Announcement

Collapse
No announcement yet.

Coolmax RM-850B failure... just wanted to share

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

    Coolmax RM-850B failure... just wanted to share

    I got this Coolmax RM-850B. 850W? Yeah, right. I think it's more like 400W or so. But anyway I got it assuming it was dead, mainly for parts. But for grins I decided to open it to sneak a peek at what exactly failed.

    So the normal PSU debugging: Plug in and try it!:

    +5VSB: DEAD

    Okay. So I'm dead in the water here and have to open it up to take a peek.

    So I see a UCC3818AN APFC controller and TNY268P 5VSB switcher. Strange, TNY268Ps should be fairly reliable.

    Okay. Need to see where I can tap off the board. So I pulled the board and see this:

    Wow... Reconnect maybe? Okay, now the bottom:

    Holy crap, now what happened to the insulation?

    Thank goodness for three wire grounding and thank goodness for FR-boards.

    Now... is this fixable?

    Actually I think this is worth a shot jury rigging a fix...

    BTW: Yes, the FR4 track protected the input fuse... (Fuse is still intact). Will need to see if the input circuitry is still good or not, like the bridge - then again there are two bridge rectifiers in parallel so that's pretty beefy...

    And dim bulb will definitely be employed trying to repair this thing...
    Attached Files
    Last edited by eccerr0r; 07-29-2017, 12:47 AM.

    #2
    Re: Coolmax RM-850B failure... just wanted to share

    This is probably your quota of nearly useless information for the day, but I got curious, and I'm going to inflict it on ya'll.

    Fire Retardant Paper-Phenolic insulating material (including PC Boards) is FR-2.

    Fire Retardant Glass-Epoxy insulating material (including PC Boards) is FR-4.

    Fire Retardant Paper-Epoxy insulating material (including PC Boards) is FR-3.

    The insulation resistance of FR-2 changes under compression (not low enough to be dangerous). FR-4's insulation resistance is so crazy high that it's hard to measure. I tried once to measure the insulation resistance at 500V of an FR-4 shoulder bushing on a power line filter, and the only time the meter needle moved was when I moved my lab stool on the floor (not touching the filter or work bench, just the rubber casters moving on the vinyl tile floor)!
    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: Coolmax RM-850B failure... just wanted to share

      You could probably kludge (for UK folk, "bodge") a repair of that PCB, but how available is that insulating plastic sheet material, which also must be fire retardant.

      That is the second nastiest example of, "This is what happens when you have a bad connection in a high current circuit," that I've seen in the past couple of weeks. The worse example involved an edge connector, damage due excessive insertion force, and 7 years of operating at 40A-60A. Fried FR-4 and melty-fried 94 V0 plastic connector shell.
      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


        #4
        Re: Coolmax RM-850B failure... just wanted to share

        This kind of thing is like this video clip classic.
        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


          #5
          Re: Coolmax RM-850B failure... just wanted to share

          was the coil glued, or just floating?
          trying to work out the cause here!

          Comment


            #6
            Re: Coolmax RM-850B failure... just wanted to share

            The coil was just floating when I pulled the board!

            Yeah, that plastic insulator needs to be flame retardant as well, and that probably will be the toughest thing to find a good fix for. Need to make it as thick as possible:

            I can't believe the designer put that track so close to the edge of the board!!! This PSU looks so nice, the case is made of extruded aluminum, but this APFC coil basically is up against the edge of the board. I wonder if this would have failed a lot sooner if it were run in a 220VAC environment, as that node should be at 340VDC peak (it's unfiltered rectified mains). The unburned end of the APFC coil is the boosted voltage, so at least that is far away from the edge of the board.


            I thought I've only seen FR4 and FR2 boards, the tell tale glass fibers shows FR4. I suppose FR3 should be more prevalent than I think but I haven't heard much talk about it. Upon another closer look at the board, perhaps this really is FR3 as I can't see fibers as I thought I saw last night after using a magnifier...

