Thank you to the guys at HEGE supporting Badcaps [ HEGE ] [ HEGE DEX Chart ]

Announcement

Collapse
No announcement yet.

From classic electrolytic to solid state capacitors

Collapse
X
Collapse
 
  • Page of 1
  • Filter
  • Time
  • Show
Clear All
new posts
Previous template Next
    #1

    From classic electrolytic to solid state capacitors

    Hi everybody!

    I was wondering if is possible to replace the common electrolytic capacitors with solid state caps just like on the newer motherboards.

    So, as stated in the title, is it possible?

    Is possible to replace common electrolytic caps with better solid state ones with the same values?

    Thank you
    Tags: None
    #2
    Re: From classic electrolytic to solid state capacitors

    Yes but generally the polys from what I've seen are only available in capacitances up to 1500uf.

    I just got some 6.3v 1000uf polys to replace lytics. They were Nichicon L8 series. If you google Nichicon L8 you can get to their website and see the other series and values available.

    You can use polyis having lower voltage and capacitance values in the VRM
    circuit that supplies the CPU. Based on the fact that most CPUs are between 1 and 2 volts it will work in this circuit.

    Topcat and others here have done this mod and will comment, I am sure.

    Outside the VRM, I suspect you want to stay with lytics unless there is an exact match in the poly range.

    Comment

      #3
      Re: From classic electrolytic to solid state capacitors

      By "solid state" I assume you mean Solid Polymer.
      That's called a "poly mod". - Very popular topic here.

      Yes it can be done although it not as straight forward as people would like it to be.
      Different circuits on the board have different needs and limitations.

      If you wanna walk through one give us a make/model and upload some pics of the beast in question.

      .
      Mann-Made Global Warming.
      - We should be more concerned about the Intellectual Climate.
      -
      Be who you are and say what you feel, because those who mind don't matter and those who matter don't mind.
      - Dr Seuss
      -
      You can teach a man to fish and feed him for life, but if he can't handle sushi you must also teach him to cook.
      -

      Comment

        #4
        Re: From classic electrolytic to solid state capacitors

        The part I don't understand why the MSI MB I worked on recently had a mix of poly and electrolyte caps near CPU (VRM?). If poly caps are so great, why MSI sneaked in couple of bad KZG caps? Planned obsolescence?

        Comment

          #5
          Re: From classic electrolytic to solid state capacitors

          >> Planned obsolescence? <<
          No. It isn't a plot. - MSI is simply too stupid to know KZG go bad.

          Mixed poly and lytic in VRM was common for a while.
          Still might be used by some manufacturers.

          If you look at VRM Vcore on all poly boards the total uF of all the Vcore caps is up over 5000uF.
          I think sometimes they mix poly and lytic in Vcore to reduce cost and other times because there aren't enough cap positions to use all polymer and still have total uF in Vcore over 5000uF.
          .
          Mann-Made Global Warming.
          - We should be more concerned about the Intellectual Climate.
          -
          Be who you are and say what you feel, because those who mind don't matter and those who matter don't mind.
          - Dr Seuss
          -
          You can teach a man to fish and feed him for life, but if he can't handle sushi you must also teach him to cook.
          -

          Comment

            #6
            Re: From classic electrolytic to solid state capacitors

            Interesting. If lytic and poly caps are in parallel and lytic are mostly to boost uF then lytic dont have to have that super low ESR, right?

            Comment

              #7
              Re: From classic electrolytic to solid state capacitors

              The caps do two things.
              The ESR passes Ripple to ground.
              The uF keeps the DC voltage steady during transients. [Like CPU going from 20%-100% utilization.]

              When you mix lytic and poly you need to look at a few things.
              - That you have enough uF for transients. From what I can tell looking at unmodified boards with polymer they shoot for over 5000uF in Vcore and usually it's a bit higher. [With 10 poly in Vcore they often use 560uF poly. With 7 or 8 Vcore caps I often see 820uF most of the time.
              - And, because poly have way low ESR, the ripple is going to tend favor passing through the poly caps [least path of resistance]. So you want to make sure there are enough poly caps in the mix to handle the full Ripple expected for -all- the Vcore caps. [That's my own 'rule' as it were. Makes sense to me anyway.]

              I'll open a new window and walk through one.
              .
              Mann-Made Global Warming.
              - We should be more concerned about the Intellectual Climate.
              -
              Be who you are and say what you feel, because those who mind don't matter and those who matter don't mind.
              - Dr Seuss
              -
              You can teach a man to fish and feed him for life, but if he can't handle sushi you must also teach him to cook.
              -

              Comment

                #8
                Re: From classic electrolytic to solid state capacitors

                Safe Vcore Poly-Mod

                Thumb? Rules:
                1: Use enough caps to get total Vcore uF over about 5000uF. [Per CPU if you are dinking with a dual socket server board.]
                2: Calculate total Ripple for all the existing caps and use enough Polys to handle that entire amount of ripple. I basically assume ALL the Ripple goes -only- through the Polys to ensure I use enough Polys to handle 100% of the Ripple current the board designer expected.
                3: Make sure you keep total ESR the same or less than original. - Don't assume the poly have less ESR than the Lytics just because they are Poly. With so many different kinds of both Poly and Lytics on the market now that isn't always true anymore. Some Poly have higher ESR than some Lytics.

