Connect with us

am transmitter - vlsi project

Discussion in 'Electronic Design' started by hananl, Oct 26, 2006.

Scroll to continue with content
  1. hananl

    hananl Guest

    I need to design an AM transmitter as a vlsi project (it needs to be
    located on smart dust later on).
    Does anyone have any good suggestions were to start from? I am familiar
    with vlsi, but don't really know how to implement such a big idea, into
    a vlsi simulation and circuit design.
    Looking on trasmitter's schemes didn't help, 'cause that's all
    resistors, amplifiers and capacitors. How do I translate it into vlsi
    Be very glad to any guidance...
  2. Joerg

    Joerg Guest

    You would need to provide more info: Frequency, power level, supply
    power budget, what's it modulated with, allowed distortion, regulatory
    requirements (harmonics etc.)...

    Look at what your available chip technology offers here. Then you might
    have to familiarize yourself with class-D amplifier technology, PWM
    schemes, synthesizing of envelopes etc. It also helps to look at how
    very modern AM transmitters in the "big league" work. The days of the
    classic big old plate modulator are numbered ;-)
  3. hananl

    hananl Guest

    The reason I don't have any specifications yet, beacuse I am on the
    begining of my project. Now all i want is to learn how to approach
    this. Maybe to read examples for vlsi projects related to trasmiters.
    Maybe to hear about a good book/pdf with examples and good explanations
    how to implement a reciever with a vlsi design.
    generally, it's for smart dust, so the power is very low, around
    several mili volts, frequecny:
    An amplitude modulated signal for typical AM broadcasts consists of a
    sinusoid with a frequency in a range from 0.535 MHz to 1.604 MHz having
    an amplitude
    that is varied (modulated) by an audio signal with frequencies of 20 Hz
    to 5KHz.
  4. Do you mean a transmitter in the US AM broadcast band? 550 to 1620

    Methinks you and your advisor should consider a few points:

    (1) An AM transmitter is usually implemented using rather simple
    analog techniques.
    One transistor as an oscillator, either LC or crystal controlled.
    Another transistor as a series modulator. Total parts count: under a
    dozen parts. You could do this on an IC I guess, but it's nowhere near
    VLSI. And to meet FCC regulations you'll need a LC tuned circuit of
    non-negligible size, so the space-saving aspect of VLSI won't be of
    much help.

    (2) You could do it digitally, but even then you just have a clock, a
    phase accumulator register, an adder and a multiplying D/A converter.
    Hmm, not all that much digital there either.
    Quite a few challenges there--- a transmitter requires a few milliwatts
    of DC power, hard to do on a dust speck. Plus a transmitter requires
    an antenna, preferably of a quarter-wavelength or more-- that's a lot
    larger than dust-size in the AM band.

    An AM band transmitter is not a very good match to digital, VLSI, or
    smart dust. Time for a rethink.
  5. Joerg ... tell me it isn't SO!!!! I've got a whole stockroom full of 6146s
    that will never fulfill their destiny.



    The days of the
  6. Joerg

    Joerg Guest

    Hello Jim,

    T'is so, I am afraid:

    Those should still hold some value in the ham radio community. Now just
    imagine what these tubes could do if operated in pulsed mode. Although
    nowadays one would probably use big FETs instead :-(

    At least you have the tubes. Out here the steel (!) tube in my old
    Rohde&Schwarz SMF has croaked. I like that generator because it has a
    real dial with coarse and vernier instead of having to press some
    buttons until the fingers cramp up.
  7. Joerg

    Joerg Guest

    Receiver? That won't help you much building a transmitter.

    I assume you meant milliwatts. Anyhow, the classical way to do this
    would be a full custom chip design. For new transmitter architectures
    that move towards the digital domain check publications such as this one:

    As Ancient Hacker has posted keep the antenna in mind. It needs to have
    certain dimensions or it won't radiate.
  8. On Thu, 26 Oct 2006 11:22:27 -0700, RST Engineering (jw) top-posted:
    [top-posting repaired]
    Just gold-plate their bases, and sell them to audiophools for $900.00
    apiece. ;-)

  9. Joerg

    Joerg Guest

    Hello Rich,
    And don't forget fancy plate connectors to go with them. Maybe like a
    small moped cylinder with gold plated fins, with an official endorsement
    paper from the RST Institute, Audio Research Department :)
  10. Well no.

