HOW-TO GUIDE
How to Measure Signal Overshoot and Ringing With an Oscilloscope
By the OhmSavvy Research Team ·
Safety first: fast edges often live near high voltage
- On a mains-powered bench scope, the probe's ground clip or ground spring is connected to mains earth. Connect it only to the circuit's ground. Touching it to any other voltage shorts that point to earth.
- The switch nodes where ringing is most often measured, in switching supplies and motor drives, are frequently not earth-referenced, and on the primary side they sit at mains potential. Use a rated differential probe or an isolation transformer on the device under test, and never float the scope. See bandwidth for switching power supplies for the isolation options.
Overshoot vs ringing
When a fast edge drives a path that has some inductance and capacitance, the voltage does not stop neatly at its new level. It carries on past it (overshoot), swings back below it, and oscillates with shrinking amplitude until it settles (ringing). The two usually appear together, but they are measured differently: overshoot as a percentage of the step, ringing as a frequency and a settling time.
The three numbers
Why measure them at all? Overshoot pushes an input above its supply rail or below ground, and every chip's datasheet sets absolute maximum limits on that. Ringing on a clock or data line can swing back through the logic threshold and be counted as an extra edge. On a power switch node, the ringing peak, not the supply voltage, is what the transistor has to withstand. In each case you need a number to compare with a datasheet limit, not just an impression that the edge looks messy.
1Set up the probe and ground before you trust any ringing
Most ringing that beginners see is created by the measurement. Three settings decide whether what you see is real:
- Probe on 10x and compensated. The 1x setting has far less bandwidth and rounds off the very edge you want to measure. A miscompensated 10x probe adds its own overshoot; fix it with probe compensation.
- Short ground spring, not the long ground lead. The clip lead's inductance and the probe tip's capacitance form a resonant circuit that rings on every fast edge. Most probes come with a short spring that clips around the tip barrel; ground it right next to the point you are probing.
- Bandwidth limit off. The 20MHz limit many scopes offer is useful for ripple but hides fast ringing. Leave it off here.
2Trigger on the edge and zoom in
Use edge triggering on the edge you want to study, rising or falling, with the level near the middle of the swing. Move the trigger point towards the left of the screen with horizontal position so the edge and everything after it is visible, then shorten the timebase until a few ring cycles fill the screen. Set V/div so the full step uses most of the screen height with room above it for the overshoot peak. If the trace is unstable, see how to trigger an oscilloscope.
3Read three levels and calculate overshoot %
You need the initial level (before the edge), the final level (where the signal settles) and the peak (the highest point just after a rising edge). Read the final level well after the ringing has died, not on a ring peak or trough.
Peak = 5.75 V, final = 5 V, initial = 0 V, so overshoot = (5.75 − 5) ÷ (5 − 0) × 100 = 15%. The same formula works on a falling edge, where the overshoot is the dip below the final low level (often called undershoot): both the top and bottom of the fraction turn negative and the result stays positive.
| Edge | Initial | Final | Peak | Overshoot |
|---|---|---|---|---|
| Rising, 5 V logic (figure) | 0 V | 5 V | 5.75 V | 0.75 ÷ 5 = 15% |
| Rising, 3.3 V logic | 0 V | 3.3 V | 3.8 V | 0.5 ÷ 3.3 ≈ 15.2% |
| Rising, 12 V gate drive | 0 V | 12 V | 13.2 V | 1.2 ÷ 12 = 10% |
| Falling, 5 V logic | 5 V | 0 V | −0.4 V | −0.4 ÷ −5 = 8% |
Many scopes offer overshoot and preshoot as automatic measurements, using their own top and base levels rather than your readings, so small differences from a hand calculation are normal. Check your manual for how your model defines them.
4Time the ring cycles for the ringing frequency
Measure from one ring peak to the next. With only a few cycles visible, measuring across two or more and dividing is more accurate than reading a single one.
The ringing frequency is a clue to its source. It is the resonant frequency of whatever inductance and capacitance are involved, f = 1 ÷ (2π√(LC)), so if it changes when you change the probe's ground connection, the probe is part of the resonant circuit.
5Read the settling time
Settling time is measured from the edge (often from its 50% crossing) to the moment the signal enters a tolerance band around the final level and stays there. The band is your choice or the datasheet's, commonly ±2% or ±5% of the step. Definitions vary, so state the band with the number.
In the example, ±2% of a 5 V step is ±0.1 V. The trough 1.5 divisions after the edge still dips about 0.12 V below 5 V, while the peak at 2 divisions is only about 0.07 V above it, so the signal settles within ±2% in under 2 divisions, less than 100 ns. Voltage cursors set at 4.9 V and 5.1 V make this easy to see.
