A thread looks like a feature cut into a hole, and it is usually drawn that way. Functionally it is a relationship. The thread has to be coaxial with the fastener, square to the face the fastener pulls against, and deep enough that the fastener engages properly without bottoming. If any of those three is wrong, the fastener either will not start, will not seat, or will not stay tight under load and heat.
That is why the order of operations matters so much on welded work. Welding moves metal. A thread cut before welding is cut in a position that welding may change. A thread cut after welding is cut in the position the part actually ended up in. Neither order is automatically right, and the correct choice depends on which of the three relationships is at risk.
A third possibility is usually the best one where geometry allows it: put the thread somewhere the weld cannot reach. Moving a fastener hole a short distance away from a seam costs nothing at the drawing stage and removes the argument entirely.
Heat from a weld spreads into the surrounding metal and leaves unevenly. A boss with a thread in it, welded on one side, will bend slightly towards the weld as it cools. The thread itself does not shrink, but its axis tilts. A fastener threaded into it then tightens against a face that is no longer square to it, and the joint feels tight while carrying its load on one edge of the contact.
The second effect is oxide and scale. Welding stainless leaves heat tint and, close to the weld, a layer that is harder and more abrasive than the parent metal underneath. A tap run into a hole with that layer in it wears quickly, and the thread it produces is rough in exactly the region where the first threads engage and carry most of the load. You can see the kind of stainless assemblies where this arises on our TIG stainless steel welding pages.
A thread that sits well away from the weld survives fine. A thread in the middle of a flange that is welded at its outer edge sees little heat and almost no movement. That part is better machined complete and welded afterwards, because the machining happens in a rigid and simple setup and the weld is far enough away not to matter.
A thread close to the weld does not survive as well. The heat softens the crests slightly, the bore can go oval, and scale fills the roots. A thread that was cut clean and then welded can end up needing to be chased anyway, which means nothing was gained by cutting it early and a setup was spent for no reason.
Machining a thread after welding puts the thread where the metal actually is, and that is the accuracy argument in its favour. The cost is access and setup. The assembly is now an awkward shape, it has to be held without crushing a tube or a thin flange, and the hole may sit at an angle that requires the part to be tilted. On a welded subassembly that is a fixture problem, and fixtures for welded assemblies are the expensive kind.
There is also the interrupted cut where a hole passes through or near a seam. The tap or the boring bar crosses from weld metal into parent metal, and the cut is uneven at the boundary. That is manageable if it is planned, and unpleasant if it is discovered halfway through the hole.
Chasing an existing thread removes a small amount of material to clean up distortion and scale. It assumes the thread is in the right place and only needs to be made usable again. Cutting a thread from a drilled hole establishes the position as well as the form. The two operations answer different questions, and a note on a drawing saying that a thread is to be cleaned after welding is asking for the first rather than the second.
Where position is critical, chasing is not enough. If the boss has tilted, cleaning the thread gives a clean thread on a tilted axis, and the fastener still seats badly. In that case the answer is to machine the seat face and the thread in the same setup after welding, so that both are corrected from one datum at the same time. That single decision removes more leaks than any thread sealant.
Stainless fasteners in stainless threads gall. The two surfaces are similar, they are often dry, and the contact pressure inside a thread is high. Once galling starts, the fastener seizes part way in and the thread is damaged on removal, usually on the car rather than on the bench.
Aluminium is worse in a different way. A steel fastener in an aluminium boss will corrode the boss over time, and the thread comes out with the fastener when it is finally undone. A dry lubricant, a thread insert or a different pairing all help, and all of them depend on the thread being clean and correctly formed to begin with. That brings the argument back to the order of operations rather than to the thread itself.
The material pairing also changes how much force a thread can take before it strips. That is a design calculation, not a shop floor choice, and it belongs on the drawing rather than in a conversation after the first part fails.
A sensor thread is a special case because it has to seal as well as hold. The sensor seats on a washer or a taper, so the thread position and the seat face have to be related to each other, and both have to be square to the axis of the sensor. If the boss is welded after the thread is cut, the seat face moves and the sensor either leaks or bottoms before it seals.
The reliable sequence is to weld the boss, machine the seat face, and cut the thread in the same setup. The relationships are then established from one datum in one operation, and the part cannot disagree with itself. It costs a setup after welding, and it removes a category of leak that no amount of sealant will fix later.
Thread questions are cheap at the drawing stage and expensive at the assembly stage, so they are worth settling before any tooling is built.
You can see the kind of assemblies these questions apply to on our machined and fabricated parts pages, the second group of small components, the brackets and adapters, our titanium fabrication components and the titanium welded parts that carry them.
