobody uses them anymore.
              It still amuses me when I see devices sold in for example Australia with this text:
              "Apparaten måste anslutas till ett jordad vägguttag"
              It's Swedish for that the device must be connected to a grounded outlet, I'm sure most Australians can catch that, or any other country such a PSU is sold in

              Originally posted by momaka View Post
              Also, apparently your -5V rail is capable of 18 Amps! SWEET!
              Geez, I did not look at the label, that is absolutely hilarious!
              "The one who says it cannot be done should never interrupt the one who is doing it."

              Comment


                #8
                Re: EZ-Cool ATX-650JSP ATX PSU recap.

                Originally posted by technicalspecialist View Post
                What did old boards even actually use -5V for? Did some ISA cards need -5V?
                Yup, certain ISA cards from what I've read before.

                Originally posted by technicalspecialist View Post
                I will replace the GI caps. I may be able to fit 12.5mm FR caps in without extending the leads too much if I adjust the pi coils and other caps a bit .
                Sounds like a plan. I too have "adjusted" components around in the past to get bigger caps to fit. Just when you're done installing the bigger caps, make sure there is no excessive pressure on the leads, as that can cause failure over time. For example, if a cap is really tightly soldered down to the PCB and there are wires near the cap that tug on it, they may eventually snap the lead connections inside the cap. Likewise, if you have to push your cap(s) hard to get them to stay where you want them, that too can make them fail.
                So make the space for the bigger cap(s) and make sure it fits there easily before soldering it down.

                Originally posted by technicalspecialist View Post
                I don't trust chemicon caps anyway as I've replaced bulging ones before on motherboards.
                Those would be their KZG series. They had a ton of problems with those up to 2008 or so, along with a few other ultra-low ESR series that were again only seen on motherboards (like KZJ).

                However, their KY, KYB, and KZE series are absolutely fine. Many PSU manufacturers use them and they simply don't fail unless heavily abused past their specs. The LXZ, LXV, and LXY series are another choice. These are not as low ESR as teh KY, KYB, and KZE, but they've been out for a long time (since the early-mid 90's) and proven themselves over the years. So I wouldn't discredit the whole brand just because of a few bad motherboard series.

                Though you're observation about Panasonic is correct - they indeed appear to have the best track record when it comes to reliability. All the other Japanese brands have had at least one or two series with some sort of problems (with Nichicon, it was HM, HN, and HZ from 2001-2005, and some of their old PSU-grade series such as PR often leaked from the bottom; Rubycon MCZ series are sensitive to heat; and Sanyo WF and to some extend WG, appear to have similar problems as Chemicon KZG).

                Originally posted by technicalspecialist View Post
                I have a whole drawer of TO-247 Schottky's taken out of PSUs that were used for the 5V output.
                I'll check the data sheets to make sure they are rated for at least 100V (why?) reverse voltage and 30A .
                If they were used on the 5V rail, they are very likely only rated for either 40 or 45V.

                The reason for the 100V rating is because half-bridge, push-pull, and full-bridge PSU topologies (i.e. ones where you have a center-tapped transformer output) will usually have 2x higher voltage on the rectifiers than what single and double-forward topologies do. So as one side of the rectifier sees a positive voltage square wave (and rectifies it), the other will be seeing the same square wave, but with negative voltage. And because the Cathode-side of the rectifier is already connected to the 12V rail (i.e. 12V DC above ground), the difference between that and the negative square wave will be 12V + the absolute voltage value of the negative square wave. You might think that's only 12V + 12V = 24V, but it's not. You also have to factor in that PSUs function on the principle of PWM, so the +/-square waves will not have a peak at 12V. It's going to be more like 15-30V, depending on the PSU design. If it's closer to the latter, each diode in the rectifier could be seeing up to 12 + 30 = 42V reverse voltage. And if you ever looked at a square wave on an oscilloscope, they always have little "ears" near the edges. These "ears" can present voltage transients of another 2-5V. Add everything up, and you'll be looking at over 45V reverse voltage. Hence why a 45V part CANNOT be used.

                The next higher voltage for rectifiers is typically 60V, and some half-bridge PSUs actually do use a rectifier with that voltage rating for their 12V rail(s). Bug again, depending on design, some PSUs will not be happy with a 60V-rated Schottky either. This has to do with the reverse-recovery time. Ultra-fast (i.e. not Schottky) rectifiers can recover from the reverse voltage very quickly before they conduct again and vice-versa. Schottkies are slower, and when you use them near their maximum rated voltage, that can make then perform even worse.

