“How much does it cost to build one of these?” It depends…
When an aerospace customer needs a new printed circuit assembly built for a legacy system, invariably their first question is “how much will it cost?” Building a new assembly can be relatively straightforward—or impossible—to accomplish. It depends. Is it build-to-print? Assuming the engineering dataset is complete, and you can wrestle it to the ground, build-to-print can be cost-effective. Reverse engineering, or reproducing the engineering data set and getting regulatory acceptance, is by far the costlier path. Here I’ll address some of the challenges of build-to-print for legacy circuits.
New Old Stock. The first consideration is whether the components called out in the drawing package are still available. And make sure the available stock is actually new, as in never been used, because refurbished old stock will not pass muster. If the components aren’t available, in many cases the drawing package may have notes about acceptable substitutions for parts. If these aren’t available either, you’ll have to find a cross-referenced part or build new original parts.
Cross-Referencing Old to New Parts. In some cases, the component industry has already provided the acceptable engineering data to cross reference old components to new. For example, resistor manufacturers have already produced the necessary engineering information for industry to accept modern metal film resistors in place of legacy carbon composition.
New Original Parts. For out-of-stock active components, the original dies may exist and can be used to produce new components in the original package. Rochester Electronics is good example of a firm that can produce new parts from original dies, or in some cases produce new dies from the engineering dataset. While the cost of a short run of parts may be high, it still may be cheaper than a significant redesign in order to accommodate current generation parts.
Component Certificates. What is the strategy for components that lack OEM certificates? Unfortunately, many NOS parts don’t have the original manufacturer certificates. In this case a certificate can usually be produced by a component testing lab. In some conditions, environmentally testing the printed circuit assembly utilizing these components may be an acceptable alternative and more representative validation of its intended use.
Minimum Order Quantities (MOQ). Even though there may be new old stock available, and you can get certificates produced, it may have unacceptable minimum quantities. If so, you’re back to cross-referencing or building new original parts.
PCB Fabrication. In some cases, Gerber files do not exist and film is all that’s available. Is the PCB film in serviceable quality? Even so, old film probably isn’t compatible with current fab equipment. You’ll have to digitize the film to produce usable Gerber files for fabrication of the PCB. You’ll also want to check the fabrication processes called out in the drawing package. For example, is the conformal coating process called for in the drawing still in use today? If not, you’ll have to use current accepted fabrication processes and get a waiver from the DER (Designated Engineering Representative).
Surface Finishes. Here’s an additional challenge. Modern component finishes are almost universally Pb-free. So tin whisker mitigation strategies should be employed as it introduces a failure mode not anticipated in the original design engineering activities. One solution can be re-plating Pb-free component pin surfaces with tin lead.
Test and Burn-In. Most printed circuit assemblies originally required a functional test fixture designed and built by the OEM for functional testing of completed PCAs. If this equipment exists and is in serviceable condition, testing can be accomplished in straight forward manner. Otherwise the drawing package for the test fixture must be acquired and the fixture built.
So how much will it cost to produce a legacy PCA? It depends… on how successfully these build-to-print challenges can be addressed. If you can’t wrestle build-to-print to the ground, you may end up taking the reverse engineering path. I’ll address those challenges next time.