Announcement

Collapse
No announcement yet.

MSI MS-7184 (Amethyst M-GL6E) – cap diagram (and a repair log)

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

    MSI MS-7184 (Amethyst M-GL6E) – cap diagram (and a repair log)

    And here is the other board that came with the ASUS I posted about in this thread: an MSI MS-7184. From what I have seen in past searches, this appears to have been a really popular socket 939 board back in the day for HP/Compaq, eMachines, and Gateway tower PCs. Badcaps.net store even has a cap kit for it.

    The two reasons I bought this motherboard: it was cheap (got a good deal on it with the ASUS motherboard above) and because I was quite impressed by the build quality of that MSI MS-7191 motherboard I posted about last week. In fact, the build quality is what swayed me more than anything, because by the time I saw the MS-7184 on eBay, I had already set up my secondary PC with the MS-7191 and was really liking it.

    So here is how the board looked when I opened the box from eBay (note: I always take pictures when I get stuff from eBay, because sometimes the seller will cheap out on packaging, and there have been a few times where I ended up with a damaged item because of that – but luckily, this was NOT case here):



    https://www.badcaps.net/forum/attach...1&d=1499310663
    https://www.badcaps.net/forum/attach...1&d=1499310663
    https://www.badcaps.net/forum/attach...1&d=1499310663

    And a back shot:


    As you can see, there are quite a few Teapo SM caps near the RAM slots and the chipsets. None of them were bulged, though. As such, I decided to test the motherboard as is. This is where things got interesting (but if you don't care about my story/testing shenanigans, feel free to skip down - all the way to the bottom of the post – to get to the cap diagram.)

    I connected the motherboard to my test PSU, keyboard, and monitor. When I hit the power button, all fans kicked on for a split second and that was it. Looked almost like there was a short-circuit somewhere.

    I tried it one more time, and this time the motherboard turned ON. Unfortunately, the screen remained off, so the motherboard appeared dead. But then I noticed that, strangely enough, the keyboard Num Lock LED was turning ON and was not getting stuck in ON or OFF state when I kept hitting the Num Lock key (a good POST indicator for older motherboards). So it looked like the motherboard might have actually POSTed. I turned OFF the motherboard, reseated the RAM and did a few other things, then performed another test. But just exactly like the first test, the first time I tried to turn ON the motherboard, the PSU crowbar-ed. Then I tried again, and the mobo appeared to turn ON and POST. So something was definitely very screwy here.

    That's when I decided to test with an external PCI-E video card. Sure enough, when doing that, the motherboard POSTed fine and was displaying a normal image on my screen (after doing the funky PSU short-circuit the first time, of course).

    So why didn't the onboard video work then? And what was causing the board not to work when cold the first time?

    As I started disconnecting the cables on my test setup, I noticed something very funny: the onboard VGA D-sub nearly came off with my test VGA cable. How is that even possible? Did I seriously not notice that the VGA D-sub was damaged? Looking back at the pictures I took – there it was, the damaged D-sub sticking out and clearly visible on that back shot:
    https://www.badcaps.net/filedata/fetch?id=2200562

    After some careful disassembly and inspection of the D-sub, I noticed that the D-sub had many pins that were all mashed and shorted together. Perhaps someone pulled the D-sub out and then shoved it back in. One of the many mashed pins was a 5V supply connected to a PSU 5V rail through a reset-able poly-fuse. And sure enough, one of the pins it was shorted to was ground. This explains why the motherboard was making the PSU go into short-circuit protection: when powered for the first time, the reset-able poly-fuse allowed enough current to go through it to trip the PSU. Once it did that and got hot, it would cut out, thus removing the 5V-rail short to ground and allowing the motherboard to turn ON the second time.

    As expected, when I completely desoldered the VGA D-sub connector, the motherboard no longer had any problems turning on the first time. But it still bugged me to know if the onboard video actually worked after all of this or not. So I decided to repair the VGA D-sub on the motherboard. At first, I looked through my stash of hardware for a spare VGA connector. But then I decided repairing the old one.

