Connect with us

Low drop discrete p-mosfet linear regulator

Discussion in 'Electronic Design' started by Steve Sousa, Mar 18, 2005.

Scroll to continue with content
  1. Steve Sousa

    Steve Sousa Guest

    Hello:

    I need a circuit that can regulate a lithium batery voltage in the range
    3.5~4.2 down to 3.3V, max 400mA.
    I came up with the following circuit:

    fdc638p

    .-->|-.
    VCC d | | s Out
    o-----------o-+^+-o----o--------o----------o
    ||| | |
    === | .-.
    |g | | |
    | | | | R2
    | | '-'
    | | |
    | >| |
    | |------o
    | /| |
    | | .-.
    o------' | |
    | | | R3
    | '-'
    | |
    .-. |
    | | |
    | | R1 |
    '-' |
    | |
    | |
    | |
    o---------------o---------------o----------o
    GND
    (created by AACircuit v1.28.5 beta 02/06/05 www.tech-chat.de)

    Where R2 is chosen to drop 0.7V when the output is 3.6.
    Any comments/sugestions? I think i should use a zener, but couldn't figure
    out where to put it, without adding a lot of parts.
    I would like to use only p-mosfets or bipolar transistors.

    Thanks in advance.
     
  2. Steve Sousa

    Steve Sousa Guest

    Hello:
    The faster one posts the more mistakes:
    There are to errors in my previous post. First, when the circuit is fed 4.2
    volts the mosfet's internal diode will put 3.5V on the output, not good.
    In the paragraph below the circuit, where it says: "output is 3.6." should
    read "output is 3.3"
     
  3. It would help if you were to state the acceptable
    range of output voltages and the temperature
    range over which you require that accuracy.

    The capacitive load that will exist should also
    be given to enable stability analysis and help
    narrow the viable design approaches. For
    the same reason, if transient loads will not
    be absorbed by the load capacitance, they
    should be specified as well.
    The circuit is going to exhibit a TC of about
    13 mV per degree C, within a few mV. If
    the output is to power logic requiring a 5%
    supply, that drift alone would constrain the
    operating temperature to stay within a 12
    degree C band. But the initial tolerance
    is going to be poorer than that, due partly
    to variation in BJT beta and partly to the
    varying threshold voltage of the MOSFET.

    The MOSFET is backwards.
    It would also help if you were to indicate the
    cost constraints applicable to the solution.
    I guess a zener is also a candidate, (although
    low voltage zeners are very soft references.)

    Before designing improvements, it would
    be smart to nail down the requirements
    and separate them from the wishes.

    You may see a number of circuits offered
    here as an "improvement". They may not
    satisfy your actual requirements unless
    they are stated ahead of the time those
    circuits are concocted. But if you have
    stated your requirements, the circuits
    offered will be scrutinized against them,
    at least by some diligent participants.
    You're most welcome.
     
  4. Robert Baer

    Robert Baer Guest

    Why not use an adjustable 3-terminal regulator, like the LM285 to
    control a pass transistor?
    A MOSFET is unacceptable, because one needs about 3V to drive it on;
    a DMOS or JFET would be better choices, and a bipolar might serve better.
     
  5. Fred Bloggs

    Fred Bloggs Guest

    That 400ma seems like a lot for Li battery application. You are wasting
    your time with this circuit- 3.3V is a standard LDO regulator:
    http://www.national.com/pf//LP/LP3872.html
     
  6. Steve Sousa

    Steve Sousa Guest

    I had my eye on lp3871 or lp8345, but wondered if it couldn't be done
    cheaper.
    The batery is 950mAh. 400mA is an estimated very worst case, tipical is
    around 60 to 230, but most of the time the circuit goes to low power sleep,
    less than 6mA.
     