                ~~
                We are going to assume the 'stock' Vcore has 8x MCZ 3300uF.
                [one MCZ] - ESR = .009 ohm .. Ripple = 3230 mA
                [8 in parallel] - ESR = .001125 ohm .. Ripple = 25840 mA

                So we want:
                - Total uF = 5000uF or more [preferably over 6000uF] - Based on factory all Poly boards.
                - Total Ripple = 25840 mA or more [in just the Polys] - Based on stock circuit.
                - Total ESR = .001125 ohm or less [ideally less] - Based on stock circuit.

                Figuring uF
                5000uF/8 => 625uF [and 5000uF is a minimum] so 680uF Polys would be fine.
                [Though 820uF Poly would be a good choice too because they are common and easy to get. Probably cheaper too.]
                I will follow the math above and use 680uF for the example.

                Going the -all- Poly route is easy to figure because you just have to look at the specs for a single cap and compare.
                -
                A Poly that would NOT work: Chemicon PS series 680uF 2.5v
                [one] - ESR = .010 ohm .. Ripple = 5230 mA
                Plenty room on the Ripple but the ESR is higher.
                [Lowering total uF drastically -and- raising ESR is a bad combination.]
                A Poly with the same ESR as the original Lytic is -probably- okay but I like more 'room' so I try to avoid it and use lower ESR Polys.
                - There are lots of 680/820uF Polys that have less than .009 ohm ESR so all you have to do is pick one.

                Examples of mixing caps.
                - Lets say we have some Chemicon PSA 820uF 4v left over from another project but we don't have 8.
                [one] - ESR = .006 ohm .. Ripple = 5860 mA
                -
                It takes 5 to achieve [equal or more] the 25840 mA minimum ripple we calculated form the 'stock' circuit. [5x of these handle to 29300 mA.]
                -
                5x 820uF = 4100uF ... need at -least- 900 more uF in Lytics [easy].
                -
                For ESR 5x .006 ohm in parallel => 0.0012 ohm and the target is less than .001125 ohm [won't take much].
                -
                A single 1000uF Lytic would technically be enough but this is a place where more [within reason] is better.
                -
                Lets look with a single of the 3300uF MCZ with 5 of the 820uF PSA.
                - Total uF = 7400 uF
                - Total Ripple = 32530 mA
                - Total ESR = 0.0010588 ohm
                - No problem..
                -
                -
                Lets look with a pair of 1500uF ZL with 5 of the 820uF PSA.
                [one Rubycon ZL 1500uF 6.3v] - ESR = .023 ohm . Ripple = 1820 mA
                - Total uF = 7100 uF
                - Total Ripple = 32940 mA
                - Total ESR = .0010866 ohm
                - No problem..
                [I picked ZL because without Polys they usually aren't good enough for VRM use [since P3 anyway] but in combination with Polys they can work out fine.]
                -
                -
                What -I- would probably use in the mixing example here is 2x 1500uF to 2200uF from some Series comparable to MBZ or MCZ. That would put everything well beyond the minimums.
                .
                I have seen people stick a single Poly in an all Lytic VRM. That can work but you you REALLY need to do the math. ESR on caps in parallel behaves like resistors in parallel. If that single Poly has a way low ESR compared to the other caps it will tend to take all the Ripple current away from the Lytics and the Ripple might exceed the Polys rating.
                [Think like having 8 big ohms resistors in parallel and then replacing one with a small ohms resistor. Most of the current tries to go through the small ohms resistor so if it can't handle enough watts it will fry. - Ripple to a cap is as Watts is to a resistor.]
                .
                .
                And a note about mounting.
                It's small but the trace run resistance does [or theoretically could] cause slightly more 'noise' [Ripple] to be at one end of all the paralleled caps in Vcore than at the other. For this reason it is best to put a Poly at the outer ends of the VRM trace run and as much as practical spread the Lytics evenly in between those two end caps. Other than the end Polys it doesn't have to be exact, just don't group all the Polys or Lytics together. Spread them out some.

                Something like:
                [P][.L.][P][P][P][P][.L.][P]

                [P][P][.L.][P][P][.L.][P][P]

                [P][P][.L.][P][P][P][P][.L.][P][P]

                [P][.L.][P][P][.L.][P][P][.L.][P]
                .
                .
                Last edited by PCBONEZ; 12-10-2010, 09:00 AM.
                Mann-Made Global Warming.
                - We should be more concerned about the Intellectual Climate.
                -
                Be who you are and say what you feel, because those who mind don't matter and those who matter don't mind.
                - Dr Seuss
                -
                You can teach a man to fish and feed him for life, but if he can't handle sushi you must also teach him to cook.
                -

                Comment

                  #9
                  Re: From classic electrolytic to solid state capacitors

                  Part of that is the higher uF caps store more 'juice' in reserve for transients.
                  IF you have a big transient you don't want all the current flowing through one small area of the traces.
                  .
                  That's also why I'd use two 1500/1800uF instead of one 3300uF. [In the example.]
                  Mann-Made Global Warming.
                  - We should be more concerned about the Intellectual Climate.
                  -
                  Be who you are and say what you feel, because those who mind don't matter and those who matter don't mind.
                  - Dr Seuss
                  -
                  You can teach a man to fish and feed him for life, but if he can't handle sushi you must also teach him to cook.
                  -

                  Comment

                  Previous template Next
                  Working...
                  X