Announcement

Collapse
No announcement yet.

DTK 400W PSU Power Good signal 0V fault

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

    DTK 400W PSU Power Good signal 0V fault

    Hello. I have this old PSU for my old computer. DTK PTP-4007P.
    The computer shuts down quickly when it boots up. Sometime in the middle of the boot. The computer is old, AMD Athlon 64 2800+ based setup (it has same low end graphics card, low power), the PSU is more than capable of powering it.

    I have noticed that when the computer was running, the Power Good (PG) signal from the PSU was high (above 4V), but when the computer shuts itself off, the PG signal goes to 0V.

    The PSU provides all voltages, but the PG is 0V even when the PSU is on (PS-ON shorted to GND).

    The only thin I have noticed is that the fan inside the PSU barely turns. I have measured the voltage at the fan connector on the board and it only gives about 7V. I have tried the better fan, and it turns better, but of course the voltage is still 7V.
    I am guessing this is also monitored, and that is why PG signal goes low.
    But I have traced the fan's connector pins a bit, and it seems that it goes from 12V (yellow) to one of the pin directly. And the other pin via the diode goes to a 5V rail (red). That means that the 7V is correct. But that is very strange. Although it does not have the original fan, I am pretty sure that it uses a standard 12V fan. And I remember years before, the fan was working normally, like a regular 12V fan. So I really have no idea.

    The PSU is built around standard TL494 controller and it uses an LM339 Op-Amp for other functions (including PG signal output).
    The PG output is on the pin 13 of the LM339. The input pins are 10 (-) and 11 (+). The voltage on the pin 13 is of course 0V. The voltages on the pin 10 is 2.5V and the pin 11 is 0.4V. The pin 13 goes to the base of a BJT, that goes to the PG wire.
    The Op-Amp pins are very hard to trace, they go all over the place, in circles, via various resistors and so on.

    I have checked the input caps, they are fine. I used the LCR meter. I checked all the diodes and transistors and resistor around the LM339. I just can not find the problem.

    I can not find schematics anywhere. I have only found some forum topic from years ago, where the same PSU is mentioned, but it is the topic that I started...

    Here are the pics:
    https://drive.google.com/file/d/1prN...ew?usp=sharing
    https://drive.google.com/file/d/1k52...ew?usp=sharing
    https://drive.google.com/file/d/1OVW...ew?usp=sharing
    https://drive.google.com/file/d/1-6o...ew?usp=sharing
    https://drive.google.com/file/d/1aep...ew?usp=sharing
    https://drive.google.com/file/d/1BKR...ew?usp=sharing

    #2
    Re: DTK 400W PSU Power Good signal 0V fault

    With the voltage at the LM339 comparator pin 11 being lower than the voltage at pin 10, the output at pin 13 should be "low", which it is. So that comparator section is working. You said that the voltage at pin 13 then goes to the base of a transistor, whose collector voltage is PG. With the base low the collector should be high, but is not.

    The problem seems to me to be in that transistor circuit (unless .4V is sufficient to turn on the transistor).
    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: DTK 400W PSU Power Good signal 0V fault

      Sorry, I made a mistake. The pin 13 goes directly to the PG wire.

      Something at the inputs of the Op-Amp is the problem (probably leading to the pin 11), but it is very hard to trace on the board.

      Pin 11 goes to the voltage divider, a 100 Ohm series resistor through a diode to the pin 14 (the output of the another comparator). But pin 11 also goes in different direction, eventually leading to the transistor.

      Now I am checking the other comparator. Pin 14 is 0V, pin 8 is 1.25V and pin 9 is 0V.

      Now I am tracing the pin 8 and 9. I found a cold solder joint on the 1uF 50V capacitor. Soldered it. Checked the PG, it is still 0V. Checked all the voltages again,and the blue wire (-12V) is 0V. I am not sure if I checked that voltage before or not.

      Now I am checking out the -12V rail.


      EDIT: I have changed the two 220zF 16V capacitor on the -12V rail and now that rail is fine it measures -11.2V. But the PG signal is still 0V.
      Last edited by PanicMechanic; 05-05-2021, 03:16 AM.

