Announcement

Collapse
No announcement yet.

BFG Tech BFGR550WGSPSU [LW6550SE Rev. K] – recapping & modding

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

    BFG Tech BFGR550WGSPSU [LW6550SE Rev. K] – recapping & modding

    Rather than post in the PSU build quality pictorial thread, again I am splitting off the recapping info into its own thread. Also, I plan on doing some possible modding/experimentation on this PSU, so I figured that info would also be better to go here.

    The PSU in question is this one:

    Detailed post with parts and build quality is here:
    https://www.badcaps.net/forum/showpo...postcount=3085

    So this PSU has very simple output filtering: two 2200 uF caps on each major rail (3.3V, 5V, and 12V.)
    Here is a cap diagram showing the caps on each rail:


    The 12V rail received 2x United Chemicon KYB, 16,V 2200 uF caps. These are close replacements for the original Teapo SC, in terms of specs.

    The 5V rail received 2x Rubycon YXJ, 10V, 2200 uF caps – an excellent match for the two bad Teapo SC 10V, 2200 uF

    For the 3.3V rail, I put 1x Rubycon ZLH, 6.3V, 2200 uF cap in place of the BH, 10V, 2200 uF cap. The ZLH is probably very over-kill. But the 3.3V rail of this PSU uses a mag-amp circuit for regulation. Mag-amp circuits usually don’t mind deviations in specs, particularly when lower-ESR caps are used. The second cap on that rail was a Teapo SC, 10V, 2200 uF that was bad. Since my stock of Ruby ZLH caps is getting very low and like to keep a few spare in case of an “emergency repair”, I tested the BH cap on my cap meter, determined it is still OK, and put it on the second (output) position. The Ruby ZLH should be handling most of the ripple/filtering, so that BH cap may not be stressed as much.

    For the -12V rail, the original cap was a single FCon GF, 16V, 100 uF. There is no 7912 linear regulator, but there is a PI coil. Still, I think 100 uF is way too little, despite the fact that the -12V rail is not really used anymore. So I replaced that cap with a Nichicon PM, 16V, 220 uF cap in 8x13 mm can. I was thinking of even using a Nichicon PJ, 16V, 470 uF… and honestly, I might end up swapping it. But for a “test run” to see how this PSU performs, I figured the upgraded capacity from 100 uF to 220 uF is already a good enough improvement.

    Finally, the 5VSB: on this one, I put a Nichicon HN, 16V, 1500 uF cap. This is an ultra-low ESR cap intended for motherboards and not really very tolerant of hot circuits. However, the 5VSB circuit of this PSU uses a DM0265r PWM-FET off-line switch IC, which makes the 5VSB fairly efficient. I measured the PSU drawing about 2 Watts at the wall with an old Pentium 4 motherboard connected (with WOL/WOR disabled for lower power draw.) So with that, there isn’t much heat generated from the 5VSB, and that Nichi HN should be fine. Moreover, 5VSB circuits tend to use flyback topology, which tends to produce a lot of ripple current. The Nichi HN have very high ripple current rating, so the circuit won’t mind. As for the 2nd cap on the 5VSB – this was also a bad 10V, 2200 uF Teapo SC. Removed that and replaced it with the FCon GF, 16V, 1000 uF cap (that was originally in place where the Nichicon HN is now.) Result: 5VSB is happy and stable. No winning noise from the ultra-low ESR of the Nichicon HN cap either.

    I also replaced a few small caps, but not every one. On the diagram above, I labeled a “Startup” cap on the primary side. This is for the 5VSB, but obviously the entire PSU will not work at all if that cap was bad. It’s a FCon KM, 50V, 47 uF, 6.3x11 mm. I replaced it with a Nichicon PM 50V, 47 uF, 6.3x11 mm.

    Likewise, there is another cap on the diagram above, labeled as “Secondary Auxiliary”. This rail provides power to the PWM chip and BJT driver transistors on the secondary side when the PSU is in Soft-off mode. With this rail unstable, the PSU may either not start or not work right. So it’s better to replace it. This cap is also an FCon KM, 50V, 47 uF like the startup cap. I replaced it with a Nichicon PM, 50V, 47 uF.

    And last but not least, there are two Teapo SC, 50V, 6.8 uF caps on the primary that are part of the driver circuit for the main PS BJTs. I didn’t have any 6.8 uF caps in stock, so I tested one of these, found it reading OK, and left it alone for now. It is recommended that these caps are replaced in half-bridge PSUs. I probably could have put together a 4.7 uF and a 2.2 uF cap in parallel for a total of 6.9 uF, but I didn’t feel like diddling around. If I buy caps in the future, I’ll just order some 6.8 uF caps then. Until that time, that is all the recapping I’ve done with this PSU.

    For anyone that wants to do an absolute full recap on this PSU, the PSU has these small caps that you will need:
    4x 50V, 2.2 uF, 5x11 mm
    2x 50V, 10 uF, 5x11 mm
    1x 50V, 4.7 uF, 5x11 mm

    And here is a picture of my finished PSU recap (at least for now):


    Also, a high-res solder-side shot, in case someone has one of these PSUs with SMD resistors burned:
    https://www.badcaps.net/forum/attach...1&d=1592073646

    In terms of modding, I haven’t done any yet. But I may increase the capacity on the -12V rail to 470 uF, as mentioned. Another thing to look into would be to add a plastic sheet under the back side of the cooling fan to hopefully optimize/improve the cooling inside the PSU better (not sure how good or bad it is right now, but it doesn’t look optimal to me, at least logically thinking about it.) On that note, I might also trace and test the fan controller circuit, and maybe try to improve it. The fan is not super-loud right now, but it is noticeable. Depending on how well the PSU is cooling, I may try to lower its speed a little more, which should improve both noise and put less dust inside the PSU. Not to mention the fan will likely last longer. In regards to that, PSU unit #2 (I bought two of these PSUs) had a seized fan and burned output inductor. Granted there was a ton of dust inside that one, suggesting the PSU ran for many hours, I still think it may have lasted longer if the fan was spinning slower (but enough to cool the PSU.)

    So those are the things that can still be improved on this PSU.
    I still haven’t fully tested my recapped PSU under a heavier load. But with an old crappy Pentium 4 motherboard and a 2 GHz Pentium 4 Northwood CPU (PC drawing about 60-65 Watts tops), the PSU was running cool, as it should. Given it has a 6-pin and an 8-pin PCI-E power connectors, I should be able to come up with a system that can generate some significant load on the 12V rail.

    Again, as mentioned in the build quality pictorial thread, I doubt this PSU can provide 550 Watts of power. It seems well-built, but I think it's probably rated closer to 350-400 Watts, honestly. It will very likely be able to provide up to 300 Watts on the 12V rail, which IMO is enough for a lot of single CPU, single GPU PC builds. And I very likely won't try to push anything more from it.
    Attached Files
    Last edited by momaka; 06-13-2020, 12:51 PM.

    #2
    BFG Tech BFGR550WGSPSU [LW6550SE Rev. K] – unit #2

    Like I mentioned in the previous post and the power supply built quality pictorial thread, I have two of these power supplies. Here is unit #2:


    Looks the same on the surface, but… a lot more dusty
    https://www.badcaps.net/forum/attach...1&d=1592259172

    … and with slight damage to the case.


    Inside pictures:




    It may not look that bad in the pictures above, but it was quite packed with dust – particularly between components inside. This PSU was definitely in service for many hours. Looks like what finally killed it was the fan. After getting seized, it let the PSU overheat and probably crash. That output coil for the 5V and 12V rails is burned to a crisp. Moving or bending any wire on the output toroid made its lacquer coating flake right off. It must have gotten really hot, because not only were all of the output caps bulged from getting cooked, but even the output wires have darkened (particularly note the yellow wires on the right side.)

    Frankly, I wasn’t even sure if this 2nd unit would work. It seemed worth a shot at trying to repair it, though. So here goes that. First, I removed the output toroid for inspection and the output caps.




