Bypass caps and lead length
A schematic carries 1 µF, 100 nF and 1 nF side by side, and a capacitor with its legs left long is worth a fraction of the same part cropped short. Both of those are the same fact — the wires to a capacitor are part of the capacitor.
The last page said to put a 100 nF across the power pins of every chip and to keep it close. Close to what, and how close, and why does the schematic have three of them?
Pick a mounting and watch the bars at the bottom.
A lead is an inductor
Every millimetre of wire is worth roughly a nanohenry, and the loop the two legs enclose is worth more. That inductance is in series with the capacitor whether you drew it or not, and the two of them together are a series resonant circuit: below a certain frequency the capacitance dominates and the part behaves like a capacitor, above it the inductance dominates and the part behaves like a small inductor.
That crossover is the self-resonant frequency, and it is 1 ÷ (2π·√(L·C)). Above it, the capacitor you fitted to swallow a fast edge is in the way of it.
Nothing is wrong with the part. It is still the right value, it still measures right on a meter, and it has stopped doing the job.
This is what "next to the chip" means. It is not tidiness. Two centimetres of track between a decoupling capacitor and the pin it is decoupling is enough inductance to move its useful ceiling down by a factor of two or three — and the transients you fitted it for are the fast ones.
Which is why there are three of them
Look at the bars again. A 10 µF has enormous charge and an early ceiling. A 1 nF holds almost nothing and carries on into the hundreds of megahertz. Neither covers the range on its own, and you cannot buy a capacitor that does, because the ceiling comes from the mounting rather than from the value.
So a supply rail on a board that has been thought about gets several in parallel:
- 1 µF to 10 µF near where power enters, and next to anything with a motor or a relay in it. This is the bulk — it handles the slow, large sags.
- 100 nF at the supply pins of every IC. The one on every schematic.
- 1 nF to 10 nF alongside it on fast digital parts and anything with a radio. Small enough to still be a capacitor where the 100 nF has given up.
They are not redundant and they are not a belt-and-braces habit. Each one covers a band the others cannot reach.
What to do with your hands
- Solder ceramics flat to the board with the legs cropped. Standing one up on full-length legs, which is the natural thing to do on stripboard, costs you most of its range.
- Put it on the same side as the chip, between the supply pin and ground, not across the rail somewhere convenient.
- Keep the loop small. The area enclosed by the path out of the capacitor, through the chip and back is what sets the inductance. A ground plane collapses that loop, which is most of why four-layer boards behave better than two-layer ones.
- Use SMT if the board is SMT. An 0603 on its own pads has a fraction of the inductance of any through-hole part, which is why production boards look the way they do.
Next: 3.3 V against 5 V, and what happens when you connect a pin of one to a pin of the other.
Edit this page — content/fundamentals/electricity/bypass-and-lead-length.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.