    The audio signal is mixed with the carrier, and the output of that mixer
    is the original carrier, unmodified, and two sidebands on either side
    of that carrier. Taken as a whole, the amplitude varies, but the carrier
    itself stays constant.

    This is likely a useful hint, because it's far more common to see a balanced
    mixer in ICs than something called an "AM modulator".

  11. Tim Wescott

    Tim Wescott Guest

    Is it a good red-blooded American tube with a number that goes
    6-letter-number-letter, or is it one of those weirdo euro-things?

    You do know that there are tubes available NOS, as from, don't you?


    Tim Wescott
    Wescott Design Services

    Posting from Google? See

    "Applied Control Theory for Embedded Systems" came out in April.
    See details at
  12. Tim Wescott

    Tim Wescott Guest

    Technically correct, but misleading as hell. The OP's explanation is
    also technically correct, and less misleading to the transmitter
    designer, although of less use to the receiver designer.


    Tim Wescott
    Wescott Design Services

    Posting from Google? See

    "Applied Control Theory for Embedded Systems" came out in April.
    See details at
  13. Tim Wescott

    Tim Wescott Guest

    hananl wrote:
    (top posting fixed)
    Actually, most of us know what "AM" means, and we also know that "AM"
    doesn't mean "North American AM broadcast" unless you say so.

    You can make the RF portion of an AM transmitter with a single
    transistor if you have sufficient audio power available. But you can't
    modulate 100%, and you'll have significant FM riding on your AM.
    Consequently your signal will be splattered all over the band and it
    won't be as efficient as it could be.

    I can't give you an upper limit, because (a) you still haven't stated
    all your requirements, (b) you haven't said if you're going to do this
    on a process that lends itself to analog and (c) I'm not a chip designer.

    I suggest that you do more web searching, or be more specific with your
    questions, or both. You may find the ARRL Handbook to be informative
    for background study -- you may even luck out and find an AM transmitter
    block diagram.


    Tim Wescott
    Wescott Design Services

    Posting from Google? See

    "Applied Control Theory for Embedded Systems" came out in April.
    See details at
  14. Joerg

    Joerg Guest

    Hello Tim,
    It's even weirder. A Euro thing built from steel instead of glass.
    EBF11, six of them, vintage 1938 :-(

    Not that one. The EBF11 is a dual diode plus pentode with an odd form
    factor. It is only 1.5" high but also about 1.5" wide. Looks
    unbreakable, like it was designed to survive world wars two through five
    but unfortunately one of them is now experiencing episodes of
  15. SioL

    SioL Guest

    Interesting, a digital modulation approach takes 72 stages, presumably of equivalent output power?
    Doesn't that yield a pretty poor dynamic range?

    Or is it a 72-bit D/A, which sounds totally crazy.

    And they must be in perfect phase and at perfect power levels across the band.

  16. Fred Bartoli

    Fred Bartoli Guest

    SioL a écrit :
    Seems to be 72 identical PA, plus one linear one to interpolate between
    the discrete power levels. (top of p5, point 2)

  17. Well, your explanation is mighty confusing, and partially incorrdct,
    IMHO but it's not your fault.

    Originally AM modulation was done in a mighty crude but effective way--
    by putting a regular old telephone's carbon microphone in series with
    the transmiting antenna!

    That setup should give you a clue as to what "AM" really is, in the
    time-domain that is-- The carbon mike's resistance varies up and down
    as sound waves hit it-- resistance goes DOWN as a high pressure wave
    hits, then goes UP as the lower pressure hits, each happens once each
    audio cycle.