                So for this reason, if you see a half-bridge (or other center-tapped topology) PSU with rectifiers on the 12V rail rated for 100V or more, then your replacements should also be rated for 100V or more.

                Anyways, sorry for the wall of text. Hope you do find it useful, though (or at least somewhat interesting to read). There are a few of us "PSU freaks" here on the forum and sometimes we get carried away with our posts.

                Originally posted by stj View Post
                -5v could be used along with +5v for the audio section and maybe the serial ports.
                -12v works just as well though.
                On that note, I really wish newer motherboards made use of the -12V and/or -5V rails for audio use, at least for the audio amp outputs. That way, they can get rid of those pesky small coupling caps and get much flatter frequency response in the low-range, even with lower impedance loads (such as headphones directly connected to the motherboard). Instead, we get crummy 100 uF small caps with typically a series 10-20 Ohm resistors to limit the output of the amp IC so that it doesn't toast itself. But that creates all kinds of distortions in the low-end with a speaker/headphone load directly connected to the mobo. So if you're using headphones (or at least decent ones), you should definitely get a headphone amp of some sort.

                Originally posted by Per Hansson View Post
                I can vouch for that, the Scandinavian safety agencies are very strict.
                Unfortunately they have gone to pasture: almost nobody uses them anymore.
                Yeah, I think here in the US we have something similar going on with UL too. I hear they are just too expensive, so small-time manufacturers don't even bother.
                Last edited by momaka; 07-13-2018, 09:51 AM.

                Comment


                  #9
                  Re: EZ-Cool ATX-650JSP ATX PSU recap.

                  I want to reiterate what momaka said about Chemi-con. So long as you stay away from KZG and KZJ made between 2001-2007 and TMV and TMJ from that time period, and so long as you don’t reuse certain quaternary salt series such as LXF/old SXE/etc, you shouldn’t have a problem with them. It’s extremely rare that I ever find a bad or out-of-spec KZE in any equipment or on any motherboard, even when they run very hot. And LXZ, I’ve never seen bad anywhere. Period. Never seen a bad Nichicon PW either.

                  Originally posted by momaka View Post
                  and Sanyo WF and to some extend WG, appear to have similar problems as Chemicon KZG).
                  I don’t know about WG, I think WG is closer to WX in quality than KZG/KZJ/bad datecode HM/HN/HZ/WF, because unlike those series, they both appear to be very inconsistent in quality and don’t seem to fail altogether. Some 16-17+ year old batches will test on the nose for ESR/capacitance/leakage current without the slightest hint of bloating or leaking (this is simply not going to be the case with those other series, at least with 6.3V rated Nichicon HM of that age and 6.3V KZG of that age whether they appear fine or not)... and yet at other times, sometimes but not often, they won’t last 11-12 years in storage. So some batches are definitely fine, others less so. I don’t think there’s a “good” batch of the lower voltage KZG and HM of old, OTOH.

                  Panasonic, BTW, don’t have an impeccable track record. Besides having some quaternary salt series of their own (HFQ, HFZ, etc), I’ve noticed that FJ/FJS/FL made after 2006-2007 have a tendency to be heat-sensitive like MCZ/MFZ (I can’t blame the caps for failing because they are located next to hot MOSFETs and hot, “enclosed” inductors, but I know the older FL/FJ would not blow up as easily in those positions). Also, another Japanese brand called TK (Toshin Kogyo) had major problems with their ATWY and ATWB series on motherboards as well - these were notorious for going bad within 2-3 years of light use in many well cooled and well ventilated systems.

                  Replacing Fuhjyyu preemptively is smart. They don’t have to bulge or leak to be bad, as they can slowly vent, dry out, and leak by way of rubber seal diffusion and go out-of-spec anyway (it’s definitely happened before). For that matter, I would replace all the small caps as they will dry out the quickest, to reiterate the advice stated before me. And as also stated before me, ultra fast recovery rectifiers, due to having higher voltage drop characteristics, are less efficient than schottky rectifiers and will dissipate quite a bit more heat, which will in turn cause the main switchers to have to work harder. Depending on what the output rectifiers are for the +5V and +3.3V rail, and between that, the 35 size transformer, and the half-bridge topology/D209L bipolar transistors (depending on how well the bipolar transistors are driven, as their switching times are rather inferior to that of a MOSFET), this PSU has likely been either overrated 200-250 watts or so or has been “subtly” rated for its peak wattage rather than continuous.

