Connect with us

LC generator

Discussion in 'Electronic Design' started by Andy I., Jul 9, 2005.

Scroll to continue with content
  1. Andy I.

    Andy I. Guest


    I've designed an LC generator supposed to accept a wide range of tank
    capacitance values with the tank inductance fixed, and a wide range of
    tank inductance values with the tank capacitance fixed. The objective is
    to be able to measure the unknown tank capacitance/inductance as
    precisely as possible (better than 1%).

    Before I go forward and buy the missing parts and solder the whole
    thing, I thought I might ask the s.e.d. community to take a look at the
    schematics, to spot any obvious problems / easy improvements.

    I've posted the schematics in a.b.s.e. under the same Subject.

    Thank you in advance for your feedback.

    -- Andy I.
  2. Joerg

    Joerg Guest

    Hello Andy,
    To be honest I don't know why you want to do that with opamps. All you
    need is a good RF transistor, an amplifier after that which doesn't have
    to be particularly linear, a Schmitt trigger that's fast enough and then
    off into a frequency counter.

    Look at how "grid dip meters" are designed. These are strikingly simple
    circuits. Their oscillating stage is able to work from a few hundred kHz
    to several hundred MHz. Examples can be found in the ham radio
    literature and probably on the web as well. The name grid dip meter is a
    left-over from the tube days. I still have one with tubes ;-)

    Regards, Joerg
  3. Andy I. wrote...
    Why don't you talk us through your design. For example, it's not
    clear where the D.U.T. inductor and capacitors go, and where the
    output is, etc. What are your design goals, design specs, the
    operating range, accuracy, etc.?
  4. Andy I.

    Andy I. Guest

    Hello Joerg,
    I have looked up some "grid dip meter" schematics on the Net, what I've
    found suggests that those are three-point oscillators, using a set of
    interchangeable LC tanks to cover an extended frequency range.

    I think I need to explicitate some requirements to the device: the
    operational mode is to connect either an unknown capacitance in parallel
    with the tank, or an unknown inductance in series with the tank's
    inductance - the same as for the well-known aade meter. This entails
    that the L/C ratio varies very very much: from 1.2e8 to 4e2 [H/F] (C
    between 680pF and 0.2uF for the L fixed at 82uH, and for L between 82uH
    and 82mH for the C fixed at 680pF), and that only 2 points of connection
    are available for the tank, the output frequency is between about 15kHz
    and 670kHz. With these requirements, do you see a simpler circuit?
    Sure, the comparator is missing from the schematics, but that's an easy


    -- Andy
  5. Joerg

    Joerg Guest

    Hello Andy,
    Not the ones I know. They have a variable capacitor on the inside and a
    set of inductors that you can plug in. Or you can connect an unknown
    inductor to the socket. There are only these two connections.
    That is a very wide range. It will only oscillate if the DUT has enough
    Q. Lots of Q. Basically, if a good grid dip meter cannot make it
    oscillate chances are that nothing else will.

    Can't you measure the unknown capacitance or inductance in another way?
    Even a regular DVM in the lab does a pretty good job there. For
    inductors we use an analyser but mostly just because it is there.

    Regards, Joerg
  6. Joerg

    Joerg Guest

    Hello Winfield,
    If you'd donate your new 4192A to Andy he might not need to build all
    this ;-)

    Regards, Joerg
  7. Andy I.

    Andy I. Guest


    The DUT inductor goes in series with L1. The DUT capacitor goes in
    parallel with C8 (sorry for the numbering!). The output is the output of
    eitehr U8 or U1 (goes to a frequency counter).

    An etalon capacitor is temporarily connected in parallel with C8, and
    based on the frequency shift, the reference capacitance of the tank is

    Then when the DUT is connected, based on the frequency shift, its value
    is calculated.

    So the operational mode is the same as for the well-known "aade" meter.

    The objective of going pure sine wave (and the harmonics are pretty low
    here, the signal is 80mV pp at U1 output), is to attain as much absolute
    precision as possible. Better than 0.5%, and for L down to 10nH.

    Because the ratio L/C varies widely from 1.2e8 to 4e2 [H/F] (C between
    680pF and 0.2uF for the L fixed at 82uH, and for L between 82uH and 82mH
    for the C fixed at 680pF), and because only 2 points of connection to
    the tank are available, an opamp-based oscillator was chosen (U1), with
    an opamp with high open loop gain (also because of the non-negligeable
    ESR of the C under test).

    The negative feedback coefficient is set with the two FET-based
    attenuators in series. Two attenuators were set up in series because of
    the required 80dB attenuation range.

    The output signal is rectified (germanium diode here) and filtered
    before closing the oscillator AGC loop. The filter values were chosen to
    provide as oscillation-free as possible AGC regulation on one hand, and
    the maximum precision of the amplitude on the other.


