Connect a probe to nothing, or to a circuit that is switched off, and the scope shows a wave at 50 or 60 Hz. It is not a fault. It is the mains wiring around you, picked up by an input built to draw almost no current. Here is why, a five-minute experiment that shows it, and how to tell it from a real signal.
The wiring in walls, benches and power cords carries 120 or 230 V alternating at 60 or 50 Hz, and it surrounds itself with an alternating electric field. Any conductor nearby, your probe tip included, is coupled to it through a tiny capacitance. A capacitor passes alternating current, and the smaller it is the less it passes, but an oscilloscope input needs very little: it is high impedance, 1 MΩ on its own and 10 MΩ through a 10× probe, so even nanoamps make a visible voltage.
An example with round numbers (the real coupling varies with where everything sits, so take it as an order of magnitude, not a prediction). Say the tip couples to a 120 V wire through 1 pF. At 60 Hz, 1 pF has an impedance of 1 / (2π × 60 Hz × 1 pF) ≈ 2.65 GΩ. The current is 120 V / 2.65 GΩ ≈ 45 nA. Through a 10× probe's 10 MΩ that makes 45 nA × 10 MΩ ≈ 0.45 V rms, about 1.3 V peak to peak; through 1 MΩ, a tenth of that. Our unpowered XBee pin showed 0.47 V peak to peak: well within that range.
Now let a powered circuit drive the same node. A logic output holds its pin through tens of ohms, not megohms, and the same 45 nA makes a few microvolts: invisible. That is why hum shows on floating nodes and not on working ones.
The probe's ground lead adds a second path. Tip, ground lead and circuit form a loop, and a changing magnetic field (from transformers, motors, the mains cords themselves) induces a voltage around any loop, in proportion to its area. A long ground lead flapping in the air makes a big loop; a short ground spring makes a tiny one.
And a circuit powered from a two-pin charger (no earth pin) can itself float at a mains-frequency voltage relative to the scope's earthed ground until the ground clip connects the two: the charger's suppression capacitor couples a little of the mains onto its output. Clip the ground before you judge the signal.
| 1× | 10× | |
|---|---|---|
| Load on the circuit | 1 MΩ, and typically 50 to 100 pF or more (the cable) | 10 MΩ, typically 10 to 20 pF |
| Bandwidth | often only a few MHz to a few tens of MHz | the scope's full bandwidth, for a matched probe |
| Hum on a floating node | less (it loads the node ten times harder) | more (it loads the node less) |
| Small signals | better: nothing is divided down | divided by 10 before the input, so more noise relative to the signal |
A 10× probe showing more hum than a 1× on nothing is not a worse probe: it is a lighter load, which is what you want on a real circuit. Use 10× for almost everything; switch to 1× only for millivolt signals at low frequencies. Your probe's own data sheet gives its real figures.
The scope's ground clip is connected to mains earth through the scope's power cord. Clip it only to the circuit's ground, never to a point at another voltage, and never probe mains circuits with an ordinary probe.
No. The hum on a floating probe is a few nanoamps through gigohms of coupling: harmless, and only visible because the scope's input draws almost nothing. It tells you the node is floating, not that it is live.
No. AC coupling blocks DC and only the slowest changes, below a few hertz. 50 and 60 Hz pass straight through.
No. A bandwidth limit (often 20 MHz) cuts high frequencies; 60 Hz is far below it. Driving the node, or grounding the probe properly, is the fix.
A frequency measured over a few cycles on a slow time base is only as precise as the samples at its edges. The mains itself is held within a small fraction of a hertz of 60 (or 50) Hz; average more cycles and the reading closes in on it.
Because it loads the floating node with 10 MΩ instead of 1 MΩ, so the same tiny coupled current makes ten times the voltage. On a node a circuit drives, both show the real signal.
Why the probes must come off the inputs, how a probe on a dead circuit still carries a signal, and when to run it.
Why the scope waits for a triggerAuto, Normal and Single; slope and level; and why a level that looks right still catches nothing on an idle line.
Decode a serial (UART) signalAn XBee radio's serial line, from the wiring to the decoded answer: the trigger on the start bit, the time base and the decoder settings.