No announcement yet.

experience using a switching pre-regulator

  • Filter
  • Time
  • Show
Clear All
new posts

  • experience using a switching pre-regulator

    Wondering of any experience using a switching pre-regulator before a linear regulator and how much capacitance can be between. Basically, what can the switching supply handle and remain stable or not blow up. Considering a laptop supply to feed a 5V regulator for a DAC. Usually 2 or 3A switching so I suspect a few hundred mic woudl be safe but in some cases, switchers are not fond of large C loads.

    I have a box of old switching supplies, and that is far cheaper than a new transformer.

  • #2
    A buck regulator before an ldo is a common approach to reduce thermal issues with the ldo. However, any ldo used should have good psrr at the switching frequency. Check out the LT3045 which has psrr graphed up to 10MHz.


    • #3
      Yea, Built a little variable supply for testing my two DACs One needed 18V, one 5. @18, heat sink was cold. @ 5, I was concerned with it pushing safe range. Seems I want to give the LT1083 only about 5V to deal with.

      Got some hints on brute force pi filters to clean up switchers and learning about newer regulators. The 10 years I have not been paying attention while restoring a couple of MGs have progressed a lot.


      • #4
        I wouldn't bother with a Pi filter. Just a series R and a shunt C should do it. Something like half an ohm and a couple of hundred uF gets you a corner frequency of under 1kHz. Plus you isolate the switcher from the big cap. The biggest downside is the roughly half watt of dissipation of the resistor.


        • #5
          I've used lots of capacitance (10,000+ uf) on the output of switching power supplies w/o issues. Most all designs implement an overcurrent protection scheme (current limiting, foldback or "hi-cup") to avoid damaging the PS.

          Last edited by Millstonemike; Yesterday, 11:19 AM.


          • #6
            A ferrite bead can sometimes be a good thing after the switcher. It may not offer lots of attenuation at the switching frequency, but it will squash the broadband RF coming out of the switcher.


            • #7
              A pocket full of ferrite beads can do wonders!
              I did a bunch of models last night for a traditional raw supply. I was surprised at how ineffective the choke is for ripple rejection, at least in a size I have. ( 2mH. I have an old choke from a tube chassis and it is 2H) I was surprised in how effective a resistor is in place, epically if you have the raw voltage to lose. In my case, I happen to have enough that a pair of 3300 caps with a 1 Ohm does well over twice what just a 6600 cap does. Really takes out the HF content. If you can waste a bit more voltage 5 Ohms can get your ripple below the noise floor. A third stage seemed to be reaching diminishing returns. Even the model for theLT1083 showed no identifiable ripple. So, going to modify my DAC power supply today and do real life measurement to see if the model taught me anything. I have an entire tray of nice Dayton 10W resistors from 1 to 12 ohms!

              I will then go back to the SMPS feedingeither a filter or a linear reg. The same filter for a switcher might do just as well on a linear output bringing th enoise down if the ripple was dealt with as above. Choices. Fun


              • #8
                Yes, an RC is pretty foolproof if you can afford the voltage drop. Another trick is a series schottky to a filter cap.


                • #9
                  Schottky. OK, that is want I did wrong. I was modeling with a generic diode and found it to not be effective as effective as a one Ohm resister.

                  Actually, the voltage drop can be a good thing as we sometimes need to take a bit of load off the regulator to keep it cool. 4 to 5V seems to be about the optimum with these new LDO chips. Buying off the shelf, we don't have as many choices in transformers. I wonder, not knowing much about switchers, how easy it is to change their output voltage a little. Can I make a 12V into an 18? Off to GOOGLE.

                  You're a pretty smart guy even if you do like horns

                  Gad, GOOGLE can be a friend. Zener and verify the caps.


                  • #10
                    Or the voltage divider. Guess it depends. Anyway, clearly a decent hack to look at.