    With some careful tinkering, I was able to open the VGA D-sub connector and straighten/align most of the pins back in it. Three pins, however, were too damaged for that: they were snapped in half. Luckily, their front parts (where they go in the D-sub connector and make contact with the male VGA D-sub pins) were intact. So I was able to solder thin extension leads to those pins. Then I soldered the whole connector back on the board.

    The two big outer pins that hold down the VGA D-sub were also snapped. So for that, I used a paper clip – I bent it around the VGA D-sub and soldered it. Then I melted hot glue everywhere around the VGA D-sub, thus encasing the paper clip too. I tried pushing and pulling the VGA D-sub harshly a few times, but it held fine.

    Finally, I tested the motherboard with the onboard video and… IT WORKS! So that repair was a success.

    Now all that is left to do is recap this motherboard. But I really need to buy me some 6.3V/10V 1000 uF caps in 8 mm, as it seems I always have a crap-ton of motherboards that need them, and I only have a handful (might actually have to pull a few from some dead motherboards). And I also don't want to waste all of my 8 mm Rubycon MFZ and Nichicon HZ on spots that don't need low-ESR caps (already did that enough with the MS-7191). So the recapping is still not done on this motherboard yet.

    That said, I am still going to post a cap diagram / cap map here, in case anyone runs into one of these motherboards. The cap map should be especially helpful when doing a poly mod, as I indicate the max voltage on each rail.

    Here it is:


    Again, as with the ASUS motherboard, I will forgo a detailed rail description here. The cap diagram contains most of that information, already (including arrows that show which caps are connected to which MOSFETs/regulators, as well as the voltage rail). Also, it may be helpful to check out the thread for that MSI MS-7191 if you haven't already, as that motherboard and this one share a very similar design in terms of how power is regulated. The only difference is how I actually labeled the voltage rails on my cap diagram above, as the numbers are in a slightly different order.

    #1 and 2: CPU 12V rail and CPU V_core respectively.

    #3: DDR V_DIMM voltage (2.5-2.6 Volts). Linearly-regulated rail, so no need for polymers or ultra-low ESR caps at all

    #4: RAM Vtt termination voltage (typically half of V_DIMM, or 1.25-1.3 Volts). Another linear rail, so same applies regarding caps as #3 above.

    #5: Northbridge V_core/Vcc (about 1.6V, if I remember correctly). Buck-regulated rail, so good candidate for polymod, if you like.

    #6: RAM / Northbridge secondary 2.5V rail (?). Notice that MSI used a good Japanese capacitor here (Nichicon HM), just like on that MSI MS-7191 motherboard. Clearly this rail is important if MSI chose a good cap there. Luckily, that means no need to recap this rail… unless you get Chemicon KZG.

    #7: Southbridge Vcc (i.e. power for SB), 1.85V, linear rail.

    #8: 5V rail input for Northbridge VRM high side (#5 above), so use good quality low-ESR caps here. Ultra-low ESR and polymers are a good idea too.

    #9: Power for P/S2 and USB ports. 5V. No need to explain more.

    #A: 3.3V STB or 5V active – a linear, always-ON rail, regardless of motherboard power state.

    That is all for this one.

    I have more motherboards to post, but I don't have cap diagrams for most of them yet. So there may be a delay posting some of them. I suppose the motherboard forums may catch a break from me now.
    Attached Files

    #2
    MSI MS-7184 (Amethyst M-GL6E) 2nd motherboard - recap

    Remember the MSI MS-7184 motherboard from this thread?
    That’s the 2nd one. I finally got down to recapping it about a month ago. (The 1st one pictured above in post #1 is still awaiting a recap, lol.)

    It’s interesting to note this 2nd board came with different capacitors. Namely, the CPU VRM low-side (CPU V_core) caps were United Chemicon KZG series (@ 6.3V and 1500 uF.)



