Connect with us

Capacitance Meter or LCR

Discussion in 'Electronic Design' started by shayan, Feb 25, 2005.

Scroll to continue with content
  1. shayan

    shayan Guest

    Hello every one, Can i measure the continuous variation(with frequency
    above 10 kHz) in the capacitance and dielectric constant of a fine
    capacitor with capacitance meters or LCR's? Please introduce me a web
    link for these instruments.
  2. qrk

    qrk Guest

    An equally vague answer to your vague question is us a network
    analyzer with an impedance function if you need to measure impedance
    over a frequency range. Hewlett Packard (Agilent) makes these. HP4195
    and Agilent 4395 come to mind. This will allow you to measure complex
    impedance from 10 Hz to 500 MHz with the proper fixturing. Although
    the spec sheet says it will measure impedance above 100 kHz, the
    analyzer will do impedance measurements below 100 kHz if you use DC
    coupled fixtures - at least in the HP4195.

  3. shayan wrote...
    I have an hp 4275A LCR meter, which measures from 10kHz to 10MHz with
    5.5-digit resolution. You will find, as I have, that most capacitors
    will not have a significant capacitance variation over most of that
    frequency range. If you have in mind a specific part or type of
    capacitor, I would be happy to measure it for you.

    OTH, there is the issue of dielectric absorption, which affects the
    observed value of capacitors at lower frequencies. If the polarizing
    voltage is present long enough, "remnant polarization trapped on
    dielectric interfaces" adds to the capacitor's value, see AoE pages
    220-221. The hp 4274A 5.5-digit LCR meter measures from 100Hz to
    100kHz, and is a good instrument to observe this effect.

    Both of these hp instruments appear on eBay from time to time, e.g.

    For frequencies below 100kHz, capacitance bridges are better suited
    for high resolution measurements. For example, the General Radio
    1615-A and 1616 bridges have 6 and 12-digit! readout, respectively.
    They can be used down to 10Hz, but the maximum bridge-excitation
    voltage must be reduced, lowering the achievable resolution. I have
    both of these fine, elegant instruments. :<>) A google search will
    turn up used-instrument dealers, and they're supposedly still being
    produced by IET,
    At least the old GR datasheets are available there for you to read.
    Search on 1620 and 1621 precision capacitance-measurement systems.
  4. I read in that Winfield Hill <[email protected]_rowland-
    Do you not find any self-resonance effects, e.g. with a 1 uF, well below
    10 MHz?
  5. John Woodgate wrote...
    Yes, of course, especially if the capacitor's leads are long enough,
    but this is easily corrected for. Actually, my hp 4191A RF impedance
    analyzer is better suited for complex-impedance measurements; It goes
    from 1 to 1000MHz, and has 4.5-digit readout resolution. I'm looking
    for a used 4192A (that I can afford) for more flexible measurements
    below 1MHz. And a 4194A would be very nice to have as well. <sigh>
  6. I read in that Winfield Hill <[email protected]_rowland-
    Ah, but what you find easy, the OP (and many others) would find a mite
    (indeed a 'might') difficult.
  7. Fred Bartoli

    Fred Bartoli Guest

    IIRC, the hp4275A is a bridge (more precisely an auto balanced bridge), and
    I really don't see what could prevent such bridges to give really high
    resolution measurements. I even see more reasons for an ABB to be more
    accurate than an "ordinary" bridge (note also that I don't know precisely
    how the Genrad are build, so... But I guess there's nothing very special
    about them. In case yes, I'd be curious if you had a schematics).

    For the ABB, as you might recall from some of my previous posts, I'm in the
    process of building a VNA that can also be used as an ABB (and also some
    other functions that are only pertinent to my use). I've made some small
    tests on those ugly solderless breadboards for the detector with my hp3456A
    for the measurement section and got really amazing results, given the
    ugliness level. I could easily measure the 0.13pf parasitics that's between
    two series of contacts separated by a guarding one for example, and also
    could easily resolve the fF with good stability.
    The expected linearity is on the ppm range and only depends on DC
    amplifiers, so can be improved at will (almost).
    The final product is expected to be largely as good.

    I really can't see what makes you saying this.
  8. shayan

    shayan Guest

    Hi Win. Tank you for your kindness and answering to my request.
    Indeed I'm a chemical engineer and I want to use a special capacitor
    for measuring the variation of dielectric constant in a fluidized bed.
    A fluidized bed is column in which air or another fluid is entering
    from bottom and fluidized the fine particles (5 to 500 micrometer)
    exist in the column.
    By variation in the particle concentration in the capacitor volume its
    dielectric also changed. If we could measure this changing in
    dielectric with a high sampling rate, we could calculate the particle
    concentration using convenient equations. But the traditional LCR's
    have not enough measurement speed (minimum 2 ms) and we want the
    sampling frequencies above 5 kHz (measurement speed below 0.2 ms).
    By which type of capacitance meters can I do this?
  9. Fred Bartoli

    Fred Bartoli Guest

    What is your measurement frequency range?
    What is your estimated capacitance range?
    What is you budget?

