Help building split supplies with positive regulators

Discussion in 'General Electronics Chat' started by spicytofu, Jul 17, 2009.

  1. spicytofu

    Thread Starter New Member

    Jul 17, 2009
    1
    0
    Im planning to add this to my existing supply. I know I can use NEG regs and it makes this alot easier, but I dont have any in hand. The supplies are isolated and connected through two bridges. I know the unregulated supply works since it is powering a mono class A amp; I used a DC supply symbol in place of the PS rails. Judging from this circuit, U2 looks to work fine, supplying 12V regulated output. (or -12V if you like to be technical) But I am not so sure about U1. Plus, since the input for U1 has no reference, Im not sure what voltage it will actually "see" from the ground. And if I assume that it will see a total of 48-12=38, thats awfully close to the 40V limit. Also, the GND reference is also changed now in comparison of the GND of the overall amp. Im no expert in PS design, any thoughts or opinions? Any way I can make this work using this REGs?

    BTW, I cannot tear the supply to split the bridges to make it work, or else I wouldnt be posting this. My supply rails looks like this:

    +24V
    +
    +
    +
    0V/GND
    -
    -
    -
    -24V
     
  2. Wendy

    Moderator

    Mar 24, 2008
    20,766
    2,536
    Won't work. I made this exact same mistake, then wrote this thread to help other people, you have to have something that can both give and absorb current for it to work.

    Creating a Virtual Power Supply Ground
     
    Last edited: Jul 18, 2009
  3. jsm09a

    New Member

    Jul 8, 2009
    5
    0
    Bill,

    You are probably already aware of this, but just in case ... TI makes a special purpose "rail splitter" that may be useful in circuits of this type (TLE2426).

    Best regards, Scott.
     
  4. StayatHomeElectronics

    Well-Known Member

    Sep 25, 2008
    864
    40
    If you are trying to keep is simple, the negative voltage regulator will be your easiest option. The current going out of the U1 regulator has no way to get back to the battery (or power supply) that is providing the power.
     
  5. millwood

    Guest

    but they work differently. 2426 creates a "virtual ground" off a single rail.

    here, you do have a split rail. you just want to "regulate" both of them with a positive rail regulator. the answer is NOT possible. because you cannot provide a current flow from the positive rail to ground.

    and once you fix that issue, you cannot regulate the negative rail, :).
     
  6. jsm09a

    New Member

    Jul 8, 2009
    5
    0
    You would need a different regulator (e.g. LM7824) to regulate the sum of the batteries to 24 volts, and then the rail splitter could create a virtual ground between them (since it can sink or source to the virtual ground). This would give you regulated and matched supplies on both sides.
     
  7. Wendy

    Moderator

    Mar 24, 2008
    20,766
    2,536
    TLE2426 Data Sheet

    Interesting. I wasn't aware of this device, though it makes perfect sense.

    I was a little disappointed about the 80ma spec limit. Surely they could do better. I would also love to see the schematic.

    The whole point of a virtual ground is it creates what looks a lot like a dual tracking power supply. Done properly, it would be impossible to tell from the outside.

    What the OP wants is a dual tracking power supply, or baring that, two independent regulated power supplies. My first choice is to get a negitive regulator. My second choice is to disconnect the orginal ground, and use a virtual ground.
     
    Last edited: Jul 18, 2009
  8. millwood

    Guest

    in that set-up, a several draw on the positive rail will cause the negative rail to expand, as a way to compensate the overall rail voltage shrinkage. so your positive and negative rail is "interconnected".

    it is nothing more than a emitter follower, or an output stage, or a diamond buffer, or another chip (with very low output impedance). you can use an audio power amp IC to split a rail too.

    end of the day, electronic devices don't care if something is ground, positive rail, or negative rail. all they care about is potential differentials.
     
  9. Wendy

    Moderator

    Mar 24, 2008
    20,766
    2,536
    An emitter follower wouldn't work, you need something that both sinks and sources for it to work properly. Just hanging a voltage out there is not enough, it has to have the characteristics of a power supply for it to work well. Otherwise a simple resistor divider would work (and does, in many cases).

    I covered this in detail in the other thread, even mentioning why emitter followers wouldn't work. If you want to argue a point, try reading the other persons threads. It's what go me so hot previous.
     
  10. millwood

    Guest

    really? what if I do make it work?

    and an emitter follower wouldn't both sink and source current?

    have you seen a typical output stage of an opamp that both sink and source current? what is it usually?

    what are the characteristics of a power supply and what are those of an emitter follower?

    the fact that you said that an emitter follower didn't work only means you were wrong then as you are now. it doesn't mean that an emitter follower doesn't work.
     
  11. Wendy

    Moderator

    Mar 24, 2008
    20,766
    2,536
    I've made my points. Enuf said.
     
  12. millwood

    Guest

    folks in the headphone amp business has for years used class AB buffers to create virtual ground. those buffers are essentially emitter followers (current amps) formed by a pnp/npn pair.

    sijosae's website (http://www.headphoneamp.co.kr/ftp/sijosae/Gallery/) will give you plenty examples of how this is done. scroll down to the power supply section and you can see for yourself. and explore his site more if you are up to it.

    the basic principle is quite simple: a critical characteristic of a power supply is its low output impedance, and its ability to maintain that.

    a class AB emitter follower has very low output impedance and you can easily use it to maintain the virtual ground even driving dynamic (current) load.

    those are things that have done all over the world for years, if not decades.
     
  13. millwood

    Guest

    here is a good summary of various techniques to create "virtual ground".

    http://tangentsoft.net/elec/vgrounds.html

    it has the buffer approach, or opamp approach I mentioned earlier, and then about half way down, it has a schematic using a class AB emitter follower to create a virtual ground - that's the approach that sijosae used.

    that particular circuit can also be improved for thermal stability but it is a good starting point.
     
Loading...