Announcement

Collapse
No announcement yet.

Replacing electrolytic capacitors

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

    Replacing electrolytic capacitors

    Hello guys,
    I need an explanation from more experienced guys regarding the replacement of faulty (bumped) caps.

    I recently oversaw a replacement of faulty caps on two different motherboards. On one motherboard we have 5 faulty caps, 3 x 3300 uF and 2x 1500 uF 6.3 V. I have witnesed that all caps were replaced by 2200 uF 6.3 V and it worked ok.
    Is there any recommendation about how to proceed if I don't have exact capacitors? Usually it is beeter that replacement capacitor has higher voltage and a bit higher capacity (if same is not available). Is it possible that 1500 uF can be replaced with 3300 uF for example?

    Second mother didn't boot up properly but that is for another subforum. I wonder if I can try to order exact capacitors and to try the same repair on second one.

    #2
    Re: Replacing electrolytic capacitors

    Not only you should avoid, but you should never deviate from the capacitance of the original capacitor. If it is 1500, then you should replace it with 1500only.
    Capacitors can fulfill different functions in electronic circuits.
    Some of the basic functions of capacitors are:
    1. Filtration and smoothing
    2. Frequency filters
    3. As a part of signal generator or timing circuits.
    4. Reactive ballasts
    For example, you have RC resonator, which relies on the Time Constant.
    Time constant = R*C. If you replace the capacitor with one with different capacitance, then the circuit will not function properly if at all.
    For example NE555 square wave generator relies on external RC oscilator. Some SMPS IC-s require external timing circuit in order to work.
    Another example - changing the capacitor with wrong capacitance completely changes the frequency range of a High/Low pass filter.
    Not only this, but reactance of the capacitor is 1/(2*pi*fc*C) Replacing the capacitor with one with different capacitance means that the reactance and impedance of the circuit will change.
    While in the simplest usage of the capacitor in power supply with bridge rectifier and transformer, the capacitance of the smoothing capacitor is not that critical, in many cases changing it with the wrong one can even make the whole device to stop working or blow up in you face.
    The tolerances of the electrolytic capacitors are wide. +/- 20% is something normal. And their characteristics are highly dependent on the operating temperature.
    The general rule of the thumb is:
    1. Capacitors with the same capacitance, where the operating voltage of the replacement is the same or higher than the original are interchangeable. This means replacing 6.3V capacitor with 16V capacitor most of the time will not hurt. Or 10V with 25V.
    2. Capacitance always stays the same! Otherwise the circuit may perform out of it specs. The results can be unpredictable.
    Never ever do it in the primary side of PSU-s!
    3. Never replace the capacitor with another with lower operating voltage. Will guaranteedly blow up(EXPLODE) or fail,
    even if not right at the moment, in extremely short time.
    So, capacitance - always the same, voltage can be higher, but never lower. Also, without any troubles or fear you can replace 85C capacitor with 105C capacitor. Not only this, but using higher temperature capacitors is always better. Most of the time someone uses 85C is just to save money.
    Originally posted by Askic View Post
    Is it possible that 1500 uF can be replaced with 3300 uF for example?
    No! Even if it somehow works, your state of happiness will soon turn into a sad face.
    The frequency response of the circuit changes if you replace the capacitor with other with different capacitance.
    It's the same as replacing a flat tire with one that is 2 times higher than the original.
    If the capacitor is used only for smoothing, you can eventually get away with installing capacitor with different capacitance.
    But again it will operate out of spec and most likely will fail soon. Just don't do this.
    Last edited by televizora; 12-18-2020, 02:46 PM.
    Useful conversions. I don't "speak" imperial. Please use metric, if you want to address me.
    1km=1000m=100000cm, 1inch=2.54cm, 1mile=1609.344meters, 1ft=30.48cm 1gal(US)=3.785liters, 1lb=453grams, 1oz=28.34grams

    Comment


      #3
      Re: Replacing electrolytic capacitors

      Televizora, thank you very much for the explanation. This answers my question why it did work one one motherboard but not on another. Since faulty caps were mainly around the CPU where there are mosfets and inductors, i.e. power line, we concöuded that their main task is to filter out disturbances, and it shouldbe fine.