            FR2 is very soft and mostly I see this with small, low voltage boards.

            I have an idea for the plastic sheet: I suppose the plastic sheet from another dead PSU gets to be used once more! (and probably will reuse the old sheet, patched with Kapton tape, as a second layer.) And this time ensure there are no sharp edges that can puncture the sheets or the tape.

            Comment


              #7
              Re: Coolmax RM-850B failure... just wanted to share

              you could glue a bit of fiberglass pcb without tracks to the case or plastic.
              it used to be used as an insulator before plastics were any good.

              Comment


                #8
                Re: Coolmax RM-850B failure... just wanted to share

                The clearance is so bad on this PSU.

                Perhaps I am stuck with adding a few layers of electrical tape?

                Comment


                  #9
                  Re: Coolmax RM-850B failure... just wanted to share

                  "FR3" is the designation I'd heard of before. I think paper-epoxy PCBs are fairly common in lower end consumer products. I much prefer G10/FR4, and that's what I've seen at the last several places I've worked (I may have seen some FR3 in low-end Delta PSUs, though I could be mistaken on that point, it being some 20 years ago). FR3 is decent, I'm just a bit of a snob.

                  That the inductor was loose - I didn't see any glue, and those wires underneath it scare the @#$% out of me. I'd put a layer or two of Kapton tape over them and then RTV the inductor in place. I'd even guess the inductor being loose plus shock/vibration may have compromised the original solder joint or wire.

                  Patching the burned/melted insulator with insulation from a scrapped/parted-out PSU sounds like a good solution. Just make the "new" patch a good deal larger than the hole, and maybe tack it in place with cyanoacrylate glue ("Super Glue", "Eastman 910", etc.).
                  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


                    #10
                    Re: Coolmax RM-850B failure... just wanted to share

                    Agreed, I like FR4, it's very durable, except it kills drill bits Have to use tungsten carbide bits on these, the HSS bits dull quite quickly on FR4.

                    ---

                    I ended up using many layers of electrical and Kapton tape, and tried the PSU for the first time to see what the current status of the supply is.

                    As I have a isolation transformer, I used it. I also used a 100W bulb as the dim bulb though did not use my Variac (I think dim bulb is better than variac honestly, for shorts testing!).

                    Of course I first I tested ground before hooking it all up... looks good.

                    When I first plugged in, I didn't see the initial flash. Weird. Okay, moving right along.

                    I tested +5VSB: Got 5.10V YAY! That's a good sign at least.

                    Then I hooked up a 47 ohm resistor to +12 and a 10 ohm resistor to +5V rails, along with an LED to light when power is received... because I don't have a bona fide PSU tester. I crossed my fingers (well, not really, I need my fingers to do measurements) and shorted GREEN to GND.

                    Fan came on solid (not full blast, it's one of these quiet fan controlled PSUs)!

                    LED on output CAME ON solid!

                    100W dim bulb turned on brightly! No wait... it turned off...and on... and off... and on... blink blink blink ad infinum.

                    WTFBBQ? No BBQ at least. Another BBQ would be bad. We already had a BBQ. I don't want another BBQ.

                    Hard to guesstimate the duty cycle but since there was a load on the output, the lamp should have lit..dimly...

                    As this is my first APFC PSU short debug, this is weird. Should it be blinking?

                    Next I added another load: a multimeter. No, not that. A 50W car headlight, yeah that's more like it. So total load is now about 60 watts. Turned it on...

                    The bright pulsing is still there but it's generally dimmer. Definitely some feedback loop due to the highly resistive source wire (dim bulb) and I suspect the switching transistor needed to be on fairly high duty cycle to make up for it.

                    But anyway, definitely no short detected, so I just hooked it up straight to the iso transformer and the moment of truth comes:

                    Shorted Green to GND and...

                    The headlight came on bright and solid. Meter gave me about 11.6 volts.

                    ---

                    Yeah I think I will need to RTV the APFC inductor, good point. Or at least somehow steady it.

