Electricity/Capacitors/Inside a capacitor

Inside a capacitor

A 100 nF ceramic is smaller than a grain of rice and a 100 µF electrolytic is the size of a thumb. Both are two plates and a gap — the difference is what is in the gap, and it is the reason one of them explodes if you fit it backwards.

Capacitance is decided by three things and nothing else: how big the plates are, how close together they are, and what is between them.

Move each one and watch the number at the bottom.

Build the capacitance out of its three parts
X7R ceramic · εr 3000
Between the plates
Plate area22 cm²
Gap27 µm
Capacitance
2.1 µF
εr
3000
1 F would need
32 m square
2.1 µF, and the material did most of the work. A class-2 ceramic multiplies the same plate by about 3000, which is how a 100 nF ends up smaller than a grain of rice. It is not free: that number is a rough middle of a range, and it moves with temperature, with age, and with the voltage you put across the part — sometimes by half. C0G ceramic gives up most of the value and holds still instead, which is why timing circuits use it and decoupling does not.

C = ε₀ · εᵣ · A ÷ d. Area on top, gap underneath, and εᵣ is how much better than air the stuff in the middle is. You will never need to calculate that on a bench, but the shape of it answers every "why is this part that size" question in the chapter:

  • Bigger plates, more capacitance. Directly proportional, which is why the foil inside a big electrolytic is a strip the length of your arm.
  • Closer plates, more capacitance. Also directly proportional — halve the gap and you double the value. The gap is also what holds the voltage off, so the thinner it gets the lower the part's voltage rating. Capacitance and voltage rating trade against each other in every technology there is.
  • Better dielectric, more capacitance. Air is 1. A class-2 ceramic is somewhere around three thousand, which is most of the reason a 100 nF fits on a fingernail.

Three constructions, and everything else follows

What is actually inside one
Ceramic
Values you can buy
1 pF – 100 µF
Polarity
None
How it dies
Cracks. And it quietly loses value.
A few hundred printed metal layers, interleaved, fired into one block. That stack is why a 100 nF is 1.6 mm long and why it has no polarity — both plates are the same printed metal. Two things to know about it: a board that flexes cracks ceramic, which is why they go near the edge of a panel last; and a class-2 ceramic (X7R, X5R, Y5V) is worth noticeably less than its marking once there is a voltage across it. Fit a part rated well above the rail and that mostly stops mattering.

The three you will actually hold:

  • Ceramic. Hundreds of printed metal layers fired into one block. Small, no polarity, cheap, and fast — the one thing that keeps up with a digital chip's edges. Fit one across the power pins of every IC.
  • Electrolytic. Two aluminium foils rolled up with a wet paper separator. The insulator is an oxide layer only microns thick grown onto one foil, which is where the enormous values come from and why the part has a polarity at all.
  • Film. Metallised plastic, rolled and squashed into a box. Stable and accurate, physically large, and the one you want when the value has to be the value.

Two numbers on the body, and the second one is not optional

The value is the one everybody reads. The voltage rating is the one that gets skipped, and it is the one that fails.

Fit a 6.3 V part on a 5 V rail and you have almost no margin for the spike when a motor stops. The habit worth having is to fit at least twice the rail: a 16 V part on 5 V, a 25 V part on 12 V. Electrolytics are the parts that care most — they are also the ones that dry out with heat and lose value over years, which is why a fifteen-year-old board with a dead power supply usually just needs its electrolytics replaced.

There is one more caveat and it catches people out with a meter in their hand: a class-2 ceramic — X7R, X5R, Y5V, the letters printed on the reel rather than the part — is worth noticeably less than its marking once there is a voltage across it. Not a few percent. A 10 µF rated at 6.3 V can measure closer to half that on a 5 V rail, and the smaller the case size the worse it gets. Check the manufacturer's DC-bias curve if the value has to be right; fit a part rated well above the rail if it only has to be roughly right.

Next: put a resistor in front of one and it stops being instant.

Edit this page — content/fundamentals/electricity/inside-a-capacitor.mdx

Discuss this article

Ask about this page. The answer stays here, on the page it belongs to, for whoever hits the same wall next.

Browse Fundamentals on the forum