      We‘ll order exact caps and we’ll wait.
      Btw we have almost identical problem to this one:
      https://forum.ts.fujitsu.com/forum/v...ic.php?t=43299

      Thanks again
      Last edited by Askic; 12-18-2020, 11:11 PM.

      Comment


        #4
        Re: Replacing electrolytic capacitors

        If they're capacitors for CPU voltage filtering (located usually in rows on the left of the CPU) you should not go for the EXACT brand. Higher quality brands, but ones of the same type (These are filtration and smoothing)

        I found a picture of the motherboard for that PC and found a lot of them are polymer, but sometimes manufacturers put different quality of parts on early/later boards.

        I saw on the other post there was a mention of "blue" caps. Indeed the board I found had Rubycon caps (high quality) where those blue ones are. Another picture I found did indeed have these light blue caps. not sure what brand

        If you can replace those blue ones with the same V/uf of a different brand, that would probably fix your problem. They look big enough to be input VRM caps but they're not usually where the input vrm caps are. It looks like the output caps are polymer, which really is the best

        I found the one with rubycons where those blue ones are here

        https://www.blackmoreit.com/fjspd266...imo-p2511.html

        and one where they're blue here

        https://sierra-indigo.com/fujitsu-si...ard-d2660-a12/

        Consider replacing them with Panasonic FM or Nichicon UHD
        Cap Datasheet Depot: http://www.paullinebarger.net/DS/
        ^If you have datasheets not listed PM me

        Comment


          #5
          Re: Replacing electrolytic capacitors

          Thank you very much for suggestions. I had a few cases in the past with faulty motherboard caps. Interesting almost all had faulty caps next to the CPU. I wonder is this because of the heat dissipated by the CPU?

          Comment


            #6
            Re: Replacing electrolytic capacitors

            Originally posted by Askic View Post
            Since faulty caps were mainly around the CPU where there are mosfets and inductors, i.e. power line, we concöuded that their main task is to filter out disturbances, and it shouldbe fine.
            The capacitors around the CPU I consider somewhat critical. Their capacitance is usually is highest that you can find on the motherboard, because they have to smooth the voltage, stepped down by the VRM-s. You don't want to fry the CPU or make the system unstable. This means that you should not use crappy capacitors there. And the values should be exact.
            Originally posted by Askic View Post
            I wonder is this because of the heat dissipated by the CPU?
            Yes. And because there was a problem with certain series, which is why this site exists in a first place. Some are prone to the "capacitor plague."
            These days, I use Sancon. Quality is not terrible, price is not bad, availability is high here.
            The problem is the most severe with Pentiun IV and some AMD CPU based boards. Not only this is the exact time, when the capacitor plague was raging, but the CPU-s themselves can be used as space heaters.
            According to my experience, the worst caps you can use, but still aren't a complete trash are OST, JWCO and Secon.
            Better than them are Fujicon and Sancon. The best are Nichicon, Rubycon and Hitano.
            Suscon, GLuxon and Capxon should be avoided by all means. For my first recap about 12 years ago, I used Suscon(because one of the electronic components shops here sold them and I thought that it would not make so much of a difference)
            With the old bulged caps on the board, the board worked somehow, but there were frequent blue screens.
            With the Su'scon the board refused to work at all. This is how bad they are. From these worst 3 brads,
            GLuxon is the best, but you must avoid all three by all means.
            -----
            I have to say that for critical applications, ESR matters too. If the ESR becomes high, even if the capacitance is fine,
            it's like putting a resistor in series with the capacitor. The voltage drop across it screws up the whole circuit.
            Most of the time the difference between normal and low ESR is much smaller than between defective capacitor and new capacitor.
            This means that if the circuit was working with bad low ESR capacitor, most likely it will work fine with normal, non Low ESR, but new capacitor.
            Although I advice against this kind of swaps and proper types should be used. In any case, new good capacitor is better than defective.
            Last edited by televizora; 12-19-2020, 08:30 AM.
            Useful conversions. I don't "speak" imperial. Please use metric, if you want to address me.
            1km=1000m=100000cm, 1inch=2.54cm, 1mile=1609.344meters, 1ft=30.48cm 1gal(US)=3.785liters, 1lb=453grams, 1oz=28.34grams

            Comment


              #7
              Re: Replacing electrolytic capacitors

              Televizora, thank you very much. Do you buy these brand name caps from ebay? Csn you make a recommendation?

