Yes, certainly the capacitor is only as important as the surrounding circuitry allows it to be!
As an example, suppose you had an ideal constant current coming down a wire. (So, you normally keep it grounded so it doesn't waste power, and you handle it very carefully so it doesn't start a spark -- because it will spark and arc as far as you care to open the circuit! Ideal constant currents are fun.
) Suppose you have naught but your hands and some means of keeping time (a pendulum, a metronome, a stopwatch... or just counting mississippis out loud).
If you connect the current source to the capacitor, start the timer, and disconnect it after a given time, you have delivered a known amount of charge to the capacitor. You can now measure the charge, say by connecting an electrometer. The foil electrodes deflect in front of a scale, allowing you to read a measurement (in effect, by calling out divisions of a scale, you're performing an A-D conversion manually, which can then be written down as a finite series of digits).
You could also keep the electrometer connected, while charging the capacitor, and stop the timer when it reaches a certain charge. Your main source of reading error is now determining how many pendulum swings occurred, or how many mississippis were spoken, or whatever.
In both cases, the electrometer scale must be calibrated, and you have two sources of errors: the ability to count time, and the ability to read a scale consistently (and its calibration).
You can combine these methods. Perhaps you have a known current sink, so after charging the capacitor for some time, you discharge it until it reads zero again. Compare the ratio of times, and you know the ratio of currents. Now you no longer need to know what the delivered charge was; the electrometer scale doesn't need to have any calibration, or even any linearity, so long as it remains a one-to-one function of voltage, so you can tell when it passes the point it was initially at.
You now need a calibrated current source instead of the electrometer scale, but the idea is to swap out sources of error until you have something easily calibrated and stable. An electrical current is easy to measure, set and calibrate, and has other advantages inside an integrated circuit. So this is good progress. The time measurement is still a matter, but it is actually one of the easiest and most accurate measurements that can be performed electronically.
Perhaps, when you stop the discharge (with the known current sink), you accidentally went too far: you counted one full mississippi extra. Your time is well known (you counted a whole number, no rounding or guessing at fractions), but now the discharge is too far. You could measure the residual with the electrometer again, but then its scale must be calibrated; and in finer divisions than before!
Suppose you connect another known accurate current source, of magnitude 1/N times the discharging one. Then you count until the charge comes back to (or just past) zero: now you have a triple slope integrating ADC, and the counts of that last step are exactly 1/N the weight of the discharge counts. So if you count 0-9 steps, and N=10, you go from, say, a measurement of X, to a measurement of X.Y, fully 10 times more precise. If the current ratio N can be generated geometrically, it won't need additional calibration, another big win.
In the electronic implementation, ultimately, you have transistors (usually MOSFETs, since we're talking battery operated, most likely CMOS circuitry here), which are required to sense that the voltage has reached some value, or that it's crossed through zero, or whatever. Whether you want to put the emphasis on one thing or another (a transistor, a capacitor, a resistor..) is up to your own interpretation. Obviously, an instrument cannot exist unless there is both a sensing element (a transducer), and something to observe that activity, a sensor or amplifier or observer or whatever. Concentrating on any one component is like saying a human is nothing but a brain, or a heart, or a stomach; none of which can exist on its own without the help of many supporting parts!
Tim
Home
Help
Search
Login
Register
