Tank you all for you 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?
What you want is a capacitance bridge - these are used in stretched-diaphragm capacitance pressure gauges, amongst many other application.
I'd look at a Blumlein tranformer bridge myself, which would use a bifilar-wound centre-tapped winding on a ferrite core transformer. One end of the winding would drive one side of your sensing capacitor (the other side would be grounded) and the other end would drive one end of a stable reference capacitor of the same nominal capacitance (whose other end would also be grounded, ideally to the same point as the grounded side of your sensing capacitor).
The centre-tap is then your unbalance source - hook it up to the input of a high-input impedance amplifier, and demodulate at the excitation frequency.
Tony Williams wound a nice transformer for a fast resistance bridge (15kHz excitation frequency IIRR) which you might find interesting.
Higher frequencies are certainly practical, but you start having to worry about phase shifts in the bridge-driving circuit and the amplifier.
Analog Devices now make synchronisable DDS circuits with quadrature outputs which make it very simple to produce an array of excitation and in-phase and quadrature demodulation waveforms. It would be a fun project to put such a system together, but what you probably need are the electronics from an MKS Baratron capacitance manometer
formatting link
I don't know if the exitation frequency is high enough for what you want, but the system goes back to a time when most such gauges used a
455kHz excitation frequency (the IF frequency used in AM radios of that period), so you should be okay.
If you need high speed sampling, then you may need to rig up a relatively simple circuit, but beware - there's more to capacitance than meets the eye. One of the reasons the standard meters are slow is they are designed to be to make sure their readings are accurate. Long story and way beyond the scope of the question.
First the basics. The capacitance of a plate style capacitor is directly proportional to the dielectric constant.
That being so, it's quite simple to feed a signal (we'll get to frequency) through a series resitor and measure the phase shift. As the phase is predictable [given by arctan X/R, or alternatatively, arctan V(c)/V(r)]. Phase shift can be measured in any number of ways. The simplest I know of is simply to measure the voltages (suitably amplified so they are reasonable measurable, becaues you don't want to impose a big signal for your measurement - I'll get to that) with an add-in A-D card using a dual half-flash converter, which should be adequate for your needs. (Full flash converters are expensive). Alternatively, you could use a card with a successive approximation converter, but it would have to be fast. Having done the measurement in a PC of some description, the software to do the calculation is trivial.
Before you get there, you should do an rms conversion so you have a stable signal for conversion (getting an idea why LCR meters are slow?).
To do that, with a sampling frequency of about 5kHz, you'll need to excite the circuit with about 50kHz or so, which gets us into other practical problems with capacitors. To work around them, choose a resistor that has about the same resistance as the reactance of your capacitance at 50kHz at some midpoint condition. (The phase at this point is 45%). This implies, of course, that you have some idea of the capacitance at this point. Another important point is the range of capacitance you are speaking of.
[Two major issues that arise at frequencies well below this are leakage and effective series resistance/inductance - so it could impact you.]
RMS conversions are fairly straightforward, especially if you have a pure sinusoid driving the circuit.
The reason I chose 50kHz is so the RMS converter has the bandwidth to respond to changes at a 5kHz rate. 10:1 is a good rule of thumb for such things.
As you might note, it's not that simple - the fastest 'standard' meter I know of converts at a rate of 800 samples / sec and it's very expensive.
Another possible circuit is much simpler, but depending on the capacitance in question may be slow.
Make a circuit that has a single capacitor in series with your effective capacitance in the tank. Apply a DC voltage. The voltage across each will vary as the capacitance of the tank varies (specifically, it acts as a voltage divider). The charge stored in each capacitor is equal (and equal to total circuit charge), and each capacitor develops a voltage given by V=Q/C. Note that as the capacitance of your tank varies, the total circuit charge will vary, but the ratio of voltages will accurately reflect the ratio of capacitance.
The big problem here is that to do a measurement, you need some current from the measured object, which can destroy the measurement (because it is taking the charge from the circuit). There are ways of dealing with this, using ultra high impedance inputs and input current cancellation (both easily available and fairly straightforward - input current cancellation is also straightforward to design but does require certain skills in feedback loops).
So if you have someone who knows how to do that, then you can use the second method.
So your solution is either to use a very expensive (about $5k - $10k) lab meter, or get one of the electrical guys to design and build you a circuit that you can then feed to a high speed measurement system.