                    Comment


                      #11
                      Re: Coolmax RM-850B failure... just wanted to share

                      Picture of the PSU and the kludge/bodge/jury rig/whatever you want to call it

                      I put some regular epoxy into the burned area to fill out the missing material due to the burn in hopes it would strengthen that area a bit.

                      I still need to support that APFC inductor, ideally away from the burn scar, so this is still a WIP... but comments about the original design of this PSU are open.

                      My initial thoughts -

                      - CRAP HV TOO CLOSE TO SIDE OF BOARD
                      - Riser for the APFC controller not insulated, just RTV'ed to the main filter capacitor, and the RTV broke.
                      - I would think the heatsinks and the main transformer would be larger for 850W.
                      - and as noted, the APFC inductor is not glued down properly.
                      - Crap cap brands "Micon" (most secondary caps) and "Fullcap" (primary cap)

                      Some good things
                      - Input rectifier survived a short to GND...
                      - Still a single sided PCB though there are SMT
                      - Riser boards are mixture SMT and through hole, but can probe the riser boards if needed since they're off to the side! I am quite PO'ed about this Antec 650W that I was debugging, the riser boards are in the middle making it impossible to hot probe. BTW, that Antec PSU is still dead...suspecting the output choke is shorted together but haven't confirmed it.
                      Attached Files
                      Last edited by eccerr0r; 07-29-2017, 12:41 PM.

                      Comment


                        #12
                        Re: Coolmax RM-850B failure... just wanted to share

                        Testing this PSU on my P4 right now The P4 was using that Antec SP400 that killed its 5VSB due to bad caps and subsequently its TL494. So far so good, this P4 is probably the second hungriest computer I have (I have another P4 that's worse as it has AGP graphics.)

                        Dang, my repaired PSUs in service exceeds the number of virgin PSUs. Currently my StealthXtreme2-600 and Antec NeoHE 380 remain working out of the box, knock on wood.
                        Last edited by eccerr0r; 07-30-2017, 08:26 PM.

                        Comment


                          #13
                          Re: Coolmax RM-850B failure... just wanted to share

                          Originally posted by eccerr0r View Post
                          Then I hooked up a 47 ohm resistor to +12 and a 10 ohm resistor to +5V rails, along with an LED to light when power is received... because I don't have a bona fide PSU tester. I crossed my fingers (well, not really, I need my fingers to do measurements) and shorted GREEN to GND.

                          Fan came on solid (not full blast, it's one of these quiet fan controlled PSUs)!

                          LED on output CAME ON solid!

                          100W dim bulb turned on brightly! No wait... it turned off...and on... and off... and on... blink blink blink ad infinum.

                          ...

                          As this is my first APFC PSU short debug, this is weird. Should it be blinking?
                          Yes, that is normal for APFC more or less. Depending on the APFC design, eventually the bulb flashing should stop once the primary cap comes to full voltage and if the load is light enough.

                          That said, once I have a PSU pass the incandescent light bulb test, I replace the incandescent light bulb with a 300-500 Watt heating element so that I can put bigger loads on the PSU without having it shut down. This works for both APFC and non-PFC PSUs, but APFC PSUs will behave much better with the heating element than the incandescent light bulb.

                          Originally posted by eccerr0r
                          Dang, my repaired PSUs in service exceeds the number of virgin PSUs. Currently my StealthXtreme2-600 and Antec NeoHE 380 remain working out of the box, knock on wood.
                          Hmm... I think my ratio is about 50:50 right now. Well, maybe I still have more "virgin" PSUs, still (if opening them to clean the dust out doesn't count, that is). Then again, perhaps I should also note that most of my computers are Pentium 4 era and older, so most have overbuilt HiPro, LiteON, Delta, and even a few Astec PSUs (which by the way, are of terrific quality with all high quality Japanese caps inside). But if we look at the PSUs I've collected over at my grandmother's house - those are mostly Deer/L&C/Allied PSUs that are absolutely terrible. I've had to do something to each one of them to make them work just acceptable.