    Later in "grid modulation" was figured out. here things get a bit
    fuzzy, but you could look at it as the audio moves the tube's bias
    point around and effects it's efficiency. You can also look at it as
    a "mixing" action is going on, which leads you to think a
    frequency-conversion is going on. Both viewpoints are correct, and of
    course both are misleading.

    Later on "plate modulation" came in, where you put the audio signal in
    series with the plate DC supply. Again it's obvious what's happening
    in the time-domain: the plate voltage goes up and down with the audio,
    hitting twice the voltage on the positive peaks, and zero voltage on
    the negative peaks. Looked at PURELY in the time-domain, it looks like
    the carrier is going UP and down in amplitude. In the frequency
    domain of course it looks like sidebands are popping up, which they

    Much later on, the theory of sidebands, "mixers", and "multipliers"
    was cleared up. mathematically, the old "mixers" were revealed to be
    kinda like a poor non-linear multiplier, which multipled the two input
    voltages (not terribly linealy, but good enough). And the math said
    when you multipled two sie waves you end up, in the frequency domain,
    with their sums and difference frequencies.

    But people kept calling capacitors "condensers", and multipliers
    "mixers", so the confusion continues.

    You can look at it as "multiplication" or "mixing"-- both are at least
    partially correct.

    But if you look at the output waveform on a scope, you have this
    waveform where the RF amplitude gores UP and DOWN, all the way down to
    ZERO, so in some sense the carrier sure LOOKS like it's going down to
    zero. And if you think of the B+ voltage going to zero, it's hard to
    imagine how the carrier can still be going out when the plate voltage
    is zero.

    So it's at least partially incorrect to say the carrier "stays the
    same" and "sidebands pop up".
  18. Guest

    They're glass under steel. The tube itself is a more-or-less
    conventional glass tube, apart from the stubby shape. I sawed
    one apart as a youngster. Didn't quite make it, the hood *is*

  19. To get back to the poor original poster's question, and to point out
    some POSITIVE and realistic options, instead of all our poo-poohing:

    If you want to make a transmitter, a really small one, one that can
    actually broadcast some distance more than a few millimeters, and
    still be detectable over the background noise:

    (1) Think hard about the physics of the antenna situation-- a tiny
    antenna implies a HIGH frequency. For instance, if your "dust" is
    going to be on the order of an IC chip size, the frequency, in order to
    have a 1/4 wave antenna, is going to have to be in the tens of
    gigahertzs region.

    (2) I suspect your vlsi process is not up to building gigahertz-region
    digital frequency synthesizers. It may be capable of low-gig
    rizetimes, but for a true synthesizer you'd need at least 20 times the
    output frequency to make effective synthesizer components, like adders
    and D/A's.

    (3) Also think about the power situation-- even if your vlsi is
    low-power (which it won't be at GHz sppeds), a transmitter will need
    several milliwatts of ouput power to overcome background noise level,
    and that requires several times the input milliwatts.-- figure out how
    large a battery or solar cell has to be to generate a few milliwatts
    for even a few seconds.
    Prolly a whole lot larger than ic-size.

    (4) Also think about the rules of your country's FCC. In the USA you
    can't just broadcast willy-nilly, there are specific bands and emission
    modes required (I think, unless there's some loophole). As far as I
    know, you have to stick to 100mw or so max power, and in the AM or FM
    bands, or around 13.56MHz, or twice that, or the 47 and 4xx MHz old
    wireless phone bands, or the microwave 2.6 GHz band, or a few other
    narrow spots. And I suspect you have to do AM in the AM band, FM in
    the FM band.

    Just a suggestion, but the choices seem to narrow to:

    VLSI digital synthesizer for the AM BC band, but with a long antenna
    (up to 3 meters in the USA).

    VLSI analog synthesizer in the FM band, with a several inch antenna.

    IC analog oscillator/modulator in the 4xx or 2.6 GHzMHz band, with an
    inch or so of antenna.

    ... and for power source, the tinyiest of lithium hearing-aid batteries,
    so I hope your VLSI process can run on 1.5 or 3 volts :)

    Hope this helps.
  20. SioL

    SioL Guest


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