              Comment


                #8
                Re: Replacing electrolytic capacitors

                Originally posted by Askic View Post
                Televizora, thank you very much. Do you buy these brand name caps from ebay? Csn you make a recommendation?
                I don't buy anything from ebay. I don't like buying stuff from abroad.
                Use what is available in my country. And if you do buy from abroad, better use TME or Farnell.
                Too many scams in Ebay. If you buy from TME or Farnell, Fujicon are not that bad.|
                You can go for Panasonic, Nichicon, Hitano or Rubycon if you can afford them.
                Last edited by televizora; 12-19-2020, 12:13 PM.
                Useful conversions. I don't "speak" imperial. Please use metric, if you want to address me.
                1km=1000m=100000cm, 1inch=2.54cm, 1mile=1609.344meters, 1ft=30.48cm 1gal(US)=3.785liters, 1lb=453grams, 1oz=28.34grams

                Comment


                  #9
                  Re: Replacing electrolytic capacitors

                  Aluminum electrolytic capacitors have always been a common choice for power supplies. However, they have a limited life span and are susceptible to extreme conditions at high and low temperatures.
                  What is a flexible printed circuit board?

                  Comment


                    #10
                    Re: Replacing electrolytic capacitors

                    Why most of them are rated -40c to 105c
                    Cap Datasheet Depot: http://www.paullinebarger.net/DS/
                    ^If you have datasheets not listed PM me

                    Comment


                      #11
                      Re: Replacing electrolytic capacitors

                      Originally posted by Uranium-235 View Post
                      Why most of them are rated -40c to 105c
                      I've seen many graphs that show how operating life span is getting short when ambient temperature is increased. If I'm not mistaken, constantly working around 40°C or above can significantly reduced lifespan of electrolytic capacitor.

                      Comment


                        #12
                        Re: Replacing electrolytic capacitors

                        Originally posted by Uranium-235 View Post
                        Why most of them are rated -40c to 105c
                        But their life is also rated, depending on the ambient temperature.
                        Originally posted by Lusifa View Post
                        Aluminum electrolytic capacitors have always been a common choice for power supplies. However, they have a limited life span and are susceptible to extreme conditions at high and low temperatures.
                        You don't have much choice, if you want high capacities. Every high capacity capacitor is electrolytic.
                        And from these, AL electrolytic are the cheapest.
                        Last edited by televizora; 12-21-2020, 10:59 AM.
                        Useful conversions. I don't "speak" imperial. Please use metric, if you want to address me.
                        1km=1000m=100000cm, 1inch=2.54cm, 1mile=1609.344meters, 1ft=30.48cm 1gal(US)=3.785liters, 1lb=453grams, 1oz=28.34grams

                        Comment


                          #13
                          Re: Replacing electrolytic capacitors

                          Originally posted by televizora View Post
                          Not only you should avoid, but you should never deviate from the capacitance of the original capacitor. If it is 1500, then you should replace it with 1500only.
                          I do agree with the notion that same capacitance should be used when possible... but it's not always necessary, and in fact, you can deviate quite far and quite often. To do that, however, you just need to find what the capacitor is used for in the circuit. Since we are discussing electrolytic caps here, 99.99% of the time, these will be used for filtering (voltage smoothing) of some sort. Thus, as long as you know the functional voltage the cap is expected to have (i.e. 3.3V, 5V, 12V, CPU V_core, V_DIMM, and etc.) and what kind of circuit it is connected to (buck vs. linear), then you can sometimes deviate quite quite far from the original capacitance and voltage rating.