    The good news is none of the secondary rectifiers were shorted. Miraculously, the output toroid didn’t show low resistance / short-circuit / leakage between the 5V, 12V, and -12V rails, despite looking so charred. I guess the fragile lacquer coating was still doing its job. The old caps were likely crap to begin with (mostly Teapo SC, except the 12V rail, which had KSC GL), so the heat didn’t help them any. If I recall correctly, they were all failed miserably, showing either 10’s of Ohms ESR or even open-circuit. Not sure if the lack of airflow from the seized fan killed the caps first, causing massive ripple to go through the output toroid and making it burn itself, or if the toroid overheated and killed the caps. Whichever the case, both are toast now and need replacing.

    After checking the primary side and determining that nothing else seems wrong with the PSU, I decided to try a spare output toroid from another PSU as a test. The spare output toroid I removed from a very crappy and gutless JNC PSU I disassembled over a decade ago – back before I even knew how to fix PSUs. I believe there are a few very old posts/threads of the same PSU as my JNC under the names Meico and/or M-tec, IIRC (for anyone curious to see it.) I looked through my junk bin of “crappy, but tested OK” caps and also found some to use as a sacrificial lamb. The result was this:


    Yup, it ain’t pretty.
    But it works!
    … at least with a very light load consisting of a 20W halogen bulb on the 12V rail and a 4W bulb on the 5V rail. I didn’t put heavier load because I was testing the PSU through a 450W heating element on the primary side (instead of an incandescent bulb trick, as this allows more load to be pulled from the PSU before the input AC voltage drops too low.) The PSU did squeal/whine a little, though. Not sure if it was the series 450W heating element or due to using a different output toroid. So more testing will need to be done in the future when the PSU is fixed.

    That aside, notice how small that test output toroid is, though – it’s about the same size as the 3.3V rail output toroid, if not smaller. That’s how crappy that JNC PSU was! The white wire you see behind it is what I had to add to connect the -12V rail winding, since it couldn’t quite reach the hole on the PCB.

    In any case, the good news are, this PSU works.
    The bad news are that I need to either rewind the old output toroid or find another one that’s bigger (I’ll probably try the former, first.) With that and new caps, this PSU might live another day to power up a PC. Also, given the amount of dust, I really will need to check the fan controller circuit to see if the fan could be / should be turned down to get less dust in the PSU (and also a more quiet PSU, and make the fan last longer.) The old seized fan, by the way, I was able to fix by taking it apart and cleaning its sleeve bearing thoroughly, then adding oil. It’s spins almost like new now.

    More work to be continued again, some day.
    Attached Files
    Last edited by momaka; 06-15-2020, 04:20 PM.

    Comment


      #3
      Re: BFG Tech BFGR550WGSPSU [LW6550SE Rev. K] – recapping & modding

      I think the sleeve bearing fan failed faster because it's mounted horizontally so the lubricant evaporated at a quicker rate (and the fan may have been noisier as well whilst mounted in this position). Also, the larger the fan, the faster the shaft wears out, IIRC, which can't help longevity. Of course, there are other factors, such as the ambient temperature of the operating environment - like crap caps, sleeve bearing fans don't take well to heat, or a poorly ventilated and cooled system. Did you replace all the caps, including the high voltage ones? If not, I wonder if they survived or not.

      There appears to be “350—450W” silkscreened on the PCBs. So they're probably really 450W PSUs in the best case scenario with mediocre efficiency.

      Comment


        #4
        Re: BFG Tech BFGR550WGSPSU [LW6550SE Rev. K] – recapping & modding

        This is the second time I've seen an overheated 120mm Yate Loon fan (the first time being in this Rhino PSU), I wonder if the bigger Yate Loon fans are prone to overheating when they seize up (Or maybe this is a lot more common than i think, but i do this stuff for a hobby so i don't see things like that often).
        I'm not a expert, I'm just doing my best.

        Comment

        Working...
        X