Connect with us

Should supply rails be negative?

Discussion in 'Electronic Basics' started by Lauri Alanko, Sep 1, 2013.

Scroll to continue with content
  1. Lauri Alanko

    Lauri Alanko Guest


    As we know, it is a historical accident that the charge of the
    electron is considered negative, and conventional current flows
    counter to the electrons. For a long time I thought that this wasn't
    of any consequence, since all the math was symmetrical: everything
    would work the same if we inverted the signs.

    But with semiconductors this isn't true. Electrons are more mobile
    than holes, and that is why NPNs are in general a bit better than
    PNPs, and N-channel mosfets are better than P-channels. Ideally, we'd
    always choose the polarity of a circuit so that we'd primarily get to
    use N-type transistors.

    However, there is another asymmetry that can prevent this, and that is
    the distinction between the ground and supply rails. "Ground" is the
    potential level that is connected to the chassis and possibly to
    earth. Sometimes, because of safety or EMI reasons, we want a circuit
    to be tied to ground, and that's why switches have to be high-side.
    But due to convention, high-side (i.e. non-ground-side) is at a
    positive voltage relative to ground, meaning that we either have to
    use P-type transistors or then resort to complex bootstrapping
    gimmicks (boost capacitors).

    So let's summarize:

    * Physics says we should switch on the negative rail

    * Safety says we should switch on the non-ground rail

    * Convention says these two rails are distinct

    To me this seems to imply that if compatibility with conventional
    circuits is not an issue, we'd be better off tying our positive rail
    to ground and calling the negative rail -V, and then doing easy, safe
    switching on the negative rail with N-type transistors. As an
    additional bonus, the flow of electrons in the circuit would finally
    seem natural.

    Do I have this right?

  2. No.

  3. Fred Abse

    Fred Abse Guest

    No. Far from it.

    "Ground" (I prefer to say "zero volts", when discussing circuits) has
    different meanings, depending on whether we are discussing utility
    supplies, or theoretical circuits. Zero volts in a circuit may or may not
    be connected to true earth ground, depending on circumstances.

    Supply rails may be either positive, or negative of circuit reference
    zero, and there can be several rails, some negative, some positive with
    respect to circuit zero. The whole circuit, positive rails, negative
    rails, and "zero", may be floated hundreds, or thousands of volts above
    true ground.

    Some operational amplifier circuits are an example, having both a
    positive, and a negative supply rail, and a "virtual ground", which is at
    (nearly) zero volts, relative to the supply rails. by virtue of gain and

    It's misleading to think of current as electron "flow". Electron
    displacement is a better description. Electrons travel quite slowly in

    My advice to anyone getting interested in electronics is to forget about
    electrons. Think charges, voltages, fields, and currents. Become familiar
    with circuit theory. Learn, and practice, Kirchoff's Laws, Thevenin's, and
    Norton's theorems, mesh, and nodal analysis to the point where they become
    second nature.

    Conventional current is a well-established norm. There's no point in
    re-inventing it to suit some philosophical principle. It works, get used
    to it.
  4. Lauri Alanko

    Lauri Alanko Guest

    [Positive ground would make switching more practical]
    That is most interesting, but not particularly informative. Would you
    care to elaborate where my reasoning went wrong?

  5. Lauri Alanko

    Lauri Alanko Guest

    That may be true, but you didn't address my argument at all, so I'm
    not yet better off.
    Did you even read my message? I didn't have any philosophical
    principle, but two practical points:

    * N-channel mosfets are preferable over P-channel mosfets for physical
    reasons that have nothing to do with conventions

    * A non-conducting (i.e. switched-off) part of a circuit should
    preferably be at the same potential level as the chassis or earth or
    at least some nearby large conductive surface, if such things are

    And from these I concluded:

    * The chassis/earth/surface should preferably be at a higher potential
    level than other rails so that we can easily use N-channel mosfets to
    control the connection between parts of the circuit and the other rails

    My reasoning may well be wrong (I'm a newbie posting to .basics, after
    all), but if so, I'd like to be explained why. Your post wasn't very
    helpful in this sense.
    So how do you choose which potential level gets called "zero volts"?
    If there is no reason to prefer one level over another and the choice
    is completely arbitrary (e.g. in an isolated, floating circuit where
    the rails have similar areas and shapes), then there is no reason not
    to do low-side switching, so my argument isn't relevant.

    But if there is a reason to call a particular rail "zero" (perhaps due
    to the circuit topology, or due to the conductive area of the rail),
    then I'd guess the same reason also motivates having switched-off
    parts of the circuit connected to that rail.

    (Of course the motivation may be very slight, if we are only dealing
    with low voltage and low frequencies.)
    Yes, obviously. That's what I was suggesting in my post, after all: if
    there is a rail that is for some reason the preferred potential level
    for inactive circuits (whether we call it zero or ground or whatever),
    then it seems to me that it would make practical sense to have other
    rails have _negative_ instead of positive potential relative to that.

