Announcement

Collapse
No announcement yet.

ATX PSU recap and upgrade

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

    ATX PSU recap and upgrade

    Hello guys. I would like to replace the secondary capacitors of an ATX PSU. It currently mounts cap-coil-cap filters for 3.3/5 and 12v. Teapo 1000uF. I'd like to fit in some better-performing Japanese capacitors. Do you advise me to go up with microfarads (I was thinking about 3300uf) or is that a bad idea? A friend explained this to me: "Be careful because the coils on the secondary should be sized to use the expected capacity condestors, it is not said that by increasing the ripple it is reduced, it could create noise at frequencies that with the larger capacitors are no longer filtered correctly." Thank you in advance

    #2
    Re: ATX PSU recap and upgrade

    keep the capacitance and esr similar,
    your friend is right and the caps and coils are matched to create filters.

    Comment


      #3
      Re: ATX PSU recap and upgrade

      generally, capacitors are made with a ±20% tolerance from the factory. so if the original is 1000uF, u could still replace it with a 1200uF without going out of tolerance or out of spec of the design of the original and still stay within spec.

      the other practicality issues with bumping up the capacitance too much is that higher capacitance caps will undoubtedly have a bigger can size. they may be wider (fatter) or taller. if that is the case, they may no longer fit on the board if there isnt enough room to accomodate the larger size caps.

      the next issue is that larger sized caps will also have lower esr from their bigger can size and too low esr will also cause unwanted side effects to the function of the circuit. so these are the practicality issues that arise from upsizing the capacitance of caps excessively and are not recommended for practicality reasons.

      Comment


        #4
        Re: ATX PSU recap and upgrade

        thank you for the answers. So I guess I'm going to use 1200uF. To improve the outbound filter add a third capacitor ? So cap-coil-cap-cap? Is it madness?

        Comment


          #5
          Re: ATX PSU recap and upgrade

          it will look like one huge output cap - pointless.

          Comment


            #6
            Re: ATX PSU recap and upgrade

            yea maybe i should have been more clear with my reply. not only the individual capacitance has to be within spec but the total overall bulk capacitance also has to be matched. if the filter has two 1000uF caps, this means the total bulk capacitance is 2000uF. if u add a third 1000uF cap to the filter, the total bulk capacitance would then be 3000uF which is out of spec of the original bulk capacitance of 2000uF.

            Comment


              #7
              Re: ATX PSU recap and upgrade

              Originally posted by stj View Post
              keep the capacitance and esr similar,
              your friend is right and the caps and coils are matched to create filters.
              Assuming it's a properly designed and built PSU, this is probably true.

              However, without seeing the PSU, I can't say if the O/P should use the same/similar capacity caps or not.

              Remember, manufacturers will sometimes cut corners to cut costs (or rather, make larger profits) and use smaller caps than what the PSU was designed for originally. Generally, that's the case with cheap "gutless wonder" PSUs. For those, I recommend not to follow what's in the PSU and instead use *at least* 2x 2200 uF caps on each major rail, with the 12V rail probably sufficing to have just 2x 1000-1500 uF caps if the PSU is rated for 300 Watts or less.

              Of course, if the PSU is built properly, then you should try to stick to same or similar capacitance. In general, though, going up to 2x the capacitance will not hurt anything. There are very few PSUs that I've seen "freak out" with larger capacitance, and they were both from the same maker - Sirtec. So except for those, larger capacitance actually helps bring the filtering frequency lower and makes the circuit even less likely to oscillate. In most cases, increasing the capacitance within reason will have positive effects.

              I don't know about going to 3300 uF caps if the originals were 1000 uF only. But going 1500 uF or even 2200 uF will likely be OK (if not better.)

              ESR is where you want to stay more similar to what the original circuit had. But even there, I find there are often exceptions that can be made. For example, PSUs with a linear 3.3V -generated rail won't really care that much if you use normal low-ESR caps mean for PSUs or GP 105C caps not necessarily suitable for PSUs. In either case, the ripple on a linear circuit is more or less strictly based on capacitance and not ESR. Likewise, a PSU with a mag-amp -generated 3.3V rail will typically be able to take ultra-low ESR caps (i.e. much lower ESR than the originals) without problems. The 5VSB rail also often falls in that category (though using ultra-low ESR caps on the 5VSB can sometimes make it produce audible "whining"... which may be annoying to the user.)

              All in all, PSU output caps are not an "exact science" of what you can use. You can also lower the capacitance too in many cases, if needed. But then the PSU wouldn't be able to output the same amount of power on that rail without noise becoming too high. In other words, you can increase or decrease output capacitance based on how much load you expect the PSU to see or how much output ripple you're willing to accept as "maximum". That's how manufacturers sometimes build PSUs on the same PCB, but one with more/bigger output capacitors than the other, based on how they want to rate the PSU in terms of power output, noise filtering, and cost.

              Comment

              Working...
              X