      Comment


        #4
        Re: DTK 400W PSU Power Good signal 0V fault

        I got nothing...

        Any ideas?

        Comment


          #5
          Re: DTK 400W PSU Power Good signal 0V fault

          That's a Macron Power PSU - those are very nice and reliable, but always come with garbage quality capacitors from the factory.

          I see the output capacitors in your PSU have already been changed once... or at least, I don't believe those 16 mm Teapo/Yageo caps are original to the PSU (and TBH, I didn't even think this PSU could fit 16 mm caps - whoever did this was an "enthusiast", to say the least.)

          Anyways. I'm 99.9% convinced your problem lies in the electrolytic caps - that is, all of the output ones and the small ones next to the ICs. In fact, I see a lot of the small caps are still the original "GL" brand, and those are pure garbage. You should do a full recap of the PSU. The only caps that you *don't* need to replace are the primary input 200V caps - those almost never go bad, even in these PSUs.

          Here is a Macron Power PSU I recapped a while back and still continue to use to this day:
          https://www.badcaps.net/forum/showpo...postcount=2230
          ^ This one is based on the same "ATX9912P" platform as yours.

          Also, it's ironic you posted about this PSU, because just a few months ago, I purchased another similar Casing/Macron Power PSU... this one based on the "ATX9806b" PCB. Despite being a new-old stock (NOS) and with nearly 0 hours on the clock, it had developed bad caps just sitting on the shelf. So it goes to show how bad the original caps are in these PSUs. I did a quick recap last month just to test it properly, and then finished re-capping it yesterday, along with doing a few mods today. Guess what? It's 100% solid under any load now.

          The fact that your PG is at 0V means either the output voltages are not stable or proper (and it's bad enough that the feedback circuit is detecting this) or the small caps near the ICs have gone bad, making the PG signal not work... and hence why your PC is shutting down.

          So yeah... get the soldering iron out and get those crappy caps replaced.
          Since you have an LCR meter, you can check which ones were the troublemakers... though I strongly suspect you will find more than one bad cap in there, especially when you get to the small 5x11 mm ones.

          *EDIT*
          By the way, as a future request, please upload your pictures to badcaps.net using the forum's attachment function. If someone is to ever look up this thread a few years down the road, there's a good chance those off-site images may not work (at least from what I have seen, people either remove their images from off-site hosts or the off-site host goes under and the images "get lost".)
          Last edited by momaka; 05-06-2021, 01:17 AM.

          Comment


            #6
            Re: DTK 400W PSU Power Good signal 0V fault

            Yeah, I was the one who changed the output caps on it, years ago. I had put a very low ESR Japanese capacitors in it, an after that it developed a fault. The output was oscillating at about 800 Hz and 1.2 Vpp on the 12 V line (under 100 W resistive load). Not the ripple, the ripple was low, about 20 mVpp or something. It seemed that the TL494 does not like very low ESR output caps. So I replaced those with worse caps (higher, but still "low ESR") ones, and the situation was improved. It was still oscillating, but at much lower peak voltages (but still about 200 mVpp if I remember correctly), so it is not ideal.

            Now that I remembered all of that, it made me think that those oscillations are the cause of Power Good signal going low.
            I havo to go now, but when I get back, or tomorrow, I will check the output under load again with an oscilloscope and post the screenshot here.

            Comment


              #7
              Re: DTK 400W PSU Power Good signal 0V fault

              Originally posted by PanicMechanic View Post
              Yeah, I was the one who changed the output caps on it, years ago. I had put a very low ESR Japanese capacitors in it, an after that it developed a fault. The output was oscillating at about 800 Hz and 1.2 Vpp on the 12 V line (under 100 W resistive load). Not the ripple, the ripple was low, about 20 mVpp or something. It seemed that the TL494 does not like very low ESR output caps. So I replaced those with worse caps (higher, but still "low ESR") ones, and the situation was improved. It was still oscillating, but at much lower peak voltages (but still about 200 mVpp if I remember correctly), so it is not ideal.
              Yeah, that's quite possible. Many of these older half-bridge PSUs don't like ultra-low or very low ESR caps. That said, I did use United Chemicon KZE 16V, 3300 uF cap on mine above for the 12V rail, and it seems to be pretty solid in terms of operation (at least I never had a PC crash because of it.) Otherwise, mine too oscillates a little when under light load, though I didn't measure it (don't have a scope). I can just hear it "whining" slightly when not loaded much. Once loaded up, though the noise disappears.