    But convention makes this difficult: most systems are designed so that
    other rails have a higher level than the ground, and we need to
    interoperate with them. And available components show this bias as
    well: for instance, there are far fewer switching regulators available
    that use the _higher_ input voltage level as the reference.

    So this is my concern: the choice of sign is by itself immaterial, but
    since we of course prefer to deal with positive instead of negative
    numbers, I fear it has induced practices that make us use
    semiconductors in a less than optimal fashion. It's not a very big
    problem, of course, but I still found it to be an interesting tidbit.

    Still, I'd of course be delighted to learn that I have misunderstood
    something and my concern is unfounded.

  6. Lauri Alanko

    Lauri Alanko Guest

    I think my amateurish pontifications are quite at place here. And I'm
    beginning to understand the motivations that led to the creation of
    this group in the first place.


  7. Fred Abse

    Fred Abse Guest

    I'm not going to argue, just to say that most of the engineers on the
    planet are happy with conventions the way they are.

    Go study electrical engineering in depth, study other people's designs,
    and try to understand why they did it the way they did.

    Whatever polarity convention you use, it doesn't stop you from using any
    component any way you want. The only limit is your own imagination, and
    what the laws of physics dictate.
  8. Lauri Alanko

    Lauri Alanko Guest

    To ask if my idea is correct, I have to first present it, competent or
    not. There's no way around that.

    Far too often I have encountered questions of the form "why does X do
    Y?" where X doing Y is something that follows quite naturally from all
    the rules of the system in question. Obviously the querier is
    misunderstanding something about the system, but their question doesn't
    give any hints as to what it might be. So the only way to respond is
    "why would you expect otherwise?" and hope for some clarification.

    To avoid this to-and-fro I'm straight away presenting the reasoning
    behind my idea so that it is easier for a responder to point out exactly
    where I went wrong. So far, though, no one has bothered to make use of
    the opportunity.

    Incidentally, this is also how science works: you put a paper up, with
    your argument and conclusions, and invite critique. The paper has to be
    detailed enough in order to be meaningfully criticized.

    You, however, seem to be offended by the very act of someone daring to
    present a detailed argument without sufficient credentials. In a newbie
    group. That's like dissing an undergrad for having the gall to present a
    paper at a seminar.
    Unlike, say, boasts about a relative's body count?

    Sorry. I'm here to learn, not to fight, so I'll leave the quips at that.

    However, conscience dictates that I'll at least try to make this group a
    nicer place, slim though the chances are. So let's make a feeble
    attempt. Here is a public service announcement:

    You are being a dick. Seriously.

    It is okay not to like beginners' stupid questions. The safest way to
    avoid them, of course, is to unsubscribe from .basics.

    And it's okay not to like my questions in particular. In that case,
    you'd better plonk me away. I don't mind, that's what killfiles are
    precisely for.

    But you are going out of your way to participate in a discussion with a)
    nothing of substance to contribute and with b) the express purpose of
    deterring someone from discussing a matter that they are trying to
    understand, and in particular from telling what their current conception

    Again, personally I don't mind too much. I have seen enough misbehavior
    towards newbies that I'm used to it. Sometimes it's due to frustration
    with previous newbies, sometimes due to imagined slights, sometimes due
    to just general condescension. It's not unexpected. But it is utterly
    antithetical to the purpose of this group, and likely to scare less
    thick-skinned newbies out from this group, and _that_ I do mind.

    I happen to love Usenet, and I still hope that new users would find this
    wonderful distributed discussion system that no one can censure or
    control. That's why I care about how a group looks to newbies. In fact
    I'm writing a magazine article about Usenet, and if a reader stumbles
    onto this group inspired by my article, I feel partly responsible for
    their experience. Seeing this sort of altercation in the history won't
    make it better.

    Now consider what you are doing. You are responding to an earnest
    question about electronics design - that you for some reason find
    offensive - with "you are wrong, and I won't tell you why, go to
    school". This sort of behavior will:

    * _not_ convince anyone that they are wrong

    * _possibly_ convince them that this group is useless for learning (if
    even a regular recommends asking in school instead of here)

    * _conceivably_ convince them that people in electronics are horrible

    Is this really the sort of contribution you wish to make to this group?

    I realize outburst has most likely been in vain, but I'm counting on the
    off-chance that you simply don't realize how destructive and
    unreasonable you are being, and just need it to be pointed out.

    So once more: you are being a dick. Stop it. Please.

    That's all.

  9. P E Schoen

    P E Schoen Guest

    "Lauri Alanko" wrote in message

    You're dealing with Jim Thompson here, and lately he has shown increasing
    evidence of being a "dick", especially when he goes on his disturbing
    tirades about his desire to kill liberals, and his childishly sexist remarks
    about women, and as you have seen, his gloating about the death count of
    someone who shares some of his DNA and most of his unpleasant attitude.