Is the ringing real or a probe artifact?
Real ringing comes from the circuit: long traces or cables acting as unterminated transmission lines, wiring and package inductance, very fast drivers, and the parasitic inductance and capacitance at a power switch node. Artifacts come from the measurement. The test is to change the measurement and see whether the ringing changes.
The shape is a clue as well. A miscompensated probe produces a spike or a rounded corner that decays smoothly over a large part of the period, with no oscillation. A long ground lead produces a true damped oscillation at a fixed frequency on every fast edge, and that frequency shifts when you move or coil the lead. Real circuit ringing stays the same however you probe it, as long as you probe the same point with a short ground.
01
Compensate
Trim the 10x probe on the cal output
02
Ground spring
Swap the long lead for a short spring
03
Full bandwidth
Bandwidth limit off, probe on 10x
04
Compare
Same reading at a nearby ground? Then real
| Cause | What it looks like | Check or fix |
|---|---|---|
| Long probe ground lead | Damped oscillation on every fast edge, changes when the lead is moved | Use the short ground spring |
| Overcompensated 10x probe | Spike that decays smoothly, no oscillation | Adjust the probe trimmer |
| Probe on 1x | Edges slowed, overshoot hidden | Switch the probe and channel to 10x |
| Bandwidth limit on | Ringing smoothed or missing | Turn the limit off for edge work |
| Real circuit ringing | Unchanged by the probe fixes above | Termination, layout or damping in the circuit |
How much bandwidth you need to see ringing
To show ringing at its true size, the scope and probe need bandwidth comfortably above the ringing frequency. The common 3-5x rule of thumb, explained in what oscilloscope bandwidth means, turns the 20MHz ringing above into 60–100MHz. At the scope's rated bandwidth a sine reads about 70.7% of its real amplitude, so a scope near its limit understates both ringing and overshoot.
| Ring period | Ringing frequency (1 ÷ period) | Bandwidth at 3-5x |
|---|---|---|
| 200 ns | 5MHz | 15–25MHz |
| 50 ns | 20MHz | 60–100MHz |
| 10 ns | 100MHz | 300–500MHz |
If the ring period you measure is close to the scope's own rise time (about 0.35 ÷ bandwidth), treat the reading as a lower limit: the real ringing may be faster and larger. For the edge-speed side of this, see how to measure a square wave.
Which scope you need for edge measurements
Overshoot and ringing measurements are bandwidth-hungry, so this is bench-scope work. A 50MHz scope covers ringing in the low MHz range; a 100MHz scope covers more. Compare them in our best beginner oscilloscopes roundup.
Bench scopes in our reviews
100MHz, 4 channels, 1 GSa/s; more bandwidth than the Rigol, for projects that will get more demanding.
100MHz, 2 channels, 1 GSa/s; an oscilloscope and waveform generator in one unit.
50MHz, 4 channels, 1 GSa/s; the default beginner recommendation for years.
As an Amazon Associate I earn from qualifying purchases. Prices and availability are shown on Amazon.
Frequently asked questions
How much overshoot is acceptable on a digital signal?
There is no single number. What matters is whether the peaks stay inside the receiving chip's absolute maximum input ratings and whether the ringing crosses back through its logic thresholds, which can cause false clocking. Both limits are in the datasheets of the parts involved. Overshoot that is safe on one logic family can be a problem on another.
Does a higher bandwidth oscilloscope show more ringing?
It can. A low-bandwidth scope, or one with its bandwidth limit switched on, filters out fast ringing and smooths overshoot, so the edge looks cleaner than it is. A wider-bandwidth scope reveals what is really there, but it also shows more probe-related ringing if the ground lead is long, so the probe setup matters more as bandwidth rises.
Can a 10x probe cause overshoot?
Yes. An overcompensated 10x probe adds a spike at every edge that decays smoothly without oscillating, and it can look like circuit overshoot. Adjusting the trimmer on the scope's calibration output removes it. A long ground lead causes a different artifact, a true damped oscillation, which a short ground spring usually removes.
What is preshoot on an oscilloscope?
Preshoot is a small excursion in the opposite direction just before an edge starts, so a rising edge dips slightly first. Many scopes offer it as an automatic measurement alongside overshoot, expressed as a percentage of the amplitude. It is calculated the same way as overshoot but before the edge, and definitions vary by brand, so check your manual.
Is overshoot on a power supply's output measured the same way?
The arithmetic is the same: peak excursion divided by the step size. The setup differs, though. You apply a load step, trigger on the change in output voltage and use a much slower timebase, because a regulator's control loop responds over microseconds to milliseconds. Large or slow-settling overshoot there usually points to low loop phase margin rather than to traces or cables.
