Are surface mount or through hole connections more reliable for safety critical applications?

Question

While obviously factors like safety certifications and margins of error are more important when it comes to safety critical applications, e.g. aerospace, I saw an article that said that one of the remaining areas for through-hole components, is where reliability is important:

• Through-Hole vs. Surface Mount: Through-hole components are best used for high-reliability products that require stronger connections between layers. Whereas SMT components are secured only by solder on the surface of the board, through-hole component leads run through the board, allowing the components to withstand more environmental stress. This is why through-hole technology is commonly used in military and aerospace products that may experience extreme accelerations, collisions, or high temperatures. Through-hole technology is also useful in test and prototyping applications that sometimes require manual adjustments and replacements.

But when googling through-hole vs surface mount reliability, a source painting a more mixed picture popped-up:

• Through-Hole Vs. Surface Mount: Contrasting Benefits and Uses: Disadvantages: Since THT component leads are fed through the board, the PCBs must be pre-drilled, which is both expensive and time-consuming. It also restricts components to one side of the board and limits the available routing area on multilayer boards since the holes must be drilled through all the PCB’s layers. THT’s soldering process often makes the resulting solder points less reliable than SMT solder. Additionally, the THT assembly process is more involved and therefore more expensive than SMT.

Although this source isn't exactly neutral, being a mixed through-hole surface mount vendor. Is there a straight-forward winner when it comes to the reliability of surface mount vs through hole components?

Answer 1

While not a definitive answer, every recent (last 20 years) military and space product I've worked on has has been built using SMT (Surface Mount Technology). There were some exceptions for individual components, but that was more because the part was only available in the through-hole package, and not for any reliability advantage.

Sometimes, particularly for larger packages, we have applied an adhesive under the package or an epoxy to the corners. But that was only done when the structural analysis showed it was needed.

None of these applications were for what I would call extreme acceleration environments, such as a cannon-launched shell (a smart munition). But they do have to survive launch and stage separation pyrotechnical shock, and many thousands of temperature swings.

Finally, many components are just not available in through hole packages. For example, show me a 1000+ pin FPGA in a through hole package. In those cases ways are found to make them work in the given environment.

Leaded parts are always preferred where large number of temp excursions are involved. The largest leaded part I used was an Actel (now Micro Semi) FPGA in either the 256 lead or 352 pin Ceramic Quad Flat Pack (CQFP) package, I forget which (was over 15 years ago). Even though it was SMT, because of the fine lead pitch, it could not be solder automatically and required a trained operator to hand solder it down.

Everything (COTS) I've seem above that IO count has been been in some sort of BGA or CGA package, which requires careful attention to board material (so as to minimize CTE mismatch between the package and the board) and staking and/or underfill of the package to the board.

You can get high IO count leaded packages, but they are usually custom designed and built MCMs (Multi Chip Modules).