                          Another thing to consider: don't always assume a manufacturers used a certain capacity, because that is what the circuit called for. Sometimes manufacturers will try to cut costs on their product, and they can do (and have done) that by cutting operational margins on a circuit. With capacitors, that often means using less and/or cheaper capacitors - sometimes just good enough to make sure the circuit is able to operate and not crash. In such cases, the circuit might as well have been designed to use, for example, 6x 3300 uF caps, butthe manufacturer used only 4x 3300 uF caps, because they did it to cut costs. In such case, increasing the capacitance back to 6x 3300 uF caps will NOT hurt the circuit. In fact, quite the opposite - it can make the circuit perform better or have better noise tolerance. So don't assume that just because the manufacturer used capacitance X that you have to use the same exact capacitance.

                          Furthermore, electrolytic capacitors have a +/-20% tolerance on capacitance. This mean, for example, that a 1800 uF capacitor (common value used on the CPU VRM of a few older motherboards) can measure as low as 1440 uF or as high as 2160 uF and still be considered "in-spec" for the circuit to operate properly. Thus, if you actually used a 1500 uF or a 2200 uF capacitor in that circuit that measures very close to nominal stated value, then the circuit really won't care.

                          Originally posted by televizora View Post
                          3. As a part of signal generator or timing circuits.
                          4. Reactive ballasts
                          I have not seen these two functions implemented and depending on electrolytic capacitors on any motherboard made in the last 15 years. So I don't think you need to worry about timing / signal generation circuits when it comes to replacing electrolytic capacitors. The only exception is on old (15-20 years) motherboards that use a KA7500 or DBL/TL494 PWM chip for CPU V_core generation. It was something only cheap motherboards like ESC used, so you're unlikely to run into this type of CPU VRM. But for these, the only thing to know is just not to mess with the value of the small caps around the KA7500 / TL494 PWM controller, as one or two can be used for timing or compensation. But really, this is a super-niche case I'm demonstrating here, and chances are you won't see one - much less with electrolytic caps around the KA7500 chip. IIRC, a few old motherboards that I have with that chip used ceramic caps for those functions... so again, probably not something that should be on the list to worry about.

                          Originally posted by televizora View Post
                          2. Capacitance always stays the same! Otherwise the circuit may perform out of it specs. The results can be unpredictable.
                          Exactly.
                          May perform out of specs is well-said.
                          Realistically-speaking, though, the chances are small.

                          Going higher in capacity can almost never hurt the circuit, as long as you do it within reasonable amount - like no more than 2x the capacity. With buck-type VRMs, higher capacitance actually makes regulation easier for the circuit. Only for linear-type VRMs you may run into a problem if you use much higher capacitance with much lower ESR caps, as the linear regulator may then go into short-circuit / overload protection at startup if it "thinks" the low-ESR of the caps are a short-circuit.
                          Meanwhile, going lower in capacitance is sometimes acceptable - typically only in buck-type VRMs that operate at a relatively high frequency. For example, when poly-modding, the "rule of thumb" was that half-capacitance with solid polymer caps is often OK to do. However, on older motherboards like ECS K7S5A and a few others that have a relatively low-switching frequency on the CPU VRM, going with lower capacitance will tend to increase the ripple on the CPU V_core output a little. And while the circuit will still work in most cases, you may find that OCing doesn't yield as good results anymore, due to higher ripple.

                          With that said, and to keep things a little more simple, my experience has been that you can deviate anywhere from 30% lower in capacitance to 100% higher capacitance (that is, 2x the capacitance) without any adverse effects on the circuit. Anything beyond should be considered an experiment.

                          Originally posted by televizora View Post
                          3. Never replace the capacitor with another with lower operating voltage. Will guaranteedly blow up(EXPLODE) or fail,
                          even if not right at the moment, in extremely short time.
                          That's a good general advice.
                          But is not necessarily true.
                          It all depends on what the capacitor is connected to in the circuit. If it's the 3.3V rail and the manufacturer used a 10V cap, you surely can use a 6.3V cap and it won't blow.

