Six common mistakes engineers make when specifying quartz crystals, and how to avoid them.

Most articles on specifying quartz crystals give you a checklist of frequency, stability, tolerance, load capacitance, package, and temperature range. Tick the boxes, raise the PO, and hope the part works.

In forty years of frequency control, we have watched products fail not because the boxes were wrong, but because nobody understood how the boxes interact. A crystal that meets every parameter on paper can still fail at production scale. The difference is not the parameters themselves. It is the judgement that connects them.

These are six trade-offs design engineers wrestle with that most articles will not tell you about.

1) Frequency tolerance is the easy number, but stability is where designs actually break

Tolerance is measured at 25 °C and tells you the part is accurate today, on a bench. Stability is what happens when your design enters the real world. A ±10 ppm tolerance with ±50 ppm stability is a part that drifts ten times more than its specification suggests, the moment your customer turns it on in a hot car or a cold warehouse. Both numbers need to be specified together, every time.

2) Load capacitance is not a value on the datasheet, it is a number you have to calculate

The crystal's CL specification assumes the rest of your circuit, including trace lengths, stray capacitance, and driver IC capacitance. If the maths is off by 2 pF, your oscillator will start up slowly, drift off frequency, or fail to start at all in the cold. The right approach is to match the crystal's CL to your actual circuit, which is something we model with customers at the design-in stage. Catching it then is cheap, but catching it after the boards are built is expensive.

3) Package size sells the BOM, but package thermal performance saves the programme

Smaller crystals look better in tight layouts and procurement loves them. What the small package often hides is reduced thermal robustness and higher drive level sensitivity.

The right package is not the smallest that meets specification. It is the one that survives the manufacturing process and the environment your product is actually going into.

4) The operating temperature range is not where the crystal performs, it is where it stays alive

A crystal rated -40 °C to +85 °C survives those temperatures. Stability curves are non-linear and ageing accelerates at the extremes.

For automotive, defence, or outdoor industrial applications, you should specify on performance across the range, not survival of it, and then test against your own profile rather than the datasheet's.

5) Ageing is the parameter nobody asks about, and the one that ends programmes

Crystal frequency drifts with time, slowly and predictably, but inevitably. Most datasheets quote first-year ageing as a single number, typically ±3 to ±5 ppm. What they do not quote is what happens in year ten or year fifteen, on a programme that has to hold time accuracy for the life of a defence radio or a medical device. The ageing rate slows but never stops. For long-life applications, the right question is not what the first-year number is, but what cumulative drift you will see over the programme's full operating life.

The timing budget has to be built around that answer.

6) The cheapest crystal is the most expensive part on your BOM

A standard 20MHz crystal is inexpensive. The same nominal frequency from a high-stability, low-ageing, defence-grade source, costs an order of magnitude more. For a consumer product, the cheap option might be right. For a medical device or defence radio that has to hold accuracy for fifteen years or more across temperature, vibration, and ageing, the cheap option will cost you the programme.

The conversation is not about the crystal price. It is about the cost of the field failure that the crystal price is hiding.

These parameters are a starting point, not an answer

What you need at design-in is someone who has specified crystals into real applications and knows what fails. Someone who understands where the datasheet stops and where engineering judgement starts.

Frequency control is a specialist discipline. We've spend four decades making it ours. Not as a catalogue supplier. As engineers who sit inside your design process.

Tell us what your circuit needs, what environment it lives in, and how long it has to last. We will recommend the right part, with the reasoning behind every choice.