  7. Steve Sousa

    Steve Sousa Guest

    That is something worth trying.
    The treshold for the mosfet i have is -1.5V, I have available lp2985-2.8

    Thanks
     
  8. Fred Bloggs

    Fred Bloggs Guest

    You better focus on that worst-case estimate because the price you will
    pay is having to go with an LDO with relatively large quiescent current.
    By the time you make a full regulator in discrete, you could buy 10 LDOs.
     
  9. Steve Sousa

    Steve Sousa Guest

    3.3+-5%
    -10 to +55
    I have no ideia what the capacitance of the load is, currently there is a
    470uF FK series panasonic on board, the loads are a couple of ATMega, a gps
    receiver (ublox) and a bluethoot module. That capacitor is of that value
    because it's used in other parts of the circuit, it can be changed/removed.
    That's bad.
    I placed it that way to be able to control VGS more easily, but it creates a
    problem when the input is over 4V, so i can't use that way anyhow.
    I was hoping a discrete solution for less than the cost of an lp3871, about
    1.36 Eur.
    I was under the impression that a linear regulator using a p-mosfet would be
    relatively straight-forward, since i din't have a problem getting the
    mosfet, as i use it in other parts of the circuit, i was wrong.

    I have available a lp2985-2.8, maybe i could use that and a pass transistor
    to "fake" a lp3871, so i can at least debug the thing during the weekend?

    Other info:
    The batery is 950mAh. 400mA is an estimated very worst case, tipical is
    around 60 to 230, but most of the time the circuit goes to low power sleep,
    less than 60mA.

    Thank you

    Best Regards.
     
  10. Fred Bloggs wrote...
    LP3872ET-3.3 Wait, there's more!
    You also get free 24-hour samples!
     
  11. Ian

    Ian Guest


    LP2996. Ignore what the datasheet calls it, look at what it does.

    http://www.national.com/pf/LP/LP2996.html#Datasheet

    Regards
    Ian
     
  12. Ian wrote...
    Hmm, Vout = 0.5 Vsupply. "The reference voltage is generated
    from a resistor divider of two internal 50kΩ resistors. This
    guarantees that VTT will track VDDQ / 2 precisely."

    You have a clever way to get around this functionality?
     
  13. Whooh! That should be fun. I think a good
    reference is already indicated. Do you, perhaps,
    already have one in the target system? Or is
    this the first use of one? If so, is there a higher
    supply avaliable?
    If you can guarantee a minimum capacitance,
    that is helpful. An upper bound would be nice
    and permit more latitude in the stabilization of
    the feedback loop. But it is not critical.
    Forward biased Si junctions, acting alone, are
    not much good at being references. With a few
    more parts, you can use a pair of them to make
    a reference good enough to do your job.
    So, it's in frontwards now? I guess I would
    use it however worked best anyway, so my
    comment was in case you might build the thing.
    That seems to make an op-amp solution
    reasonable. Is there any reason you have
    excluded that? Do you have any spares
    or singles that could become doubles?
    Are you sure there are no integrated
    low dropout regulators that do this?
    It will be, relatively, if there is a reference available.
    Otherwise, that will be one of the more fun parts.
    Have you looked into what a reference, PMOSFET,
    and cheap low power rail-to-rail op-amp solution
    would cost? Is that too much, considering your
    expected volume?
    Are you saying you mean to use parts you have
    laying around, or that you already have a 2.8V
    supply developed by that LDO? If the latter,
    then that supply may be a decent reference.
    But it can stay at the 400 mA (worst case) level
    for a time period exceeding the thermal time
    constant of small parts, right?
    Likewise,
     
  14. Ian

    Ian Guest

    Quite right, Win. The application I used this in had an available
    regulated voltage I could use for Vddq, so this was a very cheap,
    simple (low parts count) LDO regulator. For this application,
    the OP would also need to have a reference. That could be a
    say 2.5V fixed part. That plus a couple of resistors sets the
    output to 3.3V.

    Regards
    Ian
     
Ask a Question
Want to reply to this thread or ask your own question?
You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.
Electronics Point Logo
Continue to site
Quote of the day

-