                          Likewise, if the manufacturer used 6.3V caps on the CPU VRM, but the CPU V_core is less than 2V (i.e. anything Pentium 3 or newer, really), you can easily use 2.5V or 4V polymer caps.

                          Originally posted by televizora View Post
                          I don't buy anything from ebay.
                          Well, when it come to electrolytic capacitors, you're not missing too much. Ebay is pretty hit-or-miss, if you don't know what to look for (or look out for.)
                          I'd say over 90% of the "Japanese" capacitor listings on eBay from China are probably counterfeit capacitors. Very very few may be production overstock. This you may get a hint of by looking at what the seller has for sale. If they have all kinds of cheap Chinese junk, the caps are probably counterfeit. However, if the seller has only industrial components, some of which used and/or NOS, then the caps may actually be genuine.

                          For example, the listing by this seller is actually for genuine United Chemicon KZE caps:
                          https://www.ebay.com/itm/680uF-16V-N...5/401011429492
                          ... as are all of the electrolytic caps in his/her store.
                          Someone actually particularly asked about the caps in the above link in this thread:
                          https://www.badcaps.net/forum/showthread.php?t=89641
                          ... and it was found they are genuine.

                          So final word is: genuine capacitors DO exist on eBay. But so do counterfeits. Thus, you have to be careful what you buy.
                          Last edited by momaka; 12-26-2020, 10:50 PM.

                          Comment


                            #14
                            Re: Replacing electrolytic capacitors

                            Originally posted by momaka View Post
                            I do agree with the notion that same capacitance should be used when possible... but it's not always necessary, and in fact, you can deviate quite far and quite often. To do that, however, you just need to find what the capacitor is used for in the circuit. Since we are discussing electrolytic caps here, 99.99% of the time, these will be used for filtering (voltage smoothing) of some sort.
                            My advises are "general advises for newbies". While you are not wrong, if you address beginners, it is better not to complicate things too much. Also, the general rule says that 10V cap on place of 6.3 will work longer than 6.3V on place of 10V cap. Exactly because manufacturers try to cut down the costs, they will use the lowest voltage capacitor possible, so it will work for some time, not accounting many different factors and the fact that it will work on the edge of it's specs. Also, I never said that I was only talking about computers.
                            For example, cheaper LED light bulbs use capacitor as ballast. Instead of using resistor, when you have AC you may use capacitor and rely on it reactance - Xc. If you replace it with capacitor with different capacitance, you also change the Xc and throw the whole poorly designed circuit in havoc. The same is if the capacitor is part of a pass filter. The same is if the capacitor is timing capacitor and the circuit relies on the Time constant to function properly. For example, PSU PWM TL494 relies on external timing capacitor and resistor. The maximum value of the capacitor is up to 10uF. Because 10uF is too high for other types of capacitors, electrolytic may be used.
                            What will happen if you replace 2.2uF with 4.7uF? You alter the time constant. Doing this, you change the switching frequency of the PWM IC. This means that you change the duty cycle of the MOSFETS. The behaviour of the whole circuit changes. What will happen next depends on how well the circuit is built.
                            Capacitors are not used only to smooth after rectifying. 2 typical applications like AC ballast and Timing capacitor, require exact capacitors. Not similar, not with close to original value, but exact value. The same is when capacitor is used as pass filter.
                            If you use capacitor only for smoothing the ripple, then it is not that critical what type of capacitor you will use. In other cases the difference between the circuit blowing up or misbehaving and continuing to work as specified is having the exact capacitor.
                            Last edited by televizora; 12-27-2020, 12:31 PM.
                            Useful conversions. I don't "speak" imperial. Please use metric, if you want to address me.
                            1km=1000m=100000cm, 1inch=2.54cm, 1mile=1609.344meters, 1ft=30.48cm 1gal(US)=3.785liters, 1lb=453grams, 1oz=28.34grams

                            Comment


                              #15
                              Re: Replacing electrolytic capacitors

                              @Askic
                              just out of interest: what motherboard are we talking about?

                              A lot of my old Socket 939 boards start to receive bad caps... really annoys me.

                              Comment

                              Working...
                              X