              In any case, I still suggest you at least check the small caps first, especially the GL -branded ones. At this age, they really probably have failed. On mine, they were Fuhjyyu... which as bad as people say this brand is, isn't nearly as bad as GL and CS, IMO, as mine are still in spec. The newer Macron that I got came with PCE-TUR caps, and those also turned out to be OK. But the CS-branded caps on the output were starting to fail high leakage/capacitance.

              Comment


                #8
                Re: DTK 400W PSU Power Good signal 0V fault

                Still no PG signal.

                I checked the output voltages on the oscilloscope (no load, couldn't be bothered to make a load), and there where no oscillations, the ripple was "OK".

                I changed 3 ore 1uF 50V caps. Lubricated the fan, so it spins normally.
                I did not check the output caps, the thermal mass on the PCB is very high, so it is not easy to get them out.

                On Sunday, I will try the PSU with the computer again to see what happens. If is still shuts down, I will just get another PSU. It is no longer worth it to spend time on this one.

                Comment


                  #9
                  Re: DTK 400W PSU Power Good signal 0V fault

                  If it does not work, I will test the output caps...

                  Comment


                    #10
                    Re: DTK 400W PSU Power Good signal 0V fault

                    I have checked the output caps. They are all like new.
                    2 Yaego and 3 Jamicon 3300 uF caps measure 3000+ uF at 100 Hz and 20 miliohms at 100 KHz.

                    The only thing that does not seem right is the 7V fan voltage. Why it is not 12V? And why is it routed between 12V and 5V rail, making those 7V?

                    And -12V rail is about -10,5V or something, but that rail is always lower than nominal, right?

                    Power Good is still 0V, of course...

                    I have no idea...

                    Comment


                      #11
                      Re: DTK 400W PSU Power Good signal 0V fault

                      Originally posted by PanicMechanic View Post
                      I have checked the output caps. They are all like new.
                      2 Yaego and 3 Jamicon 3300 uF caps measure 3000+ uF at 100 Hz and 20 miliohms at 100 KHz.
                      Sounds like they are good then... but make sure they don't measure too much above 3300 uF. That is, these are electrolytic caps with 20% tolerance on the capacitance specs. Therefore, a cap measuring as low as 2640 uF or as high as 3960 uF can still be considered in spec. Of course, with cheap brands, once I see the caps starting to read close to 15-20% high on the capacitance, I consider them suspect. Many cheap capacitors fail in this failure mode: first their capacitance increases a lot... as does their internal leakage current (electrolyte attacking the oxide layer on the foil plates). Then gas build-up inside the cap ruptures the vent, causing the electrolyte to dry out, which in turn makes the capacitance decrease and ESR increase.

                      That said, it doesn't sound like your output caps are starting to fail, given they still show healthy ESR that isn't extremely low (usually, that's another thing that accompanies the increase in capacitance when cheap caps start to fail.)

                      Originally posted by PanicMechanic View Post
                      The only thing that does not seem right is the 7V fan voltage. Why it is not 12V? And why is it routed between 12V and 5V rail, making those 7V?
                      Only the cheap nasty power supplies (and perhaps maybe a few server ones) wire their fans directly to 12V. Most at least half-decent power supplies will have a fan controller in there that feeds the fan lower voltage and increases it only when the PSU starts running a bit warmer.

                      7V for these Macron PSUs is actually perfectly normal. I checked my ATX9806 the other day, and it also was about 7V cold.

                      Now, Macron did indeed do a "weird" design with the fan controller, where they wired it between 12V and -5V rail (or was it -12V rail and 5V rail, I don't remember exactly.) But they did this so that to put extra load on one of these negative rails, IIRC, for better voltage balance. Also, if the PSU really starts overheating, technically that fan controller could supply the fan with up to 17V (minus a few diode/transistor drops)... though I've never been able to make the controller do that.

