Electronics > Projects, Designs, and Technical Stuff
where does 1/2 come from in capacitor energy calculation
<< < (3/4) > >>
tszaboo:

--- Quote from: StillTrying on July 13, 2018, 04:08:52 pm ---As a SMPS spends all of its time charging and discharging the output capacitor, how can its energy efficiency ever be more than 50%. :)

--- End quote ---
I give you one better: How a switching capacitor voltage converter can work with 50%+ efficiency, while it just charges capacitors? Sometimes even 3.
Siwastaja:

--- Quote from: woodchips on July 13, 2018, 04:03:15 pm ---Umm, no.

You have two caps, each 1F 10V, one fully charged, one not. Energy 1/2CV^2 is 1/2x1x10^2 = 50J.

Connect in parallel, now have two caps total of 2F at 5V, Energy 1/2CV^2 is 1/2x2x5^2 = 25J.

To me, 25 is half of 50?

As far as I am aware, this halving always happens irrespective of how the cap is charged. Something to do with the fact that the cap charge is stored on the surface of the plates, not electrochemically as in a battery.

--- End quote ---

It's unbelievable how this misconception sticks and lives its own life. It has been discussed to death.

Short answer: No. Nope. It's not like that. Capacitors are and can be very efficient, easily 99%. Charging circuits can be 98-99% efficient as well, easily 90%. This is why capacitors are and have been used in energy storage for example, regenerative braking in train systems.

Charging a cap from a voltage source through a resistor (implicit or parasitic), or using a simple linear supply, of course wastes at least 50% of the energy, since the average voltage during the charging over that resistor is half of the source voltage. This is true with charging a cap, or driving any other load using a linear or resistor-dropper supply to drop half of the voltage.

An SMPS can adjust its output voltage and keep high efficiency over the operating range.

Discretize the problem into ten timesteps and you'll figure it out easily. At the first step, the receiving capacitor is at 0V and the full 10V voltage is over the resistance. You'll see instantly where power is dissipated and by how much. Now replace the resistor circuit with an SMPS that can directly provide you with 1V, instead of 10V. Then set it for 2V for the next timestep, etc...
intabits:
I was curious about this apparent paradox, and thought I had it worked out with this simulation, but now I'm a bit confused.
First, my explanation:- 

Using 15 ohms to exaggerate the parasitic resistances of the wires and capacitors.


S1 closes for 1 second between T=1 and T=2, and C1 charges via RC, and reaches the full 10v by T=2.
At T=3, S2 closes to transfer the charge between the caps. C1 discharges and C2 charges, reaching equilibrium at 5v.

The two spikes at T=1 and T=3 are the power dissipated in RC and RD, and the areas beneath them is the energy lost by these resistors.
It can be seen that the second spike's area, representing the missing energy,  is around half of the first. Any minor discrepancy would presumably be due to the internal resistance of the caps.

But now I realize that while the energy of the second spike can only have come from C1, I had assumed the area of the first spike represented the energy going into C1 to start with.
But now I can't see why the energy lost out of RC would at all have to be the same as the energy going into C1???.
(the large RX was needed to force C1 to initially be 0v. Not sure why, maybe some quirk of the voltage controlled switches?)
EEVblog:

--- Quote from: blueskull on July 14, 2018, 07:11:50 am ---Therefore, the other half of energy went to resistor, no matter how high the resistance is or how low it is.

--- End quote ---

Well, actually, when the resistance approaches zero, the current approaches infinite, and the energy starts to get lost as electromagnetic radiation at that point.
T3sl4co1l:

--- Quote from: blueskull on July 14, 2018, 08:18:17 am ---
--- Quote from: EEVblog on July 14, 2018, 08:06:19 am ---Well, actually, when the resistance approaches zero, the current approaches infinite, and the energy starts to get lost as electromagnetic radiation at that point.

--- End quote ---

Actually that can have better efficiency as stray inductance start to soft charge ("soft" comparing with resistance). The above analysis only assumes RC circuit without inductor.

--- End quote ---

1. There is ALWAYS loss.  Even a superconducting resonator (a more subtle kind of inductor-capacitor circuit) has a finite Q (around 10^8 -- better than most quartz crystals, giving a time constant of ~seconds at a resonant frequency of 100s MHz, not bad eh?).  Over a short time period, we can get reasonably lossless exchanges of energy, but over a long time scale, thermodynamics ALWAYS wins.

The equations for charge conservation, capacitor charging (RC or otherwise), and so on, are taken at \$ t \rightarrow \infty \$, so thermodynamics is the name of the game here. :)

2. There is no circuit without inductance, or capacitance, either.  Indeed, L and C are the electrical manifestations of physical causality: anywhere the physical length of some system is nonzero, and where the speed of light* is finite, there must be components of L and C!

*Generally, the velocity of propagation.  Equivalents can be derived in any wave system: e.g., in mechanics, the speed of sound causes finite and nonzero elastic modulus and mass.

So this teaches us a fallacy of our schematic notation: that all signals are instantaneous, or that the circuit has zero dimension, or the speed of light is infinite (all equivalent).  We must think carefully to ensure we aren't deceiving ourselves with an overly greedy simplification, and draw circuits that are physically realistic. :)

This is most applicable to SPICE modeling, where it is easy to create perfectly valid, but physically impossible, systems.

We're basically SPICEing this circuit by eye.

So there must, necessarily, be some L in the real circuit this represents.  And that L (and the C as well) will couple into ambient fields, however slight (there is no perfect shielding, either).  So that, even with a truly perfect resonator (better than any superconducting one we can actually make), the energy inside will inevitably leak into free space.  At infinite time, the final state will be the steady state given exactly by thermodynamics. :)

Tim
Navigation
Message Index
Next page
Previous page
There was an error while thanking
Thanking...

Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod