Excerpt: Steam pipe and buttweld elbows are the cheapest way into a custom turbo manifold, but material choice, layout, and weld sequence decide whether it lasts ten years or cracks in one season. Here's how to approach the build.
You've priced a bolt-on manifold for your combo and it's either not made for your engine, or it costs more than the turbo did. So you start looking at steam pipe, because half the engine bays worth looking at have a manifold built from it and the raw material is a fraction of the price of a cast or tubular unit.
Steam pipe is a good call. It's also where a lot of first manifolds go wrong, because the material is forgiving and the process isn't. The pipe is the easy part. What separates a manifold that lasts from one that doesn't is how you plan the layout, how you sequence the welds, and whether you understood what the metal was going to do once it got hot.
Here's how to approach the build.
What Steam Pipe Actually Is, and Why It Works for Manifolds
Steam pipe is schedule-rated pressure pipe, and it's sized by nominal bore rather than outside diameter. That catches people out when they're used to ordering exhaust tube by OD. A "2 inch" 50N.B steam pipe measures 60mm across the outside and somewhere between 52 and 55mm through the bore, depending on schedule. Order it thinking in OD and you'll end up with runners a size off what you wanted.
What makes it the default for DIY manifolds:
- Wall thickness. SCH40 steel runs 3.6 to 5mm depending on size, SCH10 stainless runs 3mm, against about 1.5mm for mandrel exhaust tube. Thick wall means you can put real heat into a joint without blowing through it, and enough section to survive heat cycling with a turbo hanging off it.
- Buttweld elbows in matching bore. A formed bend without owning a mandrel bender. They come on a consistent radius in long (1.5D) and short (1D), made to the same nominal bore and schedule as the pipe, and they butt square against it end to end. Bore lines up with bore, so there is no internal step for the gas to trip over.
- Price. Steam pipe and weld elbows cost a fraction of stainless mandrel bends in equivalent sizes.
The trade-off is weight, and all of it hangs off the side of your cylinder head. That matters later.
We stock both SCH10 304 stainless and SCH40 steel steam pipe in 32, 40, 50 and 65N.B, with matching elbows in both materials.
Stainless vs Mild Steel Steam Pipe: Cost, Heat, Weldability, Longevity
This is the first real decision, and both materials work. They fail differently, cost differently, and ask different things of you as a welder.
Cost. SCH40 mild steel runs roughly half the price of SCH10 stainless for the same nominal bore. On a six cylinder manifold you're looking at twelve or more 90° elbows plus a metre of pipe, so that gap adds up to real money before you've bought a flange.
Heat. Two properties matter here and they pull in opposite directions. Mild steel conducts heat about three times better than 304 stainless (roughly 50 W/m·K against 16). A mild steel manifold sheds heat into the engine bay; a stainless one keeps it in the gas where it still has work to do driving the turbine. That's the real performance argument for stainless on a turbo car, and it's why almost every professionally built manifold is stainless.
The flip side is expansion. 304 grows about 45% more than mild steel for the same temperature rise (17.2 µm/m·°C against roughly 12). Every joint you weld into a rigid structure fights itself harder in stainless. That's the reason stainless manifolds crack in places mild steel ones don't.
The 304 caveat worth knowing. A turbo manifold spends most of its life sitting in 304's sensitisation range, roughly 425 to 870°C, where chromium carbides form at the grain boundaries and the material gives up some corrosion resistance and toughness right where the grains meet. 321 is titanium-stabilised specifically to avoid this and it's what most professional header builders spec, but 321 pipe and buttweld elbows in nominal bore sizes are hard to source and expensive when you find them. Plenty of 304 manifolds have done well over 100,000km, so this isn't a reason to avoid it. Just know you're accepting a compromise. And 316 isn't the fix: it's better in salt water, not better in heat.
Weldability. Mild steel is far more forgiving. MIG with ER70S-6 will get you a serviceable manifold and TIG is easy enough. Stainless wants TIG, 308L filler, tighter heat control, and an argon back purge on every single joint. Skip the purge and the inside of the weld sugars: a grey, crystalline, brittle oxide layer that flakes off in service. Those flakes don't stay put, and the turbine wheel is directly downstream.