    Without those values, nobody can help you more than some vague
    Even orders or magnitude will help.
  10. Fred Bartoli wrote...
    In theory the 4275A is a bridge, but quite unlike a conventional bridge
    such as the GR instruments. The latter use ratio transformers, and all
    four arms of the bridge are apparent, including the reference capacitors.
    HP calls the 4275 circuit a "voltage-vector current measurement" method.
    A detailed look at the circuit schematic reveals many other components,
    although there is a balance resistor and null point. I haven't learned
    much about the nature of this balance resistor, which I imagine must not
    have any self capacitance. The schematics show five or six components
    making up each "resistor." I haven't torn my 4275A apart to examine it.
    I haven't been following your posts, what were some subject headings?

    As for the measurements, yes it's easy to get into the fF territory.
    The 4275A measures 0.01fF in its slower high-resolution mode, the same
    capability as the GR 1615-A bridge (the GR bridge has the benefit of
    allowing one to use an external variable-frequency oscillator). My
    own hybrid distance-meter design (ratio transformer + multiplier fine
    balance) had even better capability. But GR's 1616 was much better,
    with a 0.1aF most-sensitive measuring digit. They literally have a
    12-digit readout! I don't think that can be achieved without using
    precision low-impedance ratio transformers in an honest-to-god bridge.
    It would be nice if I understood your circuit.

    In the case of the 4275A, highly-linear RF phase-detector multipliers
    are necessary, which they implement with JFETs driven from an unusual
    staircase waveform circuit that suppresses the 3rd and 5th harmonics.
  11. Chris

    Chris Guest

    Hi, Shayan. Glad to hear from you again. Looking back on my previous
    post, I think I was a little impatient with your very brief problem
    description, and wasn't sure if it was a legitimate post. Sorry.

    But I believe I got the drift of where you're headed. I'm hearing you
    saying that you've got two plates suspended in a fluidized bed, and
    you're trying to find particle density by measuring capacitance to find
    dielectric constant of the fluidized powder between the plates. A
    couple of questions come to mind:

    * I'm going to assume you've got a fluidized bed with a powder being
    fluidized by air (again, your description of your problem leaves
    something to be desired -- what's the powder, what's the medium?). If
    so, you're depending on the difference between the dielectric constant
    of the air and that of the powder to provide you with something you can
    measure to infer powder density. Note that unless there's a lot of
    difference in the dielectric constant of the air and the powder
    ('tain't necessarily so), it might be that you are trying to infer more
    accuracy than you can get with that type of measurement. If the
    dielectric constants are close, the most accurate capacitance
    measurement isn't going to help, except in a very rough way. If
    there's a more dramatic difference in masses per given volume for
    powder and medium than their respective dielectric densities, might it
    be better to use a stirrer and measure torque?

    * Fluidization usually occurs with mechanical vibration of the bed.
    In a system with air bubbles, the mechanical shaking makes the air
    bubbles progressively smaller, until the powder/air mix acts like a
    liquid. I'm not sure why, in a mechanical system like a fluidized bed,
    5,000 measurements per second are desirable. Usually it takes from
    tenths of seconds to seconds for the powder and the meduim to
    homogenize. Again, a better description might help. This might be an
    esoteric application we don't know about.

    * This sounds like a fairly small (low pF range) capacitance. From a
    production/lab standpoint, if "real-time control" required such speeds,
    I might be tempted to cobble together a fairly high speed oscillator
    using the capacitance of the plates in the bed (possibly even
    suspending the rest of the oscillator in the bed itself), divide it
    down to an appropriate range with a comparator and some logic ICs, and
    after optocoupling, use a high speed counter board in a PC to collect
    and process period data. Either that, or you can use the divided down
    optocoupled signal to gate a counter, which would drive a D-to-A
    converter. You would infer capacitance from period or analog voltage,
    and then infer powder density from the inferred dielectric constant of
    the fluidized powder -- after all, your plate size remains the same,
    right? This has several advantages. And you'll have your 5KHz data

    I believe that, with reasonable care, you can get much better than 1%
    accuracy with this type of system, which might be enough if the
    differences in dielectric constants are great enough. It would require
    some cobbling, but might cost significantly less than your instrument,
    and might be a lot more reliable in a production/lab environment.

