For a 100 MHz square-wave or clock, the industry-standard guidance is to pick an oscilloscope with bandwidth of 5 times the highest frequency component, which gives 500 MHz and keeps measured amplitude error under ±2% [S3][S1]. A 100 MHz scope sees its own 3.5 ns rise time at the -3 dB point, so a 3.5 ns edge arriving from the DUT reads back around 4.95 ns, a 41% rise-time error, and harmonics from the 5th up start rolling off the same way [S3].
The real answer depends on whether the 100 MHz signal is a pure sine or a digital clock with fast edges. For sine work, scope bandwidth only needs to clear the carrier by a small margin (about 1.5x is the conservative shorthand), so 150 MHz is the practical floor for a 100 MHz sine [S4]. For digital clocks, the harmonic content lives in the edge, not the fundamental, which is why the 5x rule and the rise-time/bandwidth relationship (BW = 0.35 / t_rise) keep showing up in every manufacturer primer [S3][S4][S7].
Why the 5x rule exists for digital clocks
Oscilloscope bandwidth is defined as the frequency at which a sine input is attenuated to 70.7% of its true amplitude, the -3 dB point on the front-end amplifier's low-pass response [S3][S7]. For a non-sinusoidal waveform like a 100 MHz clock, the fundamental alone does not determine what you see on screen, because the higher-order harmonics carry the edge information; filter them and the displayed square wave starts looking like a sine [S3].
Siglent's worked example illustrates the cost of skimping: a 100 MHz scope measuring a 100 MHz / 3.5 ns signal produces 41% rise-time error, while a 500 MHz scope (5x bandwidth) drives that same measurement down to 2% error [S3]. Keysight echoes the same "five times the highest frequency component of interest" rule on its bandwidth glossary page [S1]. Tektronix's primer defines the cutoff the same way, 70.7% amplitude at f_{-3dB}, which is the engineering anchor for every scope-spec sheet in the catalog [S7].
Reading the rule: 100 MHz clock, 200 MHz scope, or 500 MHz scope?
Rohde & Schwarz's buyer guide reframes the choice as a spectrum question: a 100 MHz sine needs about 150 MHz of scope, but a 100 MHz digital bus needs 3-5x the clock rate, so 300-500 MHz becomes the floor [S4]. That is why a typical 100 MHz scope "can do a decent but far from perfect" job on a 20 MHz square wave: it attenuates harmonics above 100 MHz and rounds the corners [S2].
Comparing three concrete options against four decision criteria: (1) a 100 MHz scope on a 100 MHz clock is rated at the -3 dB point itself, so edge amplitude is already down 30% and rise-time error is in the 30-40% band [S3][S4]; (2) a 200 MHz scope (2x) cuts that error roughly in half because the dominant 3rd-5th harmonics now sit below the -3 dB knee [S3]; (3) a 500 MHz scope (5x) keeps the 5th harmonic intact and lands measured rise time within 2% of the DUT [S3]. Hanmatek's 2026 buyer note reaches a similar conclusion: 100 MHz is the entry floor, 110-200 MHz is the practical workhorse band for digital work, and you only stop there if your signals are slow analog or low-speed microcontrollers [S5].
Sample rate, memory, and probe bandwidth matter as much
Rohde & Schwarz's buyer guide points out that bandwidth is meaningless without sample rate to back it up: capturing 10 cycles of a 100 MHz clock is a 100 ns window, and at 1 Gsample/s the ADC samples every 1 ns, which is exactly the Nyquist floor for a 100 MHz fundamental [S4]. Most bench scopes pair a 100 MHz analog front end with 1 GSa/s, and 200 MHz models often keep the same 1 GSa/s (or push to 2 GSa/s on 4 channels), so the analog edge can outrun the digitizer if you spread channels [S2][S6].
Probes are the silent bandwidth limit. A standard 10x passive probe is specified to match the scope input, but the system bandwidth is the lower of scope and probe; a 200 MHz scope fed through a 100 MHz probe is still a 100 MHz system, and cheaper probes often roll off well before their label [S2]. For 500 MHz-class work, plan on a probe rated for the scope's full bandwidth and a 50 ohm termination path, otherwise the ringing you see on screen is the probe, not the DUT [S2][S4].
When 100 MHz is enough, and when it is not
A 100 MHz scope is genuinely enough for analog circuits under a few MHz, slow logic (Arduino-class 16 MHz clocks, simple GPIO), audio-bandwidth work, and power-rail ripple measurements where the 20 MHz filter on most modern scopes is the right tool [S5][S4]. EEVblog forum users running through "Learning the Art of Electronics" labs report that 100 MHz covers all the analog and microcontroller signal tracing they actually do, and 200 MHz was not worth the 50% price premium in their case [S2].
100 MHz is not enough for any of the following: 100 MHz or faster clock edges where you need measured rise time within 2-3% of the spec, SERDES lanes (where 0.35 / t_rise pushes you to 1-4 GHz of scope), memory buses at DDR3 speeds and up, RF work above the AM broadcast band, or any compliance test against USB, HDMI, PCIe, or Ethernet [S3][S4][S7]. If your 100 MHz "clock" actually has sub-nanosecond edges, the bandwidth question is set by the edge, not the repetition rate, and 500 MHz is the floor for accurate amplitude readings [S3][S4].
Putting the 5x rule into a one-line spec
For a 100 MHz digital clock with edges in the 1-3 ns range, the spec line to write into the test plan is: "Oscilloscope analog bandwidth ≥ 500 MHz, sample rate ≥ 5 GSa/s on 2 active channels, probe system bandwidth ≥ 500 MHz, with 50 ohm input path." This comes directly from the 5x rule plus the 0.35 / BW rise-time relationship, both of which the major manufacturers document on their scope-knowledge pages [S1][S3][S4][S7].
For a 100 MHz sine source where amplitude accuracy is the only goal, "≥ 150 MHz scope, 1 GSa/s, 10:1 passive probe" is the minimum viable spec [S4]. Two trackable signals to watch: (a) the continued migration of entry-level "100 MHz" scopes to 110-200 MHz analog front ends with the same ADC count, which changes the cost/benefit of the 5x rule on real catalog SKUs [S5]; and (b) probe-system bandwidth disclosures, since modern 4-channel bench scopes often ship with probes that quietly cap system bandwidth below the scope spec.
Spec-level background on the components involved: oscilloscope, signal calibrator, and signal conditioner.
This topic is covered further in EMC Supply for Chip Packaging: 2026 Spec, Material, and Capacity Read.