    All 6 of these were bulged, showing a few Ohms of ESR on my cap meter and 100-300 uF each. No surprise the seller described the board as “not working” and mentioned that it started crashing frequently before it stopped working altogether.

    What’s more weird is the two United Chemicon KZG on the left in the above pictures, which are used for filtering Vcc for the Northbridge (I called this rail “NB V_core” for some reason… ) were OK. I suppose they weren’t under a lot of stress. Or perhaps they were just from a “better” batch? They measured in spec.

    And here are the rest of the caps on the motherboard:
    https://www.badcaps.net/forum/attach...1&d=1610519233
    https://www.badcaps.net/forum/attach...1&d=1610519233
    Mostly a bunch of 8 mm diameter Teapo SM rated for 6.3V, 1000 uF. They all looked good, so I gave them a pass for the time being - at least for trying a test recap on the CPU VRM.

    Speaking of which, here’s the temporary recap I did on the CPU VRM:
    https://www.badcaps.net/forum/attach...1&d=1610519233
    I used only 3x Nichicon HZ 6.3V, 2200 uF caps, which I guesstimated would be enough to handle my test single-core AMD Sempron 3200+. Not only that, but these caps had 2005 date codes (H05xx), which means they are from the problematic period - or at least the 6.3V caps are (I haven’t had a single problem with H05xx 16V -rated caps, but almost my entire stock of 6.3V HZ series with H05xx date codes have gone bad.) Of course, the ones I used above were reformed recently and tested good, so I kept them for precisely this kind of job - a test temporary recap.

    Needless to say, the motherboard POSTed and booted just fine on the first try. Everything seemed to work (minus the PS/2 keyboard port, but I’ll get back to that later) and the board didn’t crash loading a Windows XP boot CD. Thus it was ready for a more proper recap.

    First, the CPU VRM:

    I put 7x Rubycon MFZ (@ 6.3V and 820 uF) and 3x Lelon OCRZ organic solid polymer caps (@ 2.5V and 820 uF). Both of these came from dead/scrap Xbox 360 motherboards that were beyond repairable. Basically, I filled each cap spot the MS-7184 had for the CPU VRM output. Original total capacitance with the 6x 1500 KZG caps was 9000 uF. But now with the “new” caps (11x 820 uF): 9020 uF. Essentially, I was spot-on. ESR-wise, however, those 7x MFZ and 3x OCRZ polymers have substantially lower total ESR compared to the original circuit. So this board got a nice upgrade. Moreover, it allowed me to decrease my stock of those MFZ caps. Many times, they are too tall and rarely fit anywhere else.

    The Northbridge Vcc (#5) rail also received a slight upgrade. Despite both of the 6.3V, 1800 uF KZG caps testing normal for ESR and capacitance, I replaced one of them with a 10 mm diameter Rubycon MFZ @ 6.3V and 2700 uF. Reason for going up in capacitance is because I didn’t have any 1800 uF or 1500 uF caps in 8 mm diameter. And I figured (since I’m replacing only one of the 1800 uF KZGs), if the remaining one failed, the 2700 uF MFZ should still be able to handle the filtering alone. Moreover, my stock of pulled Rubycon MFZ 6.3V, 2700 uF is now getting a bit aged. Thus, here was a good opportunity to use one. Fitting a 10 mm cap in that spot was not hard at all. However, it wouldn’t have been possible to fit two of them, so that’s why I replaced only one of the 1800 uF KZGs.

    Meanwhile, the input for both the CPU VRM (12V rail from PSU) and the Northbridge VRM (5V rail from PSU) is handled with 16V Nichicon HM caps (original to the motherboard.) These also had 2005 date codes. But being 16V caps, I just pulled one, checked it, and put it back since it measured spot-on for capacitance and ESR.