    But to respond on-topic to your inquiry, even if it had some validity (and
    I'm not saying it does not), the conventions for circuits using a positive
    rail have become so deep-rooted that only a very compelling reason would
    suffice to promote the alternative. Looking back at some of my very old
    circuits, it seems that I used a lot of PNP transistors such as the 2N1540
    (which was also Germanium), and the circuit had a negative rail and a
    positive ground, which is harder for me to conceptualize.

    Another reason for the positive rail and NPN or N-channel devices being more
    popular may be their closer correlation to vacuum tubes, which (AFAIK) do
    not have a PNP or P-channel counterpart.

    I do appreciate the time and thought you put into your post, but I think it
    is more of a philosophical proposition or "thought experiment" such as
    Albert Einstein enjoyed. I think it is always valuable to think "outside the
    box" to some extent, and "play the devil's advocate", and toss around some
    ideas which may prove to be zany, but perhaps may stimulate some discussion.

    Not everyone is a wacky as Jim Thompson, whose response may not be too
    surprising, but I do think you have been treated rudely, especially in the
    basics forum. Usenet may be on the road to extinction, and it will only
    survive by the influx and proper use of this resource by younger or newer
    people with electronics. I'm glad you recognize its unique value, and I hope
    you will stick around and help stabilize this resource.

  10. Phil Allison

    Phil Allison Guest

    "P E Schoen"Another reason for the positive rail and NPN or N-channel devices being more
    popular may be their closer correlation to vacuum tubes, which (AFAIK) do
    not have a PNP or P-channel counterpart.

    ** That would be the origin of the B+ rail convention.

    Early transistor radios use germanium PNP devices, so had positive ground

    Most old motor vehicles ( pre 1950s) had positive ground systems, usually 6V

    ..... Phil
  11. Yup, I wonder why? (My old tractor originally had a positive ground 6V battery, now all switched to 12V and neg. to Gnd)

    Though not at all an expert, ECL also uses a positive ground 'scheme'.

    To give the OP a bit of a break there does seem to be some positive bias in electronics at present. I can buy lots of positive voltage references, but negative ones are not so common.

    George H.
  12. Fred Abse

    Fred Abse Guest

  13. Lauri Alanko

    Lauri Alanko Guest

    Well, everyone deserves the benefit of doubt, even those who don't
    give it to others. It was worth a shot.
    Yes, obviously. The problems with running high-side n-channel switches
    are relatively minor, and there are well-established solutions. But it
    is annoying to think that there might be an easier way to do things
    and only convention prevents it.
    The popularity of N-type transistors is understood. What I don't get
    is how that associates with preferring a positive rail, when to me it
    seems like N-types would be more practical with a negative rail.
    Yes, of course. What else?

    A big part of learning about a system is studying different
    configurations and trying to understand how they behave. It is
    essential that the configurations are highly variable and illustrate
    all aspects of the system, and not merely constrained to those
    encountered in practice.

    For instance, when studying physics, it is quite on par to consider
    what happens when you send someone to space and back at light speed.
    It doesn't matter that this sort of thing will never ever happen in
    the real world: it illustrates how special relativity works, and you
    _have_ to be able to explain what happens if you want to claim to
    understand physics.

    However, in many online discussions of electronics I have seen
    something that I hesitatingly call an "engineering mindset": if a
    question is not related to a practical real-world problem that someone
    is trying to solve, it is not worth discussing. Those sort of people
    will no doubt think that discussion of unconventional rail
    configurations is a waste of time. (Those people also must think I'm
    mad for getting joy from getting a circuit work purely in simulation
    even if I have no intention of ever building it in practice.)

    For comparison, suppose a beginning math student just learned about
    numeral systems and exclaimed: "Hey, wouldn't it be better if we used
    a duodecimal system? After all, we divide things into threes and fours
    much more often than into fives!" A reasonable response would be
    something like: "Probably, yeah. It's not going to happen, of course,
    but you are right that it would have advantages." Even though the
    proposal is utterly unrealistic, making it demonstrates that the
    student has grasped something essential about the purpose of numeral
    systems and the tradeoffs in their design, and that is to be

    People with an "engineering mindset" would respond: "You're not
    experienced enough to challenge such basic practices. Keep on using
    the decimal system like everyone else."
    Indeed. Thanks for your reply, the first decent one I got in this

  14. Fred Abse

    Fred Abse Guest

    That's not what you said. You said:

    "If you are in electronics just think of electrons going with the arrow
    in diodes and bipolar transistor symbols and you won't go far wrong."

    In diodes and bipolar transistors, *conventional current* "goes with the
    arrow", NOT electrons.

    Voltage does not "go" anywhere, it just sits.
  15. Fred Abse

    Fred Abse Guest

    That's just to keep "other ranks" "in their place" ;-)
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