Industry Article

How Choosing Oscillators over Crystals Reduces Time to Market and Project Costs

November 05, 2019 by Bruce Potvin, SiTime

This article looks at the differences between crystal resonators and oscillators and presents examples that show when choosing crystals over oscillators could increase the total cost of a design.

This question of the actual cost of using a crystal versus a MEMS oscillator may not be at the forefront of a designer's selection process when the price of crystals seem so cheap—at least on the surface. But although the cost of crystal components is generally lower, once the total cost of design is calculated, the picture looks much different.

I have heard from many designers who have crystal design issues such as cold startup failures, oscillator circuit problems from mismatched crystals, or failure to pass EMI tests. These problems cause engineering cost overruns during development and can create costly quality issues. Plus delaying the product release date can cause costly lost opportunities. Described in this article are three situations where designers needed to decrease their overall cost of ownership when facing crystal design concerns.

 

The Basics of Crystals and Oscillators

What is the difference between a crystal (XTAL) and an oscillator (XO)? A crystal (sometimes called a resonator) is a moving/resonating passive device that connects to the external oscillating circuit in the chip that it is timing, like an SoC, microcontroller or processor as shown on the left in Figure 1.

 

the different design of a crystal and an oscillator

Figure 1. Comparing the design of a crystal resonator and an oscillator.

 

An oscillator, shown on the right in Figure 1, is an integrated timing solution that contains a resonator and an oscillator IC in one active device. For some oscillators, the resonator is based on silicon MEMS (micro-electro-mechanical systems) technology instead of the traditional quartz crystal. This architecture enables robust “plug-and-play” timing products that are flexible and very easy to design into a system.

 

Looking at the Total Cost of Ownership in Crystals Vs. Oscillators

Oscillators are easier to design with since they include functionality and features that solve common and often difficult timing design problems, as illustrated in the following cases. These examples are based on pricing from Digi-Key for XTALs and XOs with the same output frequency, frequency stability, and package size*. Added to the price of the timing component is the cost of engineering work-hours (based on $50 per hour) that are required to remedy the problem.

Each case has a different breakpoint based on production volume and engineering time. Not surprising the cost of designing with a crystal is lower when quantities are high and design costs are amortized over large volumes. Conversely, the cost of using an oscillator is lower when quantities are lower. But there’s more to the story.

What’s not factored into the following calculations is the opportunity cost due to project design delays, which can be tremendous in some markets. In some cases, there are additional costs for outside services and testing—and these also can be significant. Plus there are other penalties such as the costs for additional materials/components for board re-spins, the cost of load capacitors that are required with crystals, and the additional board space consumed by the capacitors (all of which further tilt the equation toward using an oscillator).

Note: For simplicity sake in the following examples ONLY the cost of the timing component and the engineering time are included in order to correct the crystal problem.

 

Example 1: Cold Startup Failure

Unlike crystals, MEMS oscillators do not have startup problems. In this case, 15 hours of engineering work was required to correct the crystal startup problem. Here, with a relatively quick fix, the cost-benefit of using a MEMS oscillator is realized when production volume is around 2,800 units or less. These numbers can be seen in Figure 2. 

 

cold startup failure costs

Figure 2. Cold startup failure costs.

 

Example 2: Mismatched Crystal Causes Oscillator Failure

Because oscillators are an integrated solution (combining the resonator and oscillator IC), matching errors are eliminated. Designers don’t need to worry about crystal motional impedance, resonant mode, drive level, oscillator negative resistance, or other pairing considerations. In this instance, 40 hours of engineering work is required to correct the matching issue, making the cost of using an oscillator payoff at around 8,000 units or less. These numbers can be seen in Figure 3. 

 

Mismatched crystal causes oscillator failure

Figure 3. The cost of mismated crystals causing oscillator failure.

 

Example 3: EMI Compliance Failure

The clock is often the largest contributor to electromagnetic interference (EMI) in a system and it can cause a prototype to fail compliance testing. Some MEMS oscillators offer multiple techniques for quickly and easily reducing EMI. One such technique is spread spectrum clocking. There are oscillators that include a programmable feature for adjusting the rise/fall time (slew rate) of the clock signal to lower EMI.

Crystals don’t have these features. If designers need to use shielding or add a spread spectrum clock generator IC with their crystal, this adds expense and board space. Plus, renting an anechoic chamber for additional testing could incur another $3,000. To redesign the board and retest, it can take 50 hours of engineering work, making it more beneficial to use a MEMS oscillator at volumes around 11,000 units or less. And this doesn’t include the additional materials and test facility costs mentioned above.

 

Conclusions Drawn from these Examples

In addition to direct costs, there are other factors that affect the cost of designing with crystals. For example, oscillators can drive multiple loads. That means one oscillator can replace multiple crystals, which can only clock one device.

MEMS oscillators that are based on a programmable architecture are readily available in any frequency, stability, and voltage within a very wide range. This provides great flexibility for designers in optimizing their design. In fact, such oscillators can be programmed by key distributors or even by customers in their own lab using Time Machine II.

 

comparing the benefits between oscillators and quartz crystals

Figure 4. An overview of the additional benefits of using MEMS oscillators.

 

Programmability can also reduce the cost of qualification efforts if specification changes are needed. This time-saving benefit is possible because a MEMS oscillator (before programming) can generate millions of part numbers and specification combinations—all with the same base part.

Perhaps one of the biggest indirect savings comes in the form of higher reliability and quality. Silicon-based MEMS oscillators can have higher reliability of over as much as one billion hours mean time between failure (MTBF) compared to typical quartz devices with about 25 million MTBF. These devices deliver less than two DPPM quality level which is about 30 times better than quartz devices, and they have much better survival rates against shock and vibration compared to quartz crystals. Using an oscillator in place of a crystal can provide benefits beyond cost as well (PDF). 

The higher failure rates of quartz crystals can increase costs in many ways such as the added resource costs for root-cause analysis or extra service and replacement costs. When procurement is focused on lowering component costs, looking at the big picture can ultimately save in the long run. 

Industry Articles are a form of content that allows industry partners to share useful news, messages, and technology with All About Circuits readers in a way editorial content is not well suited to. All Industry Articles are subject to strict editorial guidelines with the intention of offering readers useful news, technical expertise, or stories. The viewpoints and opinions expressed in Industry Articles are those of the partner and not necessarily those of All About Circuits or its writers.

0 Comments