Options for power supply?

Discussion in 'General Electronics Chat' started by Nicholas, Jan 23, 2013.

  1. Nicholas

    Thread Starter AAC Fanatic!

    Mar 24, 2005
    Hi all.

    I'm trying to figure out the best way to get power. Let's say I wanted
    any one voltage between 30V and 50V and around 20amps or more. I have a
    switched mode PSU that has 24V and a max of 15A. I find it difficult to
    find one with more than the 24V and 15A for a reasonable price:)

    Any advice how I can do this? maybe a transformer and bridge rectifier
    is the only way to go?


    (By the way, I'm in 220V - land)
  2. crutschow


    Mar 14, 2008
    If you need a variable voltage and not highly regulated, then a Variac and a bridge rectifier/filter may be your cheapest option.
  3. Nicholas

    Thread Starter AAC Fanatic!

    Mar 24, 2005

    I don't need a variable voltage, just something between 30V to 50V, like 40V just to
    name a number.

    So, I would need a transformer to get 220VAC to 40VAC, then a bridge rect. and
    a filter cap(value?)...would anything else be needed?

    Yes, as cheap as possible :)
  4. gootee

    Senior Member

    Apr 24, 2007
    You don't have quite enough requirements specified. To calculate the approximate minimum required reservoir (smoothing) capacitance, you would need to know the maximum load current and the maximum tolerable ripple voltage (peak-to-peak). You would also want to account for the minimum mains voltage and the transformer regulation.

    You can use an approximate form of the ideal capacitor equation to estimate the capacitance needed for a given load current and ripple requirement:

    i = C dv/dt can be written approximately as

    i = C Δv/Δt

    and re-arranged to

    C = i Δt / Δv

    where Δt is the time between charging pulses, which is 1 / (2 fmains)
    and you pick Δv, the largest tolerable p-p ripple voltage.

    Note, too, that it's often (always?) better to use multiple smaller-value caps in parallel, instead of one bigger one. But even though the paralleled caps can theoretically give much lower inductance and ESR, in practice the benefits might mostly be wasted unless they were mounted in an array, on a thin two-sided PCB, with one side ground and the other side power. (No etching required, nor desired. You would drill one small hole per cap and remove a small ring of copper from around each hole edge, on the top/component side of the pcb. The top side leads would just be snipped to a reasonable length, bent over flat against the copper, and soldered, and the bottom leads would be snipped, bent, and soldered similarly, after being inserted through the holes.) An array of 100x 1000uF caps mountd that way, in a 10x10 array, could give a total inductance of less than 1 nH!

    You might also need a REGULATED voltage. There are many discrete-component regulator circuits. And there are three-terminal regulators. Some common models (e.g. LM338, LT1084, LD1084) can handle up to 5 Amps, and some even up to 7.5 or 10 Amps. But in your case, to use one of those, you would need an external pass transistor (or maybe a bank of them) that could handle the current. Just remember that a regulator has a dropout voltage spec, for the difference between its input and output voltages. So you would need to be able to guarantee that the ripple vooltage minima could never drop down far-enough to violate the ripple voltage spec, or the regulator output could get very ugly.
  5. Nicholas

    Thread Starter AAC Fanatic!

    Mar 24, 2005

    Thanks! It seems to be a little more complex :) I'll have to look more into what I need
    I see!

    Thanks for the thorough post!
  6. Jaguarjoe

    Active Member

    Apr 7, 2010
    National Semiconductor (now TI) has data sheets for it's 3 terminal fixed and adjustable regulators that have many applications for them. One which is a high current regulator with multiple pass transistors.
    Be sure all the pass xistor currents are equal or the result could be less than anticipated due to thermal runaway.
    If you could do it with MOSFETS you would not have that problem.
  7. #12


    Nov 30, 2010
    There are many people here that can design an analog power supply before morning coffee break. Get specific about your needs and the answers you get will be more useful.
  8. Jaguarjoe

    Active Member

    Apr 7, 2010
  9. bountyhunter

    Well-Known Member

    Sep 7, 2009
    I just have to ask:

    What do you consider a reasonable price for the 800W power supply you are envisioning in your description?

    Talk about a small world..... ebay has a used Power Ten 40V/25A power supply I designed back when I worked there in about 1988. I designed every piece in that sucker including the PC boards, heat sinks, chassis, front panel and even the silkscreening for the front panel.

    That said, it probably works anyway. $300 for 1000W lab supply with digital meters is pretty cheap.

    Last edited: Jan 26, 2013
  10. vrainom

    Active Member

    Sep 8, 2011
    I take my proverbial hat off to you, sir. What is the topology of such piece of art?