GFE Solutions

Engineering fact: Why G-code still rules CNC manufacturing

G-code was developed in the 1950s. Decades later, it still drives the majority of CNC machines running in production facilities around the world — and the reasons why are more practical than nostalgic.

The language is explicit and deterministic. Every line tells the machine precisely where to move, at what speed, and with which tool. There is no abstraction layer between the programmer and the machine, which means that when something goes wrong on the shop floor, an experienced CNC engineer can open the file, read through the code, and locate the problem directly — without navigating layers of software interpretation.

Modern CAM systems generate G-code automatically from 3D models, so most programmers today rarely write it by hand. But the ones who can read and edit it manually are the ones who can actually optimise a toolpath, reduce cycle time, and catch problems before a single part is cut.

Alternatives like STEP-NC exist and have genuine technical advantages, but displacing a standard that runs on millions of installed machines worldwide is a slow process, regardless of how good the replacement is.

At GFE Solutions, CNC programming is one of our core services. Our engineers work directly with the code, not just the CAM output — which makes a measurable difference in part quality and production efficiency.

24 on-site visits to Germany over seven years. That number says more about how this engagement works than any service description could.

The client manufactures experimental synthetic fiber production lines — machines where mechanical design, electrical systems, and fluid dynamics are all interdependent. On that kind of equipment, the control logic can't be developed from a frozen spec. It has to evolve in parallel with the mechanical design, respond to decisions being made in real time, and reflect the actual behavior of the machine — not a model of it.

So our engineer joined the core project team from the start. Not as an external consultant brought in at commissioning, but as a permanent member sitting alongside the lead and design engineers throughout every project. The work is done in TIA Portal with SCADA integration. Between visits — continuous remote engagement. The arrangement has been running since July 2019 and keeps growing in scope.

Some things only happen when you stay long enough.

Engineering Fact: How tolerance stacking can kill a perfect design

Every individual part in your design looks perfect on paper. Each dimension is within spec, every tolerance has been reviewed and approved. And then the assembly doesn't fit together — or worse, it fits during production but fails in the field. Welcome to tolerance stacking.

What is it?
Tolerance stacking (also called tolerance accumulation) happens when the allowable variations of individual components add up across an assembly. Each part has its own dimensional tolerance — a small acceptable range around the nominal value. When multiple parts are assembled in a chain, those small variations don't cancel each other out. They accumulate, and in the worst case, they all push in the same direction at once.

Why it matters more than most engineers expect
A single ±0.1 mm tolerance seems harmless. But in an assembly of ten components, that same logic applied at every step can produce a total variation of ±1 mm — enough to cause misalignment, binding, premature wear, or a complete functional failure. The design was never flawed in isolation. It was flawed as a system.

How it gets missed
Tolerance stacking issues rarely appear in single-part reviews. They emerge at the assembly level, often late in the development process when changes are expensive. This is why early-stage tolerance analysis — using worst-case or statistical methods like RSS (Root Sum Square) — is not optional on complex assemblies, it's the difference between a design that works on a drawing and one that works in production.

Seen this issue derail a project? We review and validate assembly tolerances as part of our CAD and FEM engineering services — before the parts reach the shop floor.

Six years ago we started working with a pharmaceutical packaging machine manufacturer. At that point we had one task: write CNC programs for their machined components.

Before the first program was written, our team lead spent over a month at their facility. Not to sign documents or sit through presentations — to understand how their production actually works. What the quality standards mean in practice. How the machinists on the floor think. How a tolerance that looks acceptable on paper behaves differently on the actual machine.

During that time there was an opportunity to see the entire production cycle first-hand — from planning and blank material ordering, tool preparation, through to machining and final quality control. That kind of visibility changes how you write programs.

The programs are written in SolidCAM, validated in Vericut. Everything goes through the client's own system, using their methods and file structure. Nobody on their side has to adapt to how we work — we adapted to how they work, and we've maintained that for six years.

14,420 tasks. And we're still getting new ones.