That would be well and good if it were not for the fact that particle density to dielectric constant relationship is for static steady state. This may not hold against serious deviation with unknown displacement current circulating due to dD/dt, and probably makes the OP's entire approach fundamentally flawed.
Maybe you could measure the current the blower draws, correlate this to a mass flowrate (m-dot), then determine the concentration that way? Might also need to measure the gas / fluidized particle flowrate (v-dot) with a flowmeter... m-dot = c x v-dot, solve for c?
In theory this could be a real problem, but amongst the capacitance pressure gauges that I read about back in the 1960's there was one that used electrostatic force-balancing to null out the pressure difference across the sensing diaphragm, which is a very cute idea, but perfectly useless for measuring significant pressures, because electric field required to compensate for any pressure much higher than a "chemical" vacuum (one that a phsysicist would sneer at) was high enough to initiate a discharge.
The aerodynamic forces acting on the particles in a fluidised bed are going to be high enough that your measuring signal isn't going to move them around to any significant extent. I'd be more worried about the shot noise on the signal arising in the fact that you are going to have a finite number of particles in your capacitance cell at any one time.
--- I suggest that instead of a capacitance meter, per se, you fabricate the capacitor with the view of using it as the tuning element of an oscillator, and then use a gated counter to count (integrate) the oscillator cycles over a fixed time. Then, by selecting the nominal oscillator frequency you can get whatever resolution you need over that period for the expected change in dielectric constant.
Probably the thing which will introduce the most error will be the dimensional change of the capacitor with temperature, so that will need to be dealt with. One option would be to make the capacitor and its supporting structure out of materials with a very small thermal coefficient of expansion, and another would be to measure the change in capacitance with temperature using a dielectric with known characteristics over that temperature range, then to generate a lookup table which would be used to compensate for the measured in situ temperature when the device was in use.
If you'd be interested in having us design and build a system for you, email me and I'll be happy to respond with a quote.
A third option is to design your capacitance sensor to include a second, closed reference capacitor, whose expansion with temperature will track that of the sensing capacitor - this makes even more sense if you measure the capacitance of the sensing capacitor against the capacitance of the reference capacitor with a capacitance bridge, rather than the using a oscillator circuit to measure the product of the sensing capacitance with some arbitrary tuning inductance which is likely to be just as temperature dependent.
This is one of the reasons why I suggested using a capacitance bridge in my earlier post - like John Fields, I too am under-employed and in a position to act as a consultant .....
--
One of the problems with using a capacitance bridge is that the OP
would like to make 5000 measurements per second, another is that the
two caps _must_ be isothermal. Using option 1, which is a capacitor
with zero or close to zero tempco sidesteps the entire issue and
allows a simple counter to make the measurement. Of course some
consideration must be given to the inductor, but since it's not
required to be in the bath with the cap, it could easily be ovenized
with the rest of the oscillator, away (but not _too_ far away!) from
the cap.
Of course, the whole thing may be moot depending on the range and
accuracy the OP needs but which, unfortunately, along with the
temperature range the thing would be expected to work under, he didn't
state.
There's no obvious reason why a bridge can't be used to make 5000 measurements a second - it is lot easier if you keep the bridge excitation frequency high, which you want to anyway with most capacitance sensors to keep the impedance of the sensor manageably low.
There's nothing difficult about demodulating and digitising the out-of-balance signal fast enough to get 5k independent samples per second.
Making the balancing capacitor and the sensing capacitor isothermal ought not to be that difficult if you make them part of the same mechanical structure - probably easier than making a capacitor with a near zero temperature coefficient.
It's not a bath, but a fluidised bed. I do like the glib way you say "ovenised" - maintaining a stable and uniform temperature inside an oven isn't exactly a sinecure, and gets trickier as your oven gets bigger (crystal ovens are tiny).
didn't
He certainly didn't. Some fluidised beds run quite hot.
--
Perhaps not, but for this application it sounds kind of iffy to me.
If we had more information about the application it would help, but in
lieu of that I'd prefer to stick with the oscillator.
Okay - you've only used bridges, as opposed to designing them.
low.
10pF of capacitance at 15kHz is about 1Mohm.
You are offering to do this for money - I'm not doing your job for you.
a
I've had to hunt for funny low-coefficient-of-expansion alloys in my time, which is what your scheme seems to call for. The sensing cell isn't goig to be an off-the-shelf item in any event, and the amount of extra thinking required to design in a closed reference capacitance doesn't strike me as being in the same league.
ovenized
from
The fluid is usually a gas.