Longevity. Mild steel scales. Every heat cycle grows an oxide layer, the layer flakes, and the wall slowly gets thinner. On a street car that's a decade-plus problem, on something running sustained high EGT it's faster. It rusts from the outside too, and header wrap makes that considerably worse by holding moisture against the metal. Ceramic coating is the right answer for mild steel, both for the heat retention you gave up by not using stainless and for keeping the outside intact. Stainless doesn't scale the same way and will outlast mild steel if the fabrication is sound. If it isn't sound, it cracks sooner, for the expansion reasons above.
Mixing the two on purpose. You don't have to pick one material for the whole manifold. A steel head flange with stainless pipe and elbows is a combination plenty of fabricators choose deliberately, and the reason is the two properties above working in your favour: steel expands less and sheds heat roughly three times faster, so a steel flange moves less while you're welding it than a stainless one does. Given that a warped flange is the most common cause of a cracked manifold, that is worth having. The dissimilar joint just has to be run with 309L rather than 308L. Steel flanges are cheaper, too.
The short version: first manifold, street car, budget is a factor, go SCH40 mild steel and ceramic coat it. Sustained high load, high EGT, or you want to build it once and forget about it, go SCH10 stainless and do the purge properly.
Sizing the Runners and the Collector
Runner bore should sit close to the exhaust port size at the head, or slightly under it. Oversizing costs you velocity into the turbine and slows spool, which is the same "bigger is better" trap people fall into on exhaust diameter. Our exhaust sizing guide covers why velocity matters as much as flow capacity, and the logic applies upstream of the turbo too.
Rough starting points by nominal bore:
- 32N.B (42mm OD, 35mm ID): small four cylinders and rotaries. 4G63, SR20, B and K series, 12A.
- 40N.B (48mm OD, 40mm ID): the most common size for turbo builds. 1JZ, 2JZ, RB25, 13B, most 2 litre four cylinders.
- 50N.B (60mm OD, 52mm ID): big sixes, LS V8s, 20B, high output builds. Also works as the collector on a 40N.B manifold.
- 65N.B (73mm OD, 63mm ID): collector and merge duty on serious power builds.
Your head flange sets the ceiling. Ours are cut to the actual port dimensions for each engine, so pick the flange for your engine first and match pipe to it rather than the other way round.
Tools and Skills You Need Before You Start
- TIG welder. DC only is fine for both steel and stainless. A 180 to 200A machine handles 3 to 5mm wall comfortably. MIG is acceptable for mild steel and not really acceptable for stainless.
- Bandsaw, portaband, or cold-cut saw. An angle grinder with a cut-off wheel will physically do the job, but square cuts get much harder and the error compounds through every joint downstream.
- Angle grinder with flap discs, files, and a carbide burr. The burr is for dressing the inside of every joint, which is the step most people skip.
- A flat welding table, or a piece of 12mm-plus plate. You cannot check a manifold for twist against a bench that isn't flat.
- Clamps, welding magnets, a digital angle gauge, a scribe and a wraparound template for marking cuts accurately.
- Argon purge setup if you're doing stainless: regulator with a low-flow gauge, purge plugs or dams, aluminium tape.
- A sacrificial backing plate with the same bolt pattern as your head flange, or a spare head, to bolt the flange to while you weld it.
On skills: you need to be able to lay a consistent, fully penetrating weld in awkward positions before you start on a manifold. This isn't the project to learn TIG on. Run practice joints in the same material and wall thickness first, which is what our weld practice kits are for. Stainless in particular punishes a good-looking cap sitting over a cold root.
Planning the Layout Before You Cut Anything
Start at the flange, not the turbo. Bolt the head flange to the engine, or to a spare head on the bench, and work outward from there.
