Simulation is necessary to gauge system response and model performance when designing PCBs for manufacturing. However, real-world imperfections in manufacturing processes mean that component values differ from their nominal values; how much they differ depends on how tight the tolerancing is. Component tolerance is a vital aspect of board cost and performance as per-unit prices for components increase the closer the manufacturer can get to the ideal value.
| Comparing Relative Component Tolerances | |
|---|---|
| Tighter | Looser |
|
|
Component Tolerance and Population Distribution
Tolerance is part of a more extensive component manufacturer standardization approach that maximizes repeatability in production and yield while ensuring that in-circuit performance matches advertised attributes. Tolerance is usually a percentage of the nominal value, with a lower percentage indicating a closer match between the expected and actual value. Manufacturers may indicate the tolerance as a straight variance in picofarads (pF) instead of a percentage for small capacitors. In effect, designers can consider tolerance as a measure of quality; all components should meet the tolerance specified by the manufacturer, but tighter tolerances mean fewer components of the manufactured lot meet the constraint. For component manufacturers, an increase in per-unit price compensates for the reduction in acceptable yield.
Before further discussing component tolerance, it’s necessary to mention distribution patterns. Even for large bands of component tolerance (say, >±20%), not all occurrences within the distribution are equally likely; in other words, continuous uniform distribution is not evident. If that were the case, exceedingly large component lots would exhibit an equal likelihood of a value at either end of the spectrum. Instead, components exhibit a Gaussian distribution about the nominal or “ideal” value, i.e., the “sold-as” component value. For Gaussian distribution, it’s far likelier that a component value lies closer to the nominal value (or the center of the bell curve) than the extrema, with the total distribution decreasing from the center out.
How Component Tolerance Models Manufacturing Yield
Beyond per-unit cost, tolerance plays a significant role in systems’ extreme value analysis (EVA), which determines the maximum operating range at a given component tolerance. To maximize PCBA yield, designers can calculate the theoretical circuit minimum or maximum values depending on the component tolerance. Simulation software can simultaneously implement multiple checks for component tolerance, and results offer feedback for guidance (e.g., where component tolerance is less or more important). EVA segues naturally into Monte Carlo simulations that recreate and run circuits thousands of times using randomized values within the accepted operating range to ensure optimal system manufacturability.
Crucially, designers can determine the maximum tolerance to ensure a theoretical 100% yield from component values. This value doesn’t equate to a practical 100% yield during manufacturing with many different inputs to account for, but, in essence, it ensures the circuit values are as close to theoretically perfect as possible while optimizing for cost. As mentioned, tighter tolerances increase per-unit component costs, so while designers may desire maximum tolerance ratings for every component in their circuit, this quickly becomes financially infeasible with moderate-and-up production quantities. Graphically, the relationship between an increasing tolerance from nominal values against yield resembles a logarithmic decay function. Near the zero tolerance, yield is at a maximum, whereas yield falls off sharply (and then slower) with more deviation from the ideal.
Your Contract Manufacturer Optimizes Cost and Yield for PCBAs
Component tolerance significantly affects manufacturing yield, but designers will want to be cautious to avoid unnecessary cost overruns (especially for any considerable quantity of production volumes). Simulation and calculations will give designers a better idea of the minimum tolerance necessary for an ideal manufacturing yield of 100% (i.e., the component tolerance will not contribute to any out-of-specification products that fail to meet design goals). However, there are other design vectors to consider that can cause higher-than-acceptable yield losses; fortunately for designers, VSE is here to help. Our engineers are committed to building electronics for our customers, including a complete design review to identify areas needing optimization. We’ve been realizing life-saving and life-changing designs for over forty years with our valued manufacturing partners.