                          As for what I think of the PSU in this thread:

                          Not bad. I'm not a fan of the APFC circuit when the primary cap brand is crap. If not too hard, I would definitely try to disable the APFC and just put a voltage doubler in there (if in the US or where there is 120V AC) or just leave the crap cap in there for countries with 220/230/240V AC line.

                          Also, looks like the 3.3V rail has an output inductor rated for lower frequency, so it may not be the most efficient. The output inductor for the other rails looks a bit small, but at least it is light green (meaning higher-frequency rated).

                          Overall, I don't think the PSU will make it up to 850 Watts. Not because of the heatsink sizes, but because it just doesn't look like a PSU capable of it. For 850 Watts, I would expect synchronous rectifier on the output... otherwise you can expect quite a bit of losses and heat.
                          Last edited by momaka; 08-05-2017, 12:16 PM.

                          Comment


                            #14
                            Re: Coolmax RM-850B failure... just wanted to share

                            BOOOOOOOOoooOOOOOOoooooo....

                            The PSU failed once more!

                            It's once more sitting on the triage table

                            It stopped producing power when I was using the PSU. Triage commences ...


                            ... oh. No smoke, no funny smells, no bright lights, just dead.
                            Last edited by eccerr0r; 08-04-2018, 09:00 PM.

                            Comment


                              #15
                              Re: Coolmax RM-850B failure... just wanted to share

                              Looks like power rails go way over:
                              +12V reaches 15V
                              +5V reaches 7V

                              hmm...wonder if a feedback LED croaked.

                              Oh, BTW: This thing uses a SMT UC3843 I think, barely can make it out...
                              Last edited by eccerr0r; 08-10-2018, 12:05 AM.

                              Comment


                                #16
                                Re: Coolmax RM-850B failure... just wanted to share

                                It's the auto-destruct sequence! Overvoltage stuff after trying to burn the house down

                                Comment


                                  #17
                                  Re: Coolmax RM-850B failure... just wanted to share

                                  It could be, but it shut down a few milliseconds after hitting the voltages!

                                  Comment


                                    #18
                                    Re: Coolmax RM-850B failure... just wanted to share

                                    Well... Took another look at it...

                                    The solder blob was insufficient! The APFC toroid solder job was poor and it worked its way loose. So I resoldered it and powered on.

                                    Yay 12V, yay 5V!


                                    POP! (smell of cat piss)


                                    Wow. Capacitor popped almost in my face just after a few seconds of power up after a long hiatus of it being nonfunctional. Perhaps it needed a reform but it got subject to 12V right away.

                                    Was a 2200uF 16V Micon capacitor. Junk... Alas I don't think I have any replacements *sigh* guess back to the shelf it goes, thought it at least still powers up...

                                    Comment


                                      #19
                                      Re: Coolmax RM-850B failure... just wanted to share

                                      Nope! After putting the PSU back into its case, it stopped working. Ugh. Looks like another cold solder joint somewhere... *sigh*

                                      Nice, it was never working. Now I know why the cap vented... The 12V line got to 23V!
                                      Last edited by eccerr0r; 05-06-2022, 12:33 AM.

                                      Comment


                                        #20
                                        Re: Coolmax RM-850B failure... just wanted to share

                                        Wonder who's the OEM. About a quarter of those units I've seen were Sun Pros, with the rest being ATNG usually...
                                        Main rig:
                                        Gigabyte B75M-D3H
                                        Core i5-3470 3.60GHz
                                        Gigabyte Geforce GTX650 1GB GDDR5
                                        16GB DDR3-1600
                                        Samsung SH-224AB DVD-RW
                                        FSP Bluestorm II 500W (recapped)
                                        120GB ADATA + 2x Seagate Barracuda ES.2 ST31000340NS 1TB
                                        Delux MG760 case

                                        Comment

                                        Working...
                                        X