                  I also noticed this unit “seems” too have two 1000uF 200V primary capacitors? It would be interesting to pull them and see if they actually measure 1000uF or closer to 680uF-820uF, as there was a malpractice many Chinese brands partook in for years (and possibly still are?), where they would resleeve high voltage capacitors with labels that would purport a much higher capacitance than the actual value. Definitely stick with higher reverse voltage parts for the output. It’s one drawback of half-bridge PSUs, but what IS good is that unlike with forward topology PSUs, you will get the full rating of the rectifiers because in forward topology, which uses half-wave rather than full-wave rectification on the output, one of the two diodes in a double-diode package has to act as a “freewheeling” diode, to discharge the output inductor, complete the current path, and avert the output voltage from spiking negative. As the result of this, the freewheeling diode only takes one third of the load, and that means that the outputs parts will only be capable of about 70% of their rating in forward topology.

                  That said, I noticed the +12V rail uses two parts in parallel: this will not necessarily result in double the rating either, because they are in parallel, they may experience (if not properly cooled or coupled) a phenomenon known as “thermal runaway” where due to differing voltage drop characteristics (ever so slightly as no part is exactly the same), one diode package may be conducting more current than the other, which will cause one to run hotter than the other which will further lead to the other part dissipating even more heat, resulting in a cascade effect that will eventually cause at least one of the parts to potentially burn out. So two 16A parts, even in half-bridge topology, I would not consider the same as a single 32A part: more like 20-24A.

                  Of course, there are several other factors besides the rating of the output diodes that limit the output current rating as well - the width of the PCB traces, the main transformer, the output toroids (this PSU appears to use -52 material for the +3.3V coil, which is good because it’s more efficient than the older -26 material), the actual gauge of the output wires, etc. I think the manufacturers possibly take all those facets into mind when determining the final rating on the label... at least the PSU manufacturers who are honest.

                  Once again, sorry to follow up with yet another wall-of-text, but I tend to enjoy in depth discussion regarding PSUs as well.
                  Last edited by Wester547; 07-13-2018, 10:38 AM.

                  Comment


                    #10
                    Re: EZ-Cool ATX-650JSP ATX PSU recap.

                    Originally posted by momaka View Post


                    In general, Schottky rectifiers will give you better efficiency. They will also raise the 12V rail's voltage a bit, which may be helpful when the PC is used in a system that draws power primarily from the 12V rail (i.e. pretty much all computers since 2004-2005).
                    Fun fact: All Asus socket 462 motherboards draw from the +5V rail.

                    I can't recall Asus ever going to +12V for socket 462, unlike Abit, and likely Gigabyte.
                    (A good chance that Epox switched to +12V for later socket 462 motherboards as well, possibly Chaintech, Soltek and Shuttle as well.)
                    (Possibly Biostar as well?)
                    Last edited by RJARRRPCGP; 07-13-2018, 08:35 PM.
                    ASRock B550 PG Velocita

                    Ryzen 9 "Vermeer" 5900X

                    16 GB AData XPG Spectrix D41

                    Sapphire Nitro+ Radeon RX 6750 XT

                    eVGA Supernova G3 750W

                    Western Digital Black SN850 1TB NVMe SSD

                    Alienware AW3423DWF OLED




                    "¡Me encanta "Me Encanta o Enlistarlo con Hilary Farr!" -Mí mismo

                    "There's nothing more unattractive than a chick smoking a cigarette" -Topcat

                    "Today's lesson in pissivity comes in the form of a ziplock baggie full of GPU extension brackets & hardware that for the last ~3 years have been on my bench, always in my way, getting moved around constantly....and yesterday I found myself in need of them....and the bastards are now nowhere to be found! Motherfracker!!" -Topcat

                    "did I see a chair fly? I think I did! Time for popcorn!" -ratdude747

                    Comment


                      #11
                      Re: EZ-Cool ATX-650JSP ATX PSU recap.

                      Originally posted by RJARRRPCGP View Post
                      Fun fact: All Asus socket 462 motherboards draw from the +5V rail.
                      Fun fact 2: NOPE.
                      Here is your exception: A7V8X-LA

                      Comment


                        #12
                        Re: EZ-Cool ATX-650JSP ATX PSU recap.

