Frequency Counter#

A frequency counter measures the frequency of a periodic signal with higher accuracy than an oscilloscope’s built-in frequency measurement. While a mid-range scope might measure frequency to 3–4 digits, a basic frequency counter provides 6–8 digits, and a lab-grade counter can reach 10–12 digits. For oscillator validation, clock debugging, and any application where “close enough” isn’t good enough, a dedicated counter is the right tool.

What It Does#

A frequency counter measures:

• Frequency: How many cycles per second (Hz). The fundamental measurement.
• Period: Time for one cycle (1/frequency). More precise than frequency for low-frequency signals.
• Ratio: Frequency of one signal divided by another — useful for verifying frequency dividers and PLLs.
• Time interval: Time between edges on one or two channels — measures pulse width, duty cycle, and propagation delay.
• Totalize: Counts events over a gate time or indefinitely — useful for counting pulses, rotations, or other discrete events.

The core difference from an oscilloscope’s frequency readout is the timebase. A counter uses a precision crystal oscillator (often TCXO or OCXO in better instruments) as its reference, while a scope’s timebase is optimized for waveform display, not frequency accuracy. The scope tells you approximately what frequency; the counter tells you exactly what frequency.

Key Specs and What They Mean#

Feature Tiers#

Reciprocal Counting#

Traditional counters count cycles during a fixed gate time. This works well at high frequencies — a 1-second gate gives 1 Hz resolution regardless of input frequency. But at low frequencies, you wait a long time for enough cycles to count.

Reciprocal counters measure period precisely and compute frequency from 1/period. This gives high resolution at any frequency without long gate times. A reciprocal counter with a 100 MHz timebase achieves 8-digit resolution in 1 second at both 10 Hz and 10 MHz. Most modern counters use reciprocal counting.

Timebase Hierarchy#

The counter’s accuracy is limited by its timebase. From worst to best:

For most bench work, a TCXO is adequate. For calibration or production test, OCXO or GPS disciplining is worth the cost.

Gotchas and Limits#

• Timebase warm-up: OCXO timebases need 15–30 minutes to stabilize. Measurements made during warm-up may drift by several ppm. Wait for the warm-up indicator (if present) or allow adequate time.
• Trigger level: The counter triggers on signal crossings of a threshold. Noisy signals may trigger multiple times per cycle (false counts). Adjust trigger level and use low-pass filtering if available.
• Input coupling: DC coupling includes any DC offset; AC coupling blocks DC but has a low-frequency limit. For signals below ~10 Hz, DC coupling may be necessary.
• High-frequency input path: Signals above ~100 MHz typically require the 50 Ω input for proper impedance match. Using the 1 MΩ input at high frequency causes reflections and false readings.
• Prescaling: Some counters use prescalers to extend frequency range but lose resolution in the prescaled range (dividing by 8 costs 3 bits of resolution). Check whether your counter’s high-frequency range uses prescaling.

Tips#

• For best accuracy, let the counter warm up before making precision measurements — even TCXO timebases stabilize over 10–15 minutes
• Use the longest gate time practical for maximum resolution — a 10-second gate gives 10× more resolution than a 1-second gate
• When measuring a crystal oscillator, compare against the counter’s timebase — if the crystal is supposed to be 8.000000 MHz and reads 8.000047 MHz, the error is 5.9 ppm (or your counter’s timebase is off)
• For signals below 1 kHz, use period mode rather than frequency mode for faster, higher-resolution readings
• If measurements are unstable, check for noise on the signal — add a low-pass filter or reduce input sensitivity to reject noise while still triggering reliably on the signal

Caveats#

• Scope frequency readout is usually adequate — A good modern oscilloscope reads frequency to 5–6 digits, which is sufficient for “is the clock present and approximately right?” checks. The counter adds value for precision verification, not routine checks.
• Accuracy vs. resolution — A counter may display 8 digits, but if the timebase is only ±1 ppm accurate, those last few digits are meaningless. Resolution is how finely you can read; accuracy is how close the reading is to the true value.
• Frequency is an average — The counter reports the average frequency over the gate time. If the signal has jitter or frequency modulation, the counter reports the mean, not the instantaneous value. Use a scope or spectrum analyzer to see variation.
• Don’t chase digits without need — Verifying that an 8 MHz crystal is within ±50 ppm of nominal requires only 5 digits of resolution. More resolution than needed doesn’t improve the test, it just adds cost and warm-up time.

In Practice#

• An MCU that runs slightly slow or fast may have a crystal pulled off frequency by wrong load capacitors — the counter quantifies the error (e.g., “47 ppm high suggests ~1 pF too little load capacitance”)
• When debugging clock trees, measuring each stage (oscillator, PLL output, divided clocks) confirms where frequency error is introduced
• A signal that reads differently each time you measure may have frequency modulation (intentional or not), jitter, or a marginal oscillator — the counter reveals instability that the scope misses
• For UART bit rate issues, measure the actual baud rate clock — a 115200 baud that’s actually 117000 baud (1.6% error) will cause bit errors at long packet lengths
• When calibrating other instruments, the frequency counter (with a known-good timebase) is the reference — calibrate the counter first, against GPS or a known standard