Connect with us

History of decoupling capacitors

Discussion in 'Electronic Design' started by Theo Markettos, Jun 24, 2004.

Scroll to continue with content
  1. I was just wondering...

    Roughly when were power supply decoupling capacitors first deemed necessary
    in circuit design? I'm thinking mostly logic circuits here, though I can
    see that audio/RF circuits might have needed them too. I suppose every DC
    PSU has a smoothing capacitor doing roughly the same job but I was more
    thinking about adding capacitors around the circuit to provide localised
    decoupling. Were they necessary in valve (vacuum tube) circuits? If not,
    in what technology did they first appear (RTL/DTL/TTL etc)? Was there any
    prior use in another field (say RF) and they were later used in logic
    circuit?

    The earliest boards I can remember (late 70s/early 80s) seem to have them
    on, but I haven't seen many earlier boards and can't say I've been looking
    for them specifically...

    Thanks,
    Theo
     
  2. Howard Long

    Howard Long Guest

    I'm certain that DTL PCBs used them. The problem is to do with current
    spikes that occur when the gates change state. The power supply lines have
    inductance and so a fast spike has the potential to reduce the voltage on
    the power supply lines sufficiently to take the device out of spec on the
    voltage, with the potential for flip flops to unexpectedly change state. By
    having localised caps, these spikes are smoothed away.

    The caps have to be smallish values as larger caps have too much inductance:
    electrolytics, for example, are made by tightly rolling a conductor and
    dielectric together, perfect for a bit of inductance.

    Valves tend to have less dependency on the stabilised power supplies used by
    solid state logic - building stabilised power supplies for the HT valves
    require was difficult in the old days, so circuit design took that into
    account. Inately therefore, valve designs were generally less susceptible to
    current spikes.

    Because analogue circuits tend not to have state, and because they don't
    usually generate large current spikes, decoupling caps aren't used so much.

    In RF and analogue terms, coupling (rather than decoupling) caps are used
    extensively, allowing various stages to have differing DC biases. Decoupling
    caps are also often used in RF and analogue designs primarily to prevent
    gain stages going into self-oscillation.

    Kind Regards, Howard
     
  3. Reg Edwards

    Reg Edwards Guest

    Decoupling capacitors and other components have been used in audio and RF
    amplifier circuits for 100 years, ever since the earliest days of valve
    amplifiers. They were used and are still used in telephone circuits
    although not so obvious. A microhone is a mechanical amplifier and currents
    must be controlled and diverted into the right channels.

    Decoupling of individual components and pins of integrated circuits are
    precautions. Mainly because of the very high frequencies involved and chip
    and track physical dimensions being an appreciable fraction of a wavelength.
    Or an appreciable fraction of 90 degrees phase shift along tracks and across
    ground planes. Or unwanted radiation and crosstalk. You digital
    chip-catalogue merchants can escape from analogue only at your own peril.
     
  4. John Larkin

    John Larkin Guest

    The earliest audio/radio stuff used batteries, so had no hum problems,
    and used small numbers of tubes with such low gain that circuits were
    stable without supply bypasses. I could poke through my old books, but
    I'd guess that filter and bypass caps became common sometimes in the
    1920's maybe.

    John
     
  5. Tam/WB2TT

    Tam/WB2TT Guest

    Bypass capacitors were used with vacuum tube audio and RF amplifiers. One
    very important reason was to prevent feedback, and possible oscillation, by
    way of the plate supply voltage.

    Tam
     
  6. Joerg

    Joerg Guest

    Hi Theo,

    Nowadays the word "decoupling" is use to describe decoupling current
    spikes from the power supply. IOW to prevent them from entering there or
    causing a momentary sag in voltage.

    In the olden days the word also meant decoupling one stage from another,
    for example the plate DC level from the much lower gate DC bias of the
    next amplifier stage. In that respect they are really historic. Even the
    vintage telephones with cranks and separate ear pieces had a decoupling
    cap. They just didn't call it capacitor but condenser.

    Regards, Joerg
     
  7. CBFalconer

    CBFalconer Guest

    .... snip ...

    When local current spikes became significant. So called high
    frequency tube circuits always needed them. In general,
    transistors draw heavier currents than tubes, at lower voltages,
    and so are more likely to need a local supply. It's function, not
    time.
     
  8. I dissassembled an old German valve.based radio from the 1930's, and
    those decoupling capacitors were present there; especially around
    swithces; e.g. from MW to KW.
    The oldest radio hardware I have inspected was from 1924, and that
    was a very simplistic design for a MW radio. No excess capcitors there.

    -- mrr
     
  9. John Larkin

    John Larkin Guest

    Actually, capacitors - and resistors, for that matter - were expensive
    in the early days of radio. In 1921, a 1M grid-leak resistor could
    cost a dollar or two, and paper or mica caps, a few nf, were similar.
    A lot of early radios had no fixed capacitors at all.

    John
     
  10. Max Hauser

    Max Hauser Guest

    I agree with those here having experience with VT (valve) schematics.
    Mention of "power supply spikes" suggests either digital (rare in consumer
    products until cheap digital ICs circa 1970, you could see them arrive
    gradually, and the hype that came with them) or more specialized, pulsy
    circuitry as in TV scan; yet I've seen many multiple-stage VT audio
    amplifier schematics with successive R-C networks, sometimes inductors too,
    leading from one power supply feed to the next, "decoupling" stages from
    each other to discourage unwanted intercourse via power supply line [1].
    That "decoupling capacitor" language in English long predated commonplace
    digital hardware. When early transistor amplifier circuitry mimicked its VT
    predecessors closely, the decoupling networks came along too.

