probe loading
Probe loading is the uncomfortable truth that the act of measuring changes the thing you measure. Picture checking the temperature of a small cup of soup with a big cold thermometer: the thermometer itself cools the soup, so you read a temperature that the soup only has because you stuck a thermometer in it. A meter or scope probe does the same thing electrically. The instant you touch a probe to a node, the probe's own resistance and capacitance become part of the circuit, and a sensitive node can be nudged away from the value it had a moment before.
It works through two paths. First, resistance: every voltmeter and scope input has a finite input resistance that quietly sits in parallel with whatever it touches. Put a 1 MΩ scope input on a node fed from a 1 MΩ source and the two form a divider that halves the reading — you measure 0.5 of the real voltage and might conclude the circuit is broken when it is fine. A x10 probe makes the input 10 MΩ, so the same node sags by under 10 percent. Second, capacitance: even a few picofarads of probe capacitance forms a low-pass with the source resistance, rounding fast edges and faking a slower signal. At high frequency this capacitive loading often matters more than the resistance.
Why it matters: the nodes most worth probing — high-impedance op-amp inputs, oscillator tanks, sensor outputs, gate drives — are exactly the ones most easily disturbed, and a circuit that works until you probe it (or breaks only while you probe it) is the classic symptom. The defenses are simple and worth habit: prefer a x10 probe to cut both the resistive and capacitive load; keep the ground lead short to limit added inductance; and when a reading looks wrong, ask honestly whether your own probe is the cause before you blame the design.
A high-impedance node is driven through a 100 kΩ resistor. Touch a x1 probe (1 MΩ) and the divider 1 M/(100 k + 1 M) drops the reading to about 91 percent of true — a 9 percent error from the probe alone. Switch to x10 (10 MΩ) and the same node reads about 99 percent of true. Same circuit, honest reading, just a better probe.
The probe joins the circuit — resistance halves slow signals, capacitance rounds fast ones.
If a circuit works until you connect your probe, suspect the probe before the design — you may be loading a delicate node. This is most common at high-impedance points and at high frequency, where even a few picofarads matter.