    I was also tempted to leave the Teapo SM caps for the time being, because I wasn’t sure when this motherboard will go in service. However, I pulled a few anyways and measured them. I’m glad I did, because this is what they all showed, more or less:

    Over 1200 uF on a 1000 uF cap is slightly more than the 20% allowable tolerance. Looks like these caps are developing high internal leakage. My experience with Teapo SM is that generally they aren’t as bad as Teapo SC when it comes to this, so they could still last a few years. But I didn’t want to do a recap on the recap (a re-recap? ), so I decided to pull these and replace them too.

    Because I don’t have any 6.3V, 1000 uF in stock to replace those SM’s, I used a mix of 820 uF (Rubycon ZLH) and 1200 uF (Rubycon ZLQ) caps when there were 2 or more parallel spots, allowing me to keep the total capacity pretty much the same as it was originally (not that 820 uF -only wouldn’t have worked - I’m pretty sure that it would.)

    So here’s the rest of the board recapped:


    And a large res. pic.:
    https://www.badcaps.net/forum/attach...1&d=1610519233

    There! No more crappy caps! As far as UCC KZG goes, only three were left on the board: 2x 10V 1000 uF for the USB ports (1x in USB ports area and 1x near F_USB header) and 1x 1800 uF on Northbridge Vcc rail (#5). I left them to conduct a test on how long they will last. A failure for any of these will not cause issues for the board, due to the other caps I have replaced. There were also a few small G-Luxon caps left, but I check that these didn’t filter anything important as well, so I left them alone (I pulled and checked one, but it measured good.)

    This leaves only the non-functional PS/2 keyboard port. I thought the recap would fix this, but it didn’t. I even added an extra cap in that empty cap spot by the PS/2 ports. When that didn’t help either, I investigated further. Turns out, the issue was a damaged pin in the PS/2 port itself. Not sure what the original owner did there to cause that. But I’m very particular about having PS/2 ports on my motherboards, so I removed the entire PS/2 port stack and will attempt a repair one of these days.

    Anyways, that’s all for the recap on this motherboard. I replaced the Sempron64 3200+ with an Athlon64 3500+ and set up Windows XP SP2 on a hard drive to give it a test. Aside from the busted PS/2 keyboard port, everything else works on this motherboard. It’s 100% stable, too… not that I expected anything less. These motherboards are very reliable with good caps. Too bad the BIOS on this one is branded by Compaq and offers no OC options of any kind (as usual.) Southbridge chip also runs a bit on the hot side and could use a heatsink. But apart from that, it’s a solid motherboard.
    Attached Files
    Last edited by momaka; 01-13-2021, 12:33 AM.

    Comment


      #3
      Re: MSI MS-7184 (Amethyst M-GL6E) 2nd motherboard - recap

      Originally posted by momaka View Post
      Remember the MSI MS-7184 motherboard from this thread?
      Nope, must have missed that thread, will see if I have anything fun to post in it!
      Originally posted by momaka View Post
      What's more weird is the two United Chemicon KZG on the left in the above pictures, which are used for filtering Vcc for the Northbridge (I called this rail “NB V_core” for some reason… ) were OK. I suppose they weren't under a lot of stress. Or perhaps they were just from a “better” batch? They measured in spec.
      Well how about posting a picture where the date codes are actually readable then?
      "The one who says it cannot be done should never interrupt the one who is doing it."

      Comment


        #4
        Re: MSI MS-7184 (Amethyst M-GL6E) – cap diagram (and a repair log)

        Great cap buff with the addition of polymers. Which tool are you using for removing old caps ?

        *Are KZG still that bad ? Got a pack of 100pieces 1500uf/1000uf for 4$ and I can't think they will be worst that chongx or the other junk OEM pcb's use.

        Comment


          #5
          MSI MS-7184 (Amethyst M-GL6E) 2nd - KZG caps

          Originally posted by Per Hansson View Post
          Well how about posting a picture where the date codes are actually readable then?
          Ah, right, I forgot you keep a record of UCC KZG date codes.
          And excuse my 2.1 MP PowerShot camera for not being able to capture enough details in the wider shots above. It is 20 years old now, after all.