    I didn't see it mentioned here, but decoupling caps INside monolithic
    circuits have also been common practice for many years, if the chip needs
    them and can fit caps of useful size. (I'm talking pure-monolithic; much
    more can be done with specialized components added to a leadframe or header
    before the package is finished.) Oxidized silicon grown from crystal is
    actually an outstanding capacitor dielectric in multiple ways. (And it's
    amazing what a few picofarads can do, in the right spot, to the interior of
    the inductance matrix of the leadframe.) And then there are the RF ICs ...

    Capacitors have come far since the adventurous but fortunate Mynheer van
    Musschenbroek [2]. If you missed the good stuff, it's currently at

    http://tinyurl.com/33h2u

    -- Max


    [1] First I edited "intercourse" to "communication." Then I changed it
    back.

    [2] "Pure Hydrogen has been often respired by different philosophers,
    particularly by Scheele, Fontana, and the adventurous and unfortunate
    Rosier." -- Humphry Davy, _Researches Chemical and Philosophical,_ Division
    1, 1799.
     
  11. Joerg

    Joerg Guest

    Hi John,
    On the other hand in those days people were still able and willing to
    "roll their own". Some parchment type paper, or at least construction
    paper, lots of foil from chocolate wrappers plus litz wires and you had
    a capacitor. Resistor could also be "drawn" on paper.

    Regards, Joerg
     
  12. Guest

    FWIW, homemade capacitors are often the rule with homebrewed high-voltage
    stuff (ie, tesla coils). Glass bottles wrapped with foil and filled with
    salt water are common, albeit crude, but workable. One can dramatically
    increase the voltage handling of an air-dielectric variable capacitor
    (typical radio tuning cap) by submerging it in mineral oil. Try pricing a
    <lots of kilovolts> variable cap sometime.
     
  13. Joerg

    Joerg Guest

    Hello,
    Plus we'd get to drink the stuff that was in there before :)
    Did that with a high power dummy load once. But after a while some oil
    started to seep out and made a huge mess.
    Yes, I painfully rememember forking over lots of bills for a variable
    vacuum cap.

    Regards, Joerg
     
  14. fcweed

    fcweed Guest

    Where I grew up, we used the terms "decoupling" and "bypass" capacitors,
    where the latter provided a noise glitch bypass from power to ground.

    wrt. the original question, bypass/decoupling caps for digital circuits
    were already necessary with RTL. The first flip-flop circuit we built in
    the first digital logic course I took was a simple 4-stage ripple
    counter - Not much to screw up, but it didn't work. The instructor
    didn't even look at the circuit wiring, he just clipped a small cap
    between power and ground and things worked as expected - Which may have
    been the real point of the whole exercise.
     
  15. Hank Oredson

    Hank Oredson Guest


    I would guess you mean "in computers".

    The early mainframes that used tubes had plenty of them!
    I still have some .02 250V ceramics from that era, from
    Univac surplus, left over from the 1103 days ...

    When transistors were first used there were decoupling
    capacitors on each voltage on the cards that plugged
    into the backplane, and others as needed scattered around.
    Some backplanes had caps here and there as well.

    Going back a bit further (into the 1920s) one would find
    them on power supplied to audio / RF / IF stages. Usually
    wound foil capacitors with values of .005 to maybe .1 uFd.
    These were perhaps .5 cm diameter and 1 - 2 cm long.

    I guess the "right" answer is "There never have NOT been
    decoupling capacitors in any electronic device."

    --

    ... Hank

    http://horedson.home.att.net
    http://w0rli.home.att.net
     
  16. Rich Grise

    Rich Grise Guest

    The way I read it years ago, the bypass capacitors form a circuit path
    for the signal (or signal return) around the power supply.

    Cheers!
    Rich
     
  17. Rob Storey

    Rob Storey Guest

    IIRC, bypass caps became particularly important with the advent of TTL
    logic, since (again IIRC) the totem-pole outputs in these devices have a
    very brief period during switching when both output transistors are on,
    causing a short short(!!), which in turn caused supply line spikes.
    DEC used to recommend a cap (0.1uF or 0.01uF) on every IC package.

    Rob STorey
     
  18. Ken Taylor

    Ken Taylor Guest

    Many was the motorboating valve amp when a decoupling capacitor went dry.

    Ken
     
  19. Stuart Ballantine, in "Radio Telephony for Amateurs, Second Edition"
    (Philadelphia, 1923) devotes a few pages to a scheme for reducing
    instability in multistage amplifiers wherein there is a bypass
    capacitor from each tube's plate supply lead to ground and (if
    necessary) a series inductor between this capacitor and the power
    supply.

    So the concept was certainly well known and available to the general
    public by 1923. Though the lack of use of this circuit in the other
    parts of the book would tend to indicate that it was considered a bit
    extravagant.

    Steve
     
  20. John Larkin

    John Larkin Guest

    I have a few references from the 1920's. Mostly what I see is battery
    powered with hardly any caps at all. Interstage coupling is mostly via
    transformers... one transformer would give about as much gain as an
    additional R-C coupled tube stage, replacing 7 expensive parts. Tubes
    mostly ran at Idss in those days, so the transformer was ideal.
    Transformer coupling also greatly reduces the need for supply
    bypasses.

    By the mid-30s, tubes were cheaper and had much higher Gm's, voltages
    were higher, AF coupling was mostly R-C, grids were back-biased,
    tetrodes were appearing, and supply bypassing was universal.

    I like the NE25339; it's a tetrode GaAs mesfet. Needs bypassing.

    John
     
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

-