                      Originally posted by PanicMechanic View Post
                      And -12V rail is about -10,5V or something, but that rail is always lower than nominal, right?

                      Power Good is still 0V, of course...
                      Well, that might be why the PG is low.
                      Allowable voltage deviation on the -12V rail is +/-10% (but 5% for all other rails). Therefore, minimum (or is it maximum in this case ) allowable is -10.8V.

                      I somehow doubt the PSU might be able to detect that kind of a difference enough to pull PG low... but maybe it can??

                      Originally posted by PanicMechanic View Post
                      The PG output is on the pin 13 of the LM339. The input pins are 10 (-) and 11 (+). The voltage on the pin 13 is of course 0V. The voltages on the pin 10 is 2.5V and the pin 11 is 0.4V. The pin 13 goes to the base of a BJT, that goes to the PG wire.
                      OK, so I just re-read this from post #1.

                      Do you know what kind of BJT is connected to the PG wire? (As in, NPN or PNP type.) Also, how is it wired? - That is, what is the Emitter and what is the Collector connected to?

                      Given the two voltages at the inputs to the op-amp (2.5V for pin 10 and 0.4V for pin 11), this suggests the output at pin 13 should be low (0V)... which it is. So the op-amp is working correctly.

                      Thus, depending on the BJT type and the way it's wired, the low voltage on PG may be normal, or it may be abnormal.

                      If BJT is NPN with Emitter wired to ground, Collector to PG, and there is a pull-up resistor from a 5V or 3.3V source going to the PG wire as well (to BJT Collector), then PG signal should be high when its Base is low. If that's not the case, see why the pull-up resistor is not providing a voltage.

                      Also, if the BJT is PNP type with Emitter connected to PG and Collector connected to a 5V or 3.3V pull-up source (through a resistor, perhaps), with the Collector connected to PG as well... then there should be a pull-down resistor between ground and PG wire too. In this case, make sure pull-up source voltage exists and that the BJT is not open, because a PNP BJT with its Base pulled low (from the op-amp/comparator) and given the above configuration, then there should be a high level signal on PG pin.

                      Now, you may also have the reverse situation of how the BJT is wired to the PG signal (i.e. for NPN, C wired to pull-up and E wired to PG wire and pull-down resistor to ground... or for PNP, with C wired to ground and E wired to PG wire with a pull-up resistor), then 0V from the op-amp output on pin 13 should indeed be pulling the signal on PG low.

                      Originally posted by PanicMechanic View Post
                      If is still shuts down, I will just get another PSU. It is no longer worth it to spend time on this one.
                      I guess that does depend on how much time you have.
                      Personally, I do like to repair these older PSUs, as they are extremely solid and reliable. Yours looks to be even better-built than the Inno/Macron MPT-301 I showed above, with thicker heatsinks and better rectifiers. I've used mine in various PCs ever since I recapped it, and it has never given me any problems. Even the original fan, which was completely seized up, I was able to repair, and that still also works properly.
                      Last edited by momaka; 05-08-2021, 01:39 PM.

                      Comment


                        #12
                        Re: DTK 400W PSU Power Good signal 0V fault

                        My bad...

                        Originally posted by PanicMechanic View Post
                        Sorry, I made a mistake. The pin 13 goes directly to the PG wire.
                        The comparator output is directly connected to the PG wire.

                        I will check the -12V rail again tomorrow. And on another PSU as well, to see if it is closer to the -12V.

                        Comment


                          #13
                          Re: DTK 400W PSU Power Good signal 0V fault

                          It is actually -10.8V, and it rises to about -11.1V. It is supposed to be able to provide 0.8A, but when I load it with 50 Ohms, the voltage drops to -9V. That line has 2 caps in parallel, and I had changed them both a few days ago.

                          Comment


                            #14
                            Re: DTK 400W PSU Power Good signal 0V fault

                            Actually, ony a single cap for the -12V rail. 220uF 16V. Changed a few caps instead of that, Jamicon SK 220uF 16V, Nippon Chemi-Con KY 220uF 25V and even Nippon Chemi-Con KZG 1000uF 16V.
                            It behaves the same. With 50 Ohms load, the voltage drops from -11V to -9V.