Even without the closed reference capacitance, the sensing capacitor isn't going to be duck soup - the guy is going to go through a couple of prototypes to get rid of all the problems that weren't obvious at the brain-storming session.
Human fingers are still much the same size. And you probably don't want a ferrite-cored chip inductor - if I had to, I'd go for something built on an RM core, with a large air-gap at the inner mating surface, so I wouldn't have to worry about the temperature dependence of the permeability of the ferrite but would get the advantage of the wrap-around core as a shield
--
On the contrary, I've designed RTD bridges which were temperature
compensated and worked from the surface of the ocean to the bottom of
the sea of Japan, differential capacitive proximity sensors capable
of determining parallelism to within a couple of arc-minutes, as I
recall, and an interesting self-heated thermistor mass flowmeter, not
to mention the garden-variety stuff that comes up every day.
BTW, my hp4280A 1MHz C-V, C-t meter measures down to 0.001pF and can take bursts of high-resolution measurements in 10us. This capability is for Zerbst analysis, used to determine minority-carrier lifetime and surface-generation velocity. Perhaps the O.P. can purchase one of these instruments.
So you should know that if your excitation frequency is high, a fast demodulator is trivial.
manageably
you
per
you.
There are a couple of published papers in the U.K. journal "Measurement Science and Technology" which do include circuit diagrams - sadly, no demodulators, so you'll have to crib from someplace else.
Since you are American, you won't ever have heard of it - it is the British equivalent of the Review of Scientific Instruments, which you ought to have heard of. So far I've only published comments there - their referees on't seem to know much about electronics.
isothermal
with
of
In fact I've been thinking about that problem, and I suspect that the OP is going to have use a planar capacitor, measuring the capacitance between inter-digitated metal fingers printed onto a insulating surface. This leaves you stuck with a large-ish fixed capacitance between the fingers, associated with the field lines in the insulating support, but the field lines above the surface should buy you a component which will vary with the particle loading in the fluidised bed.
The obvious way of building such a capacitor is as a thick- or thin-film hybrid circuit on an alumina substrate. You'd want to make the other side of the substrate a solid ground plane, which would minimise the capacitance associated with the field in the alumina.
You balancing capacitor is then a second identical capacitor soldered back-to-back onto the first, with a dust-tight box to keep the fluidised particles out of range - you'd probably put a porous plug of sintered metal or ceramic in the wall of the box to allow slowish equalisation of gas pressure.
Making all that iso-thermal isn't exactly monsterously difficult - the alumina has a much higher thermal conductivity than the fluidising gas.
No schematic, but in the fast resistance bridge that Bill referred to earlier in the thread we managed an analogue freq response of about 2.4KHz, ADC'd at 10000 conversions per second.
15KHz bridge excitation frequency. V-bridge and V-unbalance picked off and AC amplified in similar channels. V-bridge used as the clock for the synchronous demodulation of both V-bridge and V-unbalance. 250KHz clock for a pair of Maxim digital low pass filters. Final dc signals into an ADC, where the demodulated V-bridge was used as the Vref. The demodulated V-bridge was also used to servo the amplitude of the excitation oscillator. The 250KHz clock also generated the (synchronised) 10KHz CONVERT timings to the ADC.
I suspect the bum-biter with a capacitive bridge would be that the excitation frequency *has* to be chosen to get a reasonable current in the bridge without requiring unreasonable voltages. Can't move anywhere else in the design until those numbers are sorted.... and that depends on the range of capacitance values to be measured.
None of the papers includes a 555, so you'd be out of your depth.
you
it,
Nothing political about that - American scientists and engineers can get away with ignoring the rest of the world, and often do. A couple of my published comments in the Review of Scientific Instruments point to papers in non-American journals which authors and referees should have been aware of.
An understandable attitude - but one that won't do anything to fill the lamentable gaps in your knowledge of electronics.
Those days are rapidly coming to an end. The US share of published scientific papers peaked in 1992 and has been declining by 10% per annum since. Looks like the heyday of this grossly mismanaged mess is nearly over with a recent survey showing an inflection point has finally occurred in the traditional measures of rate of technological breakthrough. I notice a lot of selfish people seem to be taking personal comfort in some sort of belief in economic inertia attenuating the severity of impact, but that type of moron is just not paying attention to the established rapidity of high technology industry growth and decline among winners and losers.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.