          Here, how's that?
          https://www.badcaps.net/forum/attach...1&d=1610602849

          I also verified that both NB Vcc caps and both USB filter caps have the same date codes (just mentioning this, since I only took picture of one each.) The CPU VRM ooutput (V_core) caps all had that same variety of date codes - i.e. either "4a" or "4c" at the end.

          Curious if you have anything to say regarding those date codes above.
          And I'll try to remember to take note next time I post about a mobo with UCC KZG... which should be sometime this month (maybe?)

          Originally posted by zunasthegreat View Post
          Great cap buff with the addition of polymers. Which tool are you using for removing old caps ?
          Thanks!

          I'm using a Circuit Specialists CSI 2900 75-Watt soldering station (rebranded Aoyue 2900, IIRC) with a D52 tip (5.2 mm bevel type). It's a T-12 -based station, so it has pretty good thermal recovery on the tip. Quality on the wiring inside was a bit not-so-great when I got it, but it worked fine. I fixed some of the potential problems (mainly spotty-looking solder joints on the control PCB, and a few pinched wires), and so far it has been faithful to me in the last 9 years I had it.

          Originally posted by zunasthegreat View Post
          *Are KZG still that bad ? Got a pack of 100pieces 1500uf/1000uf for 4$ and I can't think they will be worst that chongx or the other junk OEM pcb's use.
          I guess it depends on when they were made.
          Newer KZG seem to be OK so far, at least from what I have seen. I think 2008-2009 is around when they started getting better. Older ones seem to be pretty spotty, though. Among those older ones, however, the 16V -rated caps seem to be OK more often than not (with exception of 16V 470 uF.) The 6.3V KZG, on the other hand, seem to be very prone to failure - especially the 820 uF, 1500 uF, and 3300 uF parts. That said, I still don't trust the newer 6.3V 820 uF KZG, just in case.
          Attached Files
          Last edited by momaka; 01-13-2021, 11:48 PM.

          Comment


            #6
            Re: MSI MS-7184 (Amethyst M-GL6E) – cap diagram (and a repair log)

            Ahh okay, it just looked visually like they where identical as the VCore caps.
            They are manufactured quite far apart to be on the same board, so hard to draw any conclusions:
            CPU Vcore caps are from April 27 & 29 2005
            NB VCC & 2.5Vref caps are from July 1 2005
            "The one who says it cannot be done should never interrupt the one who is doing it."

            Comment


              #7
              Re: MSI MS-7184 (Amethyst M-GL6E) – cap diagram (and a repair log)

              Momaka, where did you get that cap tester in Image 8, where you tested the Teapo cap?
              I have an ESR meter and a capacitance tester but nowhere do I get "voltage loss" data.

              Also you mention reforming your caps. How is this done? I'm assuming the caps have to be off-board to do this.

              It seems to me that wet electrolytic caps of the Low ESR type are being phased out by manufacturers so it's almost essential to reform your older stock.

              All my older Asus Intel boards are having trouble detecting hard drives. Am I correct in thinking that caps can test good and still be leaking voltage?

              Thanks for this! Much appreciated.

              Comment


                #8
                Re: MSI MS-7184 (Amethyst M-GL6E) – cap diagram (and a repair log)

                Originally posted by bigbeark View Post
                Momaka, where did you get that cap tester in Image 8, where you tested the Teapo cap?
                eBay... though I'm sure you can find these on AliExpress, Amazon, and many other online retailers.

                Mine was simply cheapest GM328 -based tester I could find back then with the features I wanted. Nothing special about it, otherwise. There are many versions of these available (and mine probably isn't the best out there anymore either, since it has rather old firmware now.)

                So just look up "GM328" on an online retailer, and you should be able to get some results. Before you get one, though, you may want to check this thread out, since there are actually upgradable aspects on these meters.

                In fact, I think this one is already pre-ugpraded with some options, like the rotary encoder and socket IC, along with newer firmware (so it should have better resolution than mine):
                https://www.ebay.com/itm/TFT-LCD-GM3.../303819675907?