                        Originally posted by momaka View Post
                        Fun fact 2: NOPE.
                        Here is your exception: A7V8X-LA
                        Well, all the non-OEM ones that I saw, are pulling from the +5V rail, like most PIII systems.

                        Only one exception for an OEM, oops, L O L

                        Why didn't the Asus A7V8X-X (Via KT400) (2003, IIRC) have the same? And the same deal, even with the later nForce 2 versions of their motherboards, my 2004 Asus A7N8X-X. (nForce 2) Even the Asus A7N8X-E Deluxe (nForce 2) (2004, IIRC) was like that, IIRC, I can't recall using the 4-pin plug.

                        https://www.hardocp.com/article/2004..._a7n8xe_deluxe

                        https://www.pcper.com/reviews/Mother...A7N8X-E-Deluxe

                        I guess that's why I missed that. I was just counting the retail market.
                        Last edited by RJARRRPCGP; 07-14-2018, 06:50 PM.
                        ASRock B550 PG Velocita

                        Ryzen 9 "Vermeer" 5900X

                        16 GB AData XPG Spectrix D41

                        Sapphire Nitro+ Radeon RX 6750 XT

                        eVGA Supernova G3 750W

                        Western Digital Black SN850 1TB NVMe SSD

                        Alienware AW3423DWF OLED




                        "¡Me encanta "Me Encanta o Enlistarlo con Hilary Farr!" -Mí mismo

                        "There's nothing more unattractive than a chick smoking a cigarette" -Topcat

                        "Today's lesson in pissivity comes in the form of a ziplock baggie full of GPU extension brackets & hardware that for the last ~3 years have been on my bench, always in my way, getting moved around constantly....and yesterday I found myself in need of them....and the bastards are now nowhere to be found! Motherfracker!!" -Topcat

                        "did I see a chair fly? I think I did! Time for popcorn!" -ratdude747

                        Comment


                          #13
                          Re: EZ-Cool ATX-650JSP ATX PSU recap.

                          Originally posted by momaka View Post
                          Those would be their KZG series. They had a ton of problems with those up to 2008 or so, along with a few other ultra-low ESR series that were again only seen on motherboards (like KZJ).
                          All ultra low ESR wet Caps are crap, independently of the manufacturer.
                          And they could (and should) be replaced with Polymers. Kemet A750 seems like good replacements...

                          Comment


                            #14
                            Re: EZ-Cool ATX-650JSP ATX PSU recap.

                            Originally posted by Stefan Payne View Post
                            All ultra low ESR wet Caps are crap, independently of the manufacturer
                            Not sure if I agree with that.

                            The older Panasonic FJ/FL and Rubycon MBZ/MHZ were generally very reliable capacitors, at least compared to the likes of KZG/KZJ. And Nichicon HM/HN/HZ made after 2005 are also quite reliable. However, ultra-low impedance capacitors aren't very suitable to the output filter of older SMPS designs (as momaka has said before, it can introduce ringing into the DC output filter and can also cause oscillation in the control loop).

                            Comment


                              #15
                              Re: EZ-Cool ATX-650JSP ATX PSU recap.

                              Originally posted by Stefan Payne View Post
                              All ultra low ESR wet Caps are crap, independently of the manufacturer.
                              And they could (and should) be replaced with Polymers. Kemet A750 seems like good replacements...
                              Don't fall for that hype.

                              There are plenty of ultra low ESR caps that are 15+ years old (mainly Rubycon MBZ and Panasonic FL) and still perfectly fine. Just because polymers almost never show any signs of going bad doesn't mean that they don't. They already have much higher internal leakage current than most electrolytics, and over time it can get worse... possibly to the point of taking something out on your mobo without any external signs. Of course these days it's mostly chipsets that fail more than anything. So you're more likely to get a failed board from that even if it did use crappy caps.

                              Originally posted by RJARRRPCGP View Post
                              Well, all the non-OEM ones that I saw, are pulling from the +5V rail, like most PIII systems.
                              True

                              Originally posted by RJARRRPCGP View Post
                              Only one exception for an OEM, oops, L O L
                              Hehe, I guess I've probably spent too much time working on OEM PCs.
                              Last edited by momaka; 07-16-2018, 06:25 PM.

                              Comment

                              Working...
                              X