    -- Andy
  8. john jardine

    john jardine Guest

    The arrangement should work fine but seems sensitive to low (=normal?)
    values of inductor Q.
    Using C8 and L1 with a Q of say 10 (series loss of 30ohms) the U1 circuit
    won't oscillate. Not enough positive feedback due to the lossy divider
    action of R7 and the working tuned circuit new Rd of about 3kohms. (R7 needs
    a 0.47u in series). It'd be difficult to increase the loop gain to
    compensate as the fets are maxed out in the ohms area and another auto
    attenuator circuit on the +ve feedback side is beyond the pale.
    My own experience (ie. years of mistakes and stupidity :) suggests it's
    better in these 'wide ranging Rd-Q' cases, to just feed the tuned circuit
    from a controlled, good quality, current source. The current source can then
    act as the oscillator ALC element. (I say this as I've been running a
    similar project).
  9. Andy I.

    Andy I. Guest

    Hello Joerg,
    Ok, I think I will do some more search.
    I could have. But it looks like one has to make quite an investment to
    get better than 1% precision measurement for capacitance. And I wondered
    how far one can go with such a relatively simple circuit (it is not
    good for capacitances larger than 1uF, but that's ok).


  10. BFoelsch

    BFoelsch Guest

    Well, almost.

    I looked through a few drawings of grid dip meters and the ones I have
    generally do use a 2 terminal connection for the inductance, but use a dual
    variable capacitor in the Colpitts configuration. No problem for measuring
    inductance, as you suggest, but measuring capacitance would be somewhat more
    complicated. I couldn't find a drawing of a grid dip meter with a 2 terminal
    LC connection but that doesn't mean they don't exist.

    Of course, if tunnel diodes were still available, you could do the whole
    thing with about 5 parts...
  11. Johnson

    Johnson Guest

    For a good review of the principles of precise RCLD measurement, go to the
    US Patent and Trademarks Office and look at the theory described by the
    assignee Genrad. You can also search under "Impedance Measurement" in the

  12. M. Saindoux

    M. Saindoux Guest

    An inductance may give a slightly different value depending on the
    oscillation frequency.
    Inductances are measured at particular frequency so the 1% precision is more
    or less... utopian.

  13. Kevin

    Kevin Guest

    There is a similar device presented here:

    It uses a comparator so that it works over a wide range of LC ratios.

  14. Andy I.

    Andy I. Guest

    Yes, I know that circuit. I wanted a more linear circuit, in which the
    relationship between oscillation frequency and the tank resonant
    frequency would be more straightforward.

    -- Andy
  15. Andy I. wrote...
    What? Neil's relationship is textbook perfect, with the L = 1 /(w^2 C)
    formula, etc. The other formulas may look a bit complicated, but that's
    because he adds a clever method to use measurements on a 0.5% "standard"
    capacitor, included in the instrument, to provide precision calibration
    for the rest of the measurements.
  16. Andy I.

    Andy I. Guest

    Yes, they are too perfect, I think; these are not complicated formulas.
    What I am saying is that there are several other factors that should
    impact the real oscillating frequency in a complicated, non-reflected in
    the above formulas way, especially at the hi-frequency end: the
    comparator's threshold levels, the slew rate/propagation delay, the
    power supply voltage, ... Instead of trying to account for these
    factors, why not work with a regular oscillation.

    -- Andy
  17. Andy I. wrote...
    I had a similar reaction the first time I saw Neil's circuit, but after
    seeing how well it works, I realized the oscillating frequency isn't
    much disturbed by the things you mention (they affect the oscillator
    "gain"), but rather is determined by the high-Q L-C resonance, which
    overcomes those issues. His automatic calibration technique hides the
    rest of the true disturbing issues, such as stray capacitance.

    I recommend you get one of his little beauties. Quite inexpensive, too.
  18. Joerg

    Joerg Guest

    Hello BFoelsch,
    When I have to measure a capacitance in a tight spot where I can't get
    to with an analyzer I just connect a well known inductor to it. Then I
    use the dip meter to figure the resonant frequency of the resulting LC
    circuit and, bingo. But it happens so infrequently that I have put the
    dip meter on the list of units for regular battery swaps. Else it might
    just start to leak from old age.
    Tunnel diodes were all the rage when I was young. The ultimate boutique
    part. Every professor marveled at them but you couldn't get them
    anywhere or they cost an arm and a leg. So I guess they went the same
    way as the unijunction transistor (I did have one of those). I prefer
    the bread and butter stuff such as ye olde BFS17A. Available for pennies
    at every street corner and blazingly fast.

    Regards, Joerg

  19. from

    Part Number: 1N3712

    General Purpose Tunnel Diode
    Price Special: $ 68.79


  20. Andy I.

    Andy I. Guest

    There seem to exist an offer of legacy Soviet parts from various
    distributors in Russia, including the Arsenid-Gallium and Germanium
    tunnel and reversed diodes (not expensive). If only there was a good
    application today. Except the micro-bugs :)

    -- Andy
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