Publications
Backdrilling vs. Laser Microvias, Part 2: Reliability, DFM and the Case for Hybrid Design
The first part of this discussion compared backdrilling and laser microvias primarily through signal integrity, cost, manufacturability and application fit. Part 2 examines a different question: which technology will remain reliable after the board leaves the laboratory?
A via is both an electrical interconnect and a mechanical feature. It carries signal or power, connects copper structures across layers and experiences thermal expansion, vibration, shock, plating stress and current-induced heating. A technology that looks excellent in a simulation can still fail if the stackup, fabrication data or qualification process is not controlled.
The most useful conclusion is not that one technology replaces the other. Backdrilled PTH vias and laser microvias occupy different positions in the design space. Backdrilling is generally stronger, cheaper and better suited to current-carrying or harsh-environment applications. Microvias provide greater routing density and lower parasitics, particularly in fine-pitch BGA and very-high-speed regions. A hybrid architecture can allocate each technology where it creates the greatest value.
Backdrilling vs. Laser Microvias, Part 1: Choosing the Right Way to Control Via Stubs
As PCB data rates increase, the via is no longer a passive hole in the board. It becomes part of the transmission channel. A plated through-hole that extends beyond the signal layer can leave a copper barrel segment—known as a via stub—that behaves as a discontinuity. At sufficiently high frequencies, that stub can create resonances, reflections, insertion loss, jitter, eye closure and unwanted radiation.
Two widely used ways to control the problem are backdrilling and laser microvias. Both can improve signal integrity, but they solve different design problems. Backdrilling modifies a conventional plated through-hole after plating. Laser microvias create short, blind interconnects between adjacent layers and avoid the full-length barrel altogether.
Human-Centered Enterprise Architecture: An Organizational Developmental Strategy for the AI Era
The commercialization of Agentic Artificial Intelligence and enterprise-grade automation presents medium and large organizations with a defining strategic challenge. As software agents assume responsibility for transactional customer interactions, traditional management paradigms dictate aggressive headcount reductions to harvest immediate software efficiency gains.
However, empirical evidence from major enterprise transformations demonstrates that cost-first restructuring creates severe systemic vulnerabilities. Aggressive layoffs erode organizational trust, trigger widespread employee anxiety, destroy institutional memory, and cause severe service degradation—ultimately compromising brand reputation and customer retention.
Strategic Workforce Transformation - A Cost-Benefit Analysis
Organisations adopting AI in customer-facing work face three broad choices: immediate layoffs, a mixed approach combining reductions with internal reskilling, or a phased transformation that uses internal redeployment first and external hiring selectively. The best answer is rarely determined by salary arithmetic alone. It depends on the quality of automation, the cost of losing product knowledge, the time required to recruit and train new staff, service risk, employee trust, labour law and the value of new customer-facing capabilities.
This report uses a US high-volume retail/e-commerce contact-centre reference case because publicly available cost benchmarks are more accessible than a single global benchmark. The case is illustrative, not a prediction for every company. It models a 1,000-person operation over three years and compares three strategies.
The recommendation is a gated hybrid: automate suitable routine work; redeploy and reskill existing employees; use attrition and voluntary mobility before involuntary reductions where feasible; and hire externally only for verified gaps. The recommendation is not based on a claim that reskilling is always cheaper. It is based on risk-adjusted economics: preserving knowledge, reducing ramp time, maintaining customer quality and keeping the organisation able to reverse course if AI performance or unit economics disappoint.
Customer-Facing Roles in the AI Economy
Customer-facing work is moving from transaction execution towards orchestration, judgement, relationship management, insight and governance. AI can automate or assist predictable tasks, but the emerging operating model still requires people who understand customers, products, risk and context. The strongest evidence does not support a single global “jobs disappear” conclusion. It supports a task-level transition in which routine work is compressed and higher-value responsibilities expand.
Forrester presents the most disruptive view, forecasting a substantial reduction in US customer-service jobs by 2030 and highlighting the decoupling of inquiry volume from headcount. Gartner reports a more adaptive organisational response: service leaders are expanding agent responsibilities, revising skill requirements and moving employees into new roles.
OECD and ILO caution that exposure to AI is not the same as occupation-level displacement.
Minimizing EMI radiation from Via Transitions in ML PCBs - Part 2
Via transitions are among the most important EMI-control points in a multilayer PCB. Whenever a high-speed signal moves from one layer to another, the signal current and its return current must move through a three-dimensional structure containing vias, pads, antipads, reference planes, and dielectric material.
The most important design principle is simple: keep the signal current and return current close together through the entire transition. A smaller current loop produces lower inductance, less field leakage, and lower EMI.
The numerical reductions discussed in this paper should be treated as representative engineering results rather than universal guarantees. Actual performance depends on stackup, bandwidth, geometry, material properties, manufacturing tolerances, and the complete channel.
Minimizing EMI radiation from via transitions in ML PCBs - Part 1
Via transitions are among the most significant sources of electromagnetic interference in multilayer PCBs, particularly when signal bandwidth extends into the gigahertz range. A via is not merely a vertical electrical connection. It is a three-dimensional discontinuity containing inductance, capacitance, unused barrel length, antipads, reference-plane transitions, and coupling to nearby structures.
The design objective is not simply to reduce via count. It is to control the complete electromagnetic transition and keep the forward and return currents closely coupled.
Designing Via Transitions to Minimize Return Path Discontinuities
Via transitions are unavoidable in most multilayer PCBs. They allow signals to move between routing layers, connect components across the board, and support dense escape routing under BGAs and other fine-pitch packages.
However, every via transition introduces a local electromagnetic structure with its own inductance, capacitance, impedance, and return-current path. If these properties are not controlled, the transition can create a return path discontinuity (RPD). The result may be signal reflections, crosstalk, timing errors, ground bounce, mode conversion, and increased electromagnetic interference.
This white paper explains how to design ‘via’ transitions that preserve return-current continuity while remaining practical to manufacture.
Building Organisational AI Capability in 2025-26
For Léim Partners, the key question is not whether organisations should use AI, but how they can build the internal capability to use it effectively, responsibly, and at scale. The organisations that succeed will not be those that simply buy tools or launch one-off projects. They will be the ones that treat AI as an organisational transformation, embedded into workflows, governance, data structures, skills development, and decision-making.
AI Impact on Customer Facing Roles
The global enterprise landscape is witnessing a structural transformation in customer-facing operations. Driven by the rapid deployment of Autonomous Agentic AI, routine customer interaction channels are transitioning from labor-intensive execution to automated software execution. However, market research organizations hold distinct and complementary perspectives regarding the depth, timing, and long-term resolution of this disruption.
Why high-speed PCB designs fail quietly
High‑speed digital design is no longer a specialist niche – it’s the default for every serious product team working on compute, networking, and embedded systems. Yet one of the most critical aspects of signal integrity is still routinely treated as an afterthought: the current’s return path on your reference planes.