    By the way, you should watch for the development of static charge in
    the fluidized bed, and see if that's going to interfere with your
    measurements. You might be able to detect this by putting a high
    impedance voltmeter on your measurement plates with nothing attached,
    and monitoring to see if any charge is developed. Also, if you're
    using an air bed, you want to get a good handle on humidity issues,
    because that will seriously affect your measurements, as well as
    possibly affecting your process. Fluidized beds are a battle to get
    running well, and usually something of a mess. By the way, I'm sure it
    really isn't necessary to mention this, but from my experience, make
    sure everyone's read the MSDS and uses adequate breathing protection
    where necessary. Lab/machine operators sometimes don't follow even the
    best instructions.

    Thanks for posting again, and putting up with some unjustified
    impatience on my earlier post. I'd like to hear any additional details
    or project requirements.

  12. shayan wrote...
    I'm not aware of any that any that fast, although that's likely my
    own ignorance. I'm puzzled by your need for a 5kHz sampling rate,
    are you trying to measure turbulence dissipation or something, with
    a very small cell? You can make an LC oscillator that's running at
    several MHz (so its frequency changes within 200us, keeping a high
    resonance Q in mind), and take rapid measurements with a vernier-
    interpolation period counter. Period counting is normally limited
    in resolution, but the vernier-interpolation technique solves this
    problem for short measurements. See AoE pages 1020 to 1024, where
    we mention the hp5370B counter and describe how it works. According
    to the datasheet, the 5370B can take 8000 sustained measurements/sec.
    The 5370B went for $12,100 in my 1989 catalog, but I got one on eBay
    at a much lower price. Another counter with high-speed capability
    is the Stanford Research SR620, which can do 1400 measurements/sec. Costs about $5000.
  13. shayan

    shayan Guest

    Measurement frequency: minimum 5kHz

    Estimated capacitance: .001pF - 1mF

    Budget: maximum 5k$
  14. shayan wrote...
    Does this mean 5000 meas/sec, or does it mean a 5kHz
    ac measurement frequency?
  15. shayan

    shayan Guest

    We need the sampling frequency( =measurement frequency) above 5000
    sample per second

  16. shayan wrote...
  17. Winfield Hill wrote...
    An hp 5370B reciprocal interpolating counter is up on eBay, $850.
  18. Fred Bartoli

    Fred Bartoli Guest

    Yes, this resistance is a critical component (if you want some bandwidth)
    and this is a pb I haven't thought about yet. Thanks for having given the
    begining of a solution.

    Nothing very exciting I suppose. The post were about some detail points.
    Most are there, I guess :

    Hmm, not so sure. What you need is a sensitive vector measurement, and a
    null tuning mean. This will give you "coarse" nulling. Then you boos the
    detector gain to precisely measure the remaining unnulled component at the
    nulling node(s) and some math will give you the extra digits through ako
    precision virtual nulling. If you have a sensitive enough and linear enough
    detector, the figure can be astonishing and, I think, will compensate for
    not having standards to switch.
    I don't have the GR1616 manual but have found the 1615 one. The detector is
    (obviously) only for nulling, and all the sensitivity comes from the dynamic
    range given by the switched standards and the ratio transformer. Furthermore
    the vector measurement capability allow for noise averaging which the
    magnitude null detector can't provide.

    Another thing to consider is that the ratio transformer bridges are rather
    bandwidth limited.

    I'll email you about this, 'cause I can't disclose it.

    Do you really mean that they've implemented the detector at the RF level and
    not after an IF amplifier?

    Hmm, that could explain a lot of things and that's why I've prefered the
    other way (which is possible with todays components and was probably much
    more difficult when the hp4275 was designed). Another reason is that I
    certainly don't have the same ressources capabilities and I have to cheat a
    bit :)
  19. Fred Bartoli

    Fred Bartoli Guest

    That's not what I meant. I was speaking of possibly approaching those
    bridges performances (at least the easiest one) with the ABB by using the
    fact that I have a highly linear and stable vectorial detector.
    If I fell inclined and have some time when I have my two IF/detector boards
    debugged, I might try to push the limits and see where we go.
    That's not the detection that's causing problem, but rather phase stability
    of the IF amplifiers if you want really good accuracy. I've run into such
    problems and was very very happy to have really modern opamps to help
    solving this issue. I agree that doing the detection at the RF stage might
    have been tricky, but having followed the other route I can foresee the
    difficulties at the moment it was designed and I guess I understand their
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