Connect with us

Interchanging of drain and source of MOS transistor

Discussion in 'Electronic Basics' started by shuteng, May 20, 2004.

Scroll to continue with content
  1. shuteng

    shuteng Guest

    Hi all,

    1. I learnt in my varsity that the drain and source of a transistor
    can be interchanged when drawing the layout (IC) of a transistor. Have
    been thinking about it for sometime and I conclude that the drain and
    source of a MOS transistor is equally doped hence, they can be
    interused unlike that of (collector and emittor of BJT). Is my
    hypothesis correct? If so, does it also means that the source and
    drain can be interchanged on schematics?

    2. In microelectronics, if a nMOSFET is used as a pass gate, does it
    matter whether I connect the (driver) signal to the drain or source?
    What are the minimum gate voltages needed to pass the signal from the
    driver to the other devices(assuming 1V Vt)?


    a) _|_ b) _|_
    signal=3V ___| |___ signal=0V ___| |___
    D S D S

    c) _|_ d) _|_
    signal=3V ___| |___ signal=0V ___| |___
    S D S D

    I apologise for my long questions and thanks in advance
    shuteng
     
  2. -------------
    Only if their construction is symmetrical.

    ----------------------------
    Not if the answer to 1) is true.

    -------------------------------
    Go look it up in a device book such as Floyd's "Electronic Devices".
    I'm NOT going to do your reading for you. That's what your computer's
    for.

    -Steve
     
  3. Beau Schwabe

    Beau Schwabe Guest

    For the most part yes, the Source and Drain can be interchanged.
    Where it matters is if you have a single transistor with an even
    number of gate fingers. This creates an ODD number Source
    and Drain ratio, and in some sensitive circuits this parasitic
    difference can affect how the circuit will perform.

    In a pass gate, any parasitics as mentioned above will be seen by the
    signal no matter which way it is connected, so no in this case
    Source/Drain swapping does not matter. ...Unless you have multiple
    signals and matching is a concern. In this case you can still swap
    Source and Drain, however ALL of the other signals should be connected
    to the pass gates in a similar/uniform way.
    This is highly dependent on the process in which you are using and how
    the transistors are built. i.e. Gate Oxide, Salicide, etc.
     

  4. No, not exactly.

    Most people don't emphasize very much where the substrate (bulk) of the
    device is connected. The substrate connection is however very important to
    the MOSFET's operation. In normal MOSFETs used in normal fashion, the
    substrate is normally short circuited to the source contact. This
    connection is really what defines the "source" pin (and consequently the
    remaining pin as the drain pin).

    Although most people talk about a gate-source potential that produces an
    electric field that causes channel inversion, I think it is more accurate to
    claim the gate-bulk potential difference is what produces the electric field
    that causes channel inversion. It just so happens in the marjority of cases
    the source is short circuited to the substrate, so a gate-source potential
    is the same thing as a gate-bulk potential.

    If you look at the MOSFET cross sectional view you notice an NPN structure
    (for an N-channel device). When you short circuit the source to the bulk
    (by making ohmic contacts to some metal which connects the two parts) you
    effectively end up with a PN junction (between source/bulk and drain). This
    PN junction becomes forward biased whenever the source is more positive than
    the drain potential. Therefore, a standard MOSFET used in the normal
    fashion cannot block more than about one diode drop in the reverse
    direction.

    If you take a look at International Rectifier's MOSFET datasheets they
    specifically show a diode symbol as part of the MOSFET symbol to represent
    this PN junction.

    http://www.irf.com/product-info/datasheets/data/irf2804.pdf

    They also characterize the PN junction's performance in the section labelled
    "Diode Characteristics" as well as in figure 7.

    Note, my comments only apply to MOSFETs, not to other FET devices such as
    the JFET. JFETs are very different beasts despite the same pin naming
    convention.

    No, you most definitely cannot interchance the source/drain connections on a
    shematic and expect the circuit to function properly. In one direction the
    MOSFET will block only one diode drop, but in the other the full amount it
    is rated to block...


    I'm not sure I understand quite what you are depicting/asking here so I
    can't answer this...
     
  5. Beau Schwabe

    Beau Schwabe Guest

    For discrete components I would completely agree with you, but if you
    are actually "inside" building the transistor the Source and Drain are
    in fact interchangeable with respect to odd fingered transistors.
    Even fingered transistors are better for abutment when you have
    multiple transistors and you want to share the Sources or Drains, but
    exhibit slight parasitic differences because of the odd number of
    Source to Drain ratio created. Most of the time this difference will
    not matter, but in some circumstances, this will create an unbalanced
    loading of the circuit that can propagate against your favor.
    (snip)
     
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

-