On this page

1) Why you should “engineer” an entry-level crystal

In mission-critical timing (radar, satcom, network holdover), designers treat time and frequency as a system budget and often anchor performance with OCXO. In consumer and home electronics, the same idea applies at a different scale: you may not need ppb stability, but you do need repeatable start-up, predictable frequency offset, and noise resilience in real boards.

2) The three roles entry-level MHz crystals play in real systems

• MCU/SoC core clocks: boot, scheduling, timers, peripheral dividers.
• Interface reference clocks: Ethernet PHY, USB, audio/video, connectivity platforms.
• Manufacturing / lifecycle balance: mature packages, stable supply, controllable reflow behavior for high volume.

Start from the family overview and map to packages/series: Entry-Level MHz Crystals (FCX/FSX/HC).

3) Frequency planning: don’t pick MHz numbers blindly

The “right” frequency is constrained by your MCU/SoC PLL plan, interface standards, and EMI budget. Common reference points (chip-specific—always follow the datasheet):

4) Package selection: 3225 vs 2520 vs 5032 vs HC metal can

• 3225 (3.2×2.5 mm): balanced choice for gateways, smart displays, STB and dense mainboards.
• 2520 / smaller: saves area but is more sensitive to layout noise and stray capacitance variation.
• 5032 (5.0×3.2 mm): larger pads for more relaxed assembly; great for white goods control, lighting and power control boards.
• HC-49 (through-hole / SMD metal can): legacy platforms, development boards, and rework-friendly builds.

Need a broader product index? See Quartz Crystal Resonators.

5) The four levers that decide success: CL, ESR, drive, layout

5.1 Load capacitance (CL): treat stray as an input, not an afterthought

CL sets where the crystal runs (pulling). Start from the crystal’s specified CL and design C1/C2 while accounting for stray capacitance from pins, pads, and traces.

For dense boards, Cstray can move enough to shift frequency and erode start-up margin. Keep C1/C2 as tuneable options in EVT and validate across temperature.

5.2 ESR and gain margin: what actually determines start-up reliability

High ESR makes start-up harder; noise and supply disturbance consume margin. But “lowest ESR wins” is not always true—very low ESR can over-drive the loop depending on the MCU. When a project has start-up pain, read: Start-up time in crystal oscillators – causes and fixes.

5.3 Drive level: too high accelerates aging, too low causes slow or unstable start

Crystals are specified in µW drive limits. Excessive drive increases aging and can distort behavior; insufficient drive increases time-to-ready and may cause dropout under noise. Make drive validation a DVT item, not a lab anecdote.

5.4 Layout: place it like an analog loop

• Put the crystal and C1/C2 close to the MCU pins; keep traces short, symmetric, and away from layer hops.
• Keep the return path tight; avoid routing over splits; keep switching nodes and fast edges away.
• Use a clean local ground and consider a guard/keepout under the crystal loop if needed.

6) Validation checklist (EVT → DVT → PVT)

7) Fast selection table: map your design to FCX / FSX / HC

This is the quickest way to shortlist. For the full family narrative and applications, see Entry-Level MHz Crystal Family.

8) Recommended entry-level MHz crystal picks (product-embedded)

9) When to upgrade: crystal → XO / TCXO / OCXO

A bare crystal depends on the on-chip oscillator loop. Upgrade when you need lower tuning risk, more controlled output, or tighter stability:

• Reduce validation risk: you want predictable output waveform and less CL tuning → consider XO.
• Stability across temperature: wireless, sync, or accuracy budgets become visible → TCXO.
• Holdover and ultra-low noise: high-end timing and mission-critical links → OCXO.

References

• STMicroelectronics: AN2867 Oscillator design guidelines
• Microchip: AN2648 Crystal basics, layout and testing · AVR186 PCB layout best practices
• Texas Instruments: SNLA290 Crystal selection and specification (clock accuracy impact) · SNAA065A Crystal based oscillator fundamentals

FAQ

What CL targets are common for entry-level MHz crystals?

8/10/12/16 pF are common targets, but always follow the crystal datasheet and MCU oscillator spec. The correct value is the one that meets frequency offset and start-up margin across corners.

How do I estimate Cstray without overthinking it?

Start with 2–5 pF as a practical range, then iterate using frequency measurements while swapping C1/C2 in ±1–2 pF steps. Avoid heavy probe loading on XIN/XOUT.

My design starts in the lab but fails in the field—what’s the usual root cause?

Margin erosion under worst-case noise, temperature, and VDD: layout injection, incorrect CL due to strays, or ESR/drive mismatch. Validate cold + low VDD + max switching noise with multiple lots.

How can I reduce EMI from the crystal loop?

Short and symmetric routing, tight ground return, keepout from switching nodes, and avoiding long parallel runs near high-dv/dt lines. If needed, add a small series resistor to tame over-drive and edge energy.

Is it worth adding a series resistor (Rs) in a Pierce oscillator?

Often yes when you observe over-drive or abnormal start behavior. Use a small range (for example 0–100 Ω) and re-validate start-up margin and frequency.

When does a crystal become a poor choice vs an XO?

When you cannot get repeatable start-up and jitter behavior across builds, or the platform expects a defined logic waveform with tight jitter limits. XO reduces tuning risk at a modest BOM increase.

What should I include when requesting engineering support?

MCU/SoC part number, target frequency, temperature range, supply range, intended CL, measured start-up behavior, and a PCB snippet of the clock loop placement/routing.

Need design review or part shortlisting?