                Of course, if for whatever reason you really want the same exact unit as mine (not sure why), I was actually able to find one seller that had it:
                https://www.ebay.com/itm/LCD-GM328A-.../311742432137?

                ^ But again, feature-wise, mine is not the "latest and greatest" anymore. It has proven itself as a reliable tool, though.
                Just beware that if you do test capacitors with it, make sure the capacitors are fully discharged first - especially high-voltage caps.
                Also note that these testers are not intended for in-circuit measurements of components (since they are "component" testers, after all, and can get a bit "confused" as to which component to measure.)

                Originally posted by bigbeark View Post
                I have an ESR meter and a capacitance tester but nowhere do I get "voltage loss" data.
                To be completely honest, I don't really look at the "Vloss" values too much when testing caps, as I don't think it's always too accurate on these meters. You really need just capacitance and ESR - these will tell you if a cap is bad like 99.999% of the time, if not more.

                The ESR really is the most important parameter. However, if a capacitor reads grossly much higher capacitance than what is stated on its sleeve - now that's a leaky (electrically) capacitor. Here's an example of old Sanyo WF breaking down like this:
                https://www.badcaps.net/forum/showpo...&postcount=974
                ^ Note how these read perfect ESR, but the capacitance is much higher (3000-4000 uF while the capacitor is rated for 1800 uF only! ) In the case of the 1st cap in the picture in that post, the component tester was indeed able to identify the Vloss as very high (27%). However, the 2nd cap in that post, which still read much higher capacitance, did not show any abnormal Vloss. So that's why I don't really trust the Vloss too much on these meters. Ultimately, the capacitance and ESR is all one really needs to know.

                Originally posted by bigbeark View Post
                Also you mention reforming your caps. How is this done? I'm assuming the caps have to be off-board to do this.
                See this thread:
                https://www.badcaps.net/forum/showthread.php?t=50530

                And yes, the capacitors have to be off the board.

                I also probably didn't state it too clear in that thread, but reforming already faulty capacitors will not restore them in any way. The reform process is really more to be used for used or NOS capacitors that are good, but have sat for long time in storage without being powered.

                Originally posted by bigbeark View Post
                It seems to me that wet electrolytic caps of the Low ESR type are being phased out by manufacturers so it's almost essential to reform your older stock.
                For the good Japanese brands, I'd say don't worry if the cap has sat for less than 3-5 years in storage in room temperature - most will work absolutely fine without reforming. But if you're trying to use salvage Chinese/Taiwanese/Korean caps (I sometimes do it on hardware that I'm testing), I'd say do the reforming more often.

                Also, low-ESR eletrolytic caps will never be fully phased out, because power supplies still use them. Sure even some PSUs have moved to polymers too... but when it comes to cheaper-alternatives, the low-cost of wet electrolytic caps are still hard to beat. Now as for ultra-low ESR electrolytics used on older motherboards - those were indeed majority phased out for quite a few years now. However, something like Panasonic FR and FS along with Rubycon ZLH will still suffice in most cases.

                Originally posted by bigbeark View Post
                All my older Asus Intel boards are having trouble detecting hard drives. Am I correct in thinking that caps can test good and still be leaking voltage?
                No, if the caps measure good on ESR and capacitance within spec, then they are good. (Note: this means the capacitance should be within the +/-20% range of the stated value on the sleeve. Any more or any less than that means the capacitor could be marginal.)

                If they are all not detecting the hard drives, I would suspect issue(s) either with the hard drive or cable... or perhaps a default configuration on these motherboards that is somehow not able to pickup the HDDs properly (download and check the manual thoroughly for each motherboard to make sure you didn't skip some config option in the BIOS.)
                Last edited by momaka; 02-13-2021, 12:18 AM.

                Comment


                  #9
                  Re: MSI MS-7184 (Amethyst M-GL6E) – cap diagram (and a repair log)

                  Momaka, thanks so much for all this information! Made my day

                  Comment

                  Working...
                  X