                            Comment


                              #15
                              Re: DTK 400W PSU Power Good signal 0V fault

                              Hmmm. Maybe the rectifying diodes for the -12V rail is/are leaky. There should be 3 of them, I think: two for rectification, and one in series after that for dropping voltage down a bit. Should be small FR104 or similar diodes. The unpleasant news is they are located under the output toroid... so that will need to come out if you do have to replace them. But if you follow the traces on the PCB, you should still be able to check them in-circuit... though that may not be 100% reliable.

                              Also, there should be a 25V, 470 uF cap by the secondary heatsink and 5VSB transformer - make sure that's good too, as that's for the PWM IC power (auxiliary secondary rail.)

                              Originally posted by PanicMechanic View Post
                              Actually, ony a single cap for the -12V rail.
                              Yeah, I was going to say... one of these 220 uF caps is for the the -12V rail and the other is for the -5V rail.

                              50 Ohms load on -12V is 240 mA... so it shouldn't dip so low, especially if you have a motherboard connected as a load to the PSU. Now, if the PSU does not have a load when you did that test - particularly on the 5V rail, as these are older PSUs that "prefer" 5V-heavy loads - then that will keep the 12V and -12V rails a bit lower than usual. Adding load on 5V rail increases the 12V and -12V rails, and vice versa.

                              But here is a little trick you can try if you think the -12V rail is causing the issue: get a 12V *ungrounded* (double-isolated) adapter (switching type with regulation, because regular linear ones will usually have too high of a voltage unloaded) and connect it so that (+) on the adapter connects to GND on the power supply, and (-) on the adapter connects to -12V on the PSU. This will essentially feed -12V into the -12V rail and make the voltage right... but again, it has to be a 2-pin (ungrounded) adapter. Then fire up the PSU and see if PG signal still shows low. If it is, then you have an issue with something other than the -12V rail. But if PG is good, then the -12V rail was the issue.

                              Comment


                                #16
                                Re: DTK 400W PSU Power Good signal 0V fault

                                I actually did remove the output toroid, and checked everything under it. It all measured good. Now I have soldered it back, so too late to replace the diodes (but they do measure fine). :/

                                I think I have checked that cap, but I will check it again.

                                Everything I have been doing with PSU on the table (unloaded). I will try to load the 5V rail a bit.

                                Comment


                                  #17
                                  Re: DTK 400W PSU Power Good signal 0V fault

                                  OK, you where right. I have loaded the 5V rail with 5 Ohms, and the -12V rail voltage vent above -12V. And when loaded with 50 Ohms, it dropped to -11V (and a bit above).
                                  And... The PG signal is 5V.

                                  I also changed a couple of 10nF 100V film caps next to the comparator, but the PG was still 0V, untill I loaded the 5V rail.

                                  I did check the PSU a few days ago, with some 20 year old motherboard and CPU (on the table), and the fan on the CPU did spin, but the PG signal was 0V. Maybe it was not loading the 5V rail enough. It is a Celeron II, BTW. :p

                                  I have no idea. I will try it with my computer on Saturday, to see how it behaves.

                                  Comment


                                    #18
                                    Re: DTK 400W PSU Power Good signal 0V fault

                                    Well, I installed it back in the computer and it seems to work fine.
                                    Maybe it was that cold joints on that 1uF 50V cap. And the cap for the -12V rail.

                                    Comment


                                      #19
                                      Re: DTK 400W PSU Power Good signal 0V fault

                                      Originally posted by PanicMechanic View Post
                                      Well, I installed it back in the computer and it seems to work fine.
                                      Maybe it was that cold joints on that 1uF 50V cap. And the cap for the -12V rail.
                                      I guess we won't know if the issue doesn't re-appear... but that's a good thing then!

                                      Good to hear it's all working again. Like I mentioned, these Macron PSUs are very reliable after a recap, so probably yours should be too.

                                      Comment

                                      Working...
                                      X