Connect with us

simple transistor selection question

Discussion in 'Electronic Components' started by Danny Sauer, Aug 26, 2004.

Scroll to continue with content
  1. Danny Sauer

    Danny Sauer Guest

    Ok, so I know a little about electronic devices. However, I don't
    have the parts/spec books memorized like lots of people who do this
    stuff all the time. :)

    I'm making a device to make my tail lights flash in entertaining
    patterns. It's overkill, but I'm using a PIC microcontroller (a
    16f676) to handle the flashing patterns. I've got the device working
    just fine with LEDs, but actual 1157 taillight bulbs pull way more
    current, which I'm pretty sure will be too darned much for this little
    micro to handle. My plan is to replace the LEDs with a transistor to
    control each bulb individually. The bright side of an 1157 is usually
    around 25-30 watts, so at 12 volts that gives me a worst case of about
    2.5 amps. How about allowing for 3 amps at 14.4 volts - 43 watts.
    What would be a good transistor to use for this? I can switch either
    the positive or negative side, it doesn't really make any difference,
    though switching the positive side would likely be a bit more
    convenient (more wire to work with).

    Can someone recommend a supplier and a part number or something where
    I can get stuff in relatively small quantities? I only need about 8
    per device, and I'm only building one or two devices for the time
    being. If I could get something at the local Radio Shack (there are
    no electronics suppliers nearby other than that place), that'd
    probably be alright, too.

    Thanks a lot!
  2. Quite a few power transistors will work. The main things to note are:

    (1) The transistor must be *really* well "turned on" in order to have a low

    (2) The transistor shouldn't dissipate anywhere near 43 watts. That is, the
    transistor wattage has nothing to do with the wattage of your taillights.

    (3) Have you considered using relays?

    Getting back to transistors, this is not the cheapest solution, but I'd
    consider N-channel switching MOSFETS that (a) saturate fully with a
    logic-level input (e.g., IRL series, not IRF series), and (b) have an "on"
    resistance that is quite low, like 0.2 ohms or less.

    I don't have a data sheet in front of me and am about to have to go... If
    others don't chime in, let me know.
  3. Dave Platt

    Dave Platt Guest

    I'm making a device to make my tail lights flash in entertaining
    For a high-side switch, you'd want a PNP power transistor, or a PNP
    Darlington. [Darlingtons are actually two transistors, connected
    together, in a single case. You turn on the first, and the first
    turns on the second. A Darlington pair has a much higher current gain
    than a single power transistor, and is thus easier to drive.]

    You want a part with at *least* a 50-volt limit for auto applications,
    due to voltage spikes caused by a phenomenon called "load dump". The
    TIP125 and TIP127 Darlingtons have 80-volt limits, and will handle a
    maximum collector current of 5 amperes. The NTE262 is an acceptable
    substitute for both of these parts.

    You should be able to drive the base of one of these from a PIC's
    LED-capable output pin, through a current-limiting resistor. Figuring
    maybe 15 volts (to be generous) at the transistor, you'd want a
    resistor of about 750 ohms between the PIC pin and the base. This
    will draw 20 mA out of the base when you turn on the PIC's output
    driver (set the pin value to 0 and set the tristate to "drive it,
    please"), and should turn on the Darlington quite hard... enough that
    its voltage drop will be very low, and it won't dissipate a
    significant amount of power in the transformer. I'd guess you won't
    even need a heatsink. To turn off the transistor and the light,
    just disable the PIC's tristate driver for that pin.

    Oh... make sure that you're using one of the PICs which doesn't
    mind having its pin "pulled up" to a high voltage when it's
    tristated off. Most of the newer ones will tolerate this, I
    believe. If you're using one which will malfunction or fry
    if the pin is pulled high, then you'll need to drive the
    transistor indirectly. Use the PIC to drive the LED in an
    optoisolator (ideally one with a 100% or greater current
    transfer ratio), and connect the optoisolator's output
    between ground and the current-limiting resistor. is one source, and I'm sure also
    carries them. Also check the various surplus-electronics outfits on
    the Web - has TIP127s. Any well-stocked
    electronics outfit can probably sell you the NTE262 - the price will
    be a good deal higher than for the TIP equivalent but if it's
    available locally to you there won't be a shipping charge.
  4. Suraj Singh

    Suraj Singh Guest

    I think consideration of power dissipation in the transistor is definitely
    required. Suppose you are using a silicon NPN power transistor with an on
    state voltage of 0.7 Volts at 3A current will dissipate 3*0.7=2.1Watts of
    power which is critical and will require a heat sink if ambient temperature
    is not too cold ( like an outdoor temp in summer).

    You can say that you will be switching your transistor on and off and it
    will be off for 50% of the time. If your switching frequency is fast enough
    ( more that few seconds) you can divide the power by 2. Now if you are
    dissipating 1watt in a TO-220 type package you should consider using a small
    heat sink be safely operate the device at 40 degree Celsius. There will be
    some loss of power in Transistor during switching transition which you can
    neglect because switching frequency will be low enough.

    A general purpose robust transistor available in RadioShack could be used in
    your case is TIP3055 (TO-220) or 2N3055 (TO-3). It would be better to make a
    Darlington with another low power general purpose transistor as the hfe of
    3055 is a bit low.

    MOSFETs are much better switch than transistor but they burn out more

  5. Using a Darlington is wasteful because its saturation voltage is always
    at least a volt. At 3 or 4 amps, that's 3 or 4 watts to dissipate and
    you'll need a bigger heatsink.

    Use two separate transistors, such as a MJ2955 PNP power transistor with
    a current limiting resistor in series with the base, and use a 2N4401 to
    drive it, with its emitter grounded. Here's an example.
  6. Danny Sauer

    Danny Sauer Guest

    Doesn't it have to pass all of the current through, though? If use
    something only rated to pass a couple of amps, won't it likely burn up
    if the load wants to draw more current through?
    I'd like to be able to fade the lights out, and that'll either require
    the ability to rapidly switch between on and off, or some kind of
    variable resistance. I suppose I could control a potentiometer with a
    stepper motor, or something awful like that, but that's not a
    realistic option. Even if a relay was able to keep up with the
    high-speed switching, the contacts wouldn't last long in that type of
    use. So I'm pretty much left with a transistor for this situation.

    Thanks much for the replys, all. I've got a few ideas to play with,
    and at least an idea of what I'm looking for. It's much easier to
    have the list limited down to a few hundred rather than having several
    thousand possible options. :)

  7. Didn't you read my earlier followup?
  8. Jeff Walther

    Jeff Walther Guest

    I haven't seen the earlier messages in this thread, so apologies if I'm
    retreading old ground.

    Have you considered using a digital potentiometer? Xicor makes a line of
    digital pots that might work. I don't know what kind of apperage they
    support, so you may have to run the output through some amplification, but
    they are a more elegant way to convert a digital signal to a potentiometer
    type output.
  9. Danny Sauer

    Danny Sauer Guest

    My ISP doesn't have a newsfeed, so I'm posting through Google. It
    works, but sometimes (like this time) my replies are weeks late
    because posts frequently don't show up for several days. So no, I
    didn't read the earlier reply. :) That does implement pretty much
    what I was figuring on, though... Thanks.
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