Four things constrain where the turbo can actually sit:
Oil drain angle. This is the one people discover last and it wrecks otherwise good layouts. The drain has to run downhill to the sump, as close to vertical as you can manage and realistically within about 30 degrees of it, and it has to enter the sump above the oil level. Put the turbo somewhere the drain runs sideways or uphill and it'll push oil past the seals into the exhaust housing. You'll be chasing that smoke for months. Check your oil drain flange orientation as part of the layout, not after it.
Bonnet and chassis clearance. Measure with the engine sitting on its mounts, not on an engine stand. Then account for movement: 15 to 25mm of rock under load is normal on street mounts and more on soft ones. Whatever gap looks fine when everything's static has to survive that.
Heat proximity. Brake master cylinder, brake lines, wiring looms, steering shaft, coolant hoses. Anything within about 50mm of the manifold or turbine housing is a heat shielding job you've just signed up for.
Downpipe exit path. Plan the downpipe at the same time as the manifold. A layout that clocks the turbo beautifully and then leaves no room for the downpipe to clear the chassis rail is a manifold you'll be cutting apart.
Runner routing. Keep bends as gentle as packaging allows and don't stack two tight bends back to back. Long radius elbows flow better than short radius, so use short radius only where you've run out of other options. Equal length matters less on a turbo manifold than on an NA header, since the turbine responds to pressure and mass flow more than to wave tuning. Pulse separation is the thing worth protecting.
Twin scroll pairing. If you're running a twin-scroll turbo, the pairing has to follow the firing order so that cylinders firing back to back don't share a scroll. On a four cylinder firing 1-3-4-2, that's cylinders 1 and 4 in one scroll and 2 and 3 in the other. On an inline six firing 1-5-3-6-2-4 (RB, JZ, Barra), it's 1-2-3 in one and 4-5-6 in the other. Pair them wrong and you've built an expensive single scroll manifold.
Wastegate placement. This is the single most common flow mistake on home-built manifolds. The gate needs to pull a meaningful share of total exhaust flow, and it can only do that if it's teed into the collector at a shallow angle, facing the direction the gas is already going. A wastegate flange welded on perpendicular to the pipe is fighting the flow, and the result is boost creep: the gate sits wide open, boost climbs anyway, and there's nothing you can do about it in the tune. Aim for 30 to 45 degrees off the collector, downstream of where the runners have merged.
Mock the whole thing up before you cut anything you care about. Cardboard, foam, or a dummy turbo made from offcuts all work. Time spent here is time you don't spend cutting a nearly finished manifold apart.
Cutting and Mitering Steam Pipe
Buttweld elbows are what make steam pipe manifolds practical. They give you a formed, consistent radius that would otherwise take a mandrel bender. The skill is cutting them into the angles your layout actually needs.
- A 90° elbow cuts into two 45s, or into any pair of angles that sum to 90. A 45° elbow gives you two 22.5s. Between long and short radius in both angles, you can path almost anything.
- Pie cuts (slicing an elbow into wedges) let you build tight custom-angle transitions, but every pie cut is another weld and another leak path. Use them where you have to, not as a default approach.
- Mark with a wraparound template or a strip of masking tape wrapped square around the pipe. Freehand scribing looks accurate and puts you two degrees out, and two degrees compounds fast across five joints.
- Cut square. Every degree of error at a cut face becomes angular error in the finished runner, and you'll be chasing it at the turbo flange.
- Deburr the ID at every joint. The step almost everyone skips. A cut edge leaves a burr and the weld root adds more material into the bore, and all of that sits directly in the gas path on the pressure side of the turbo. Chamfer the outside edges to give the weld somewhere to go, then dress the inside back flush.
- Dry fit the whole runner before you weld anything. Butted joints hold themselves in position exactly not at all, so clamps, magnets and a few tacks are what carry a full mock-up. Check the path against the engine while it's still all in pieces.
Welding: Purging, Heat Control, and Keeping the Flange Flat
Tack everything before you weld anything. Build the complete manifold in tacks, bolt it to the engine, check every clearance, then take it off and weld it out. A manifold welded runner by runner pulls itself out of position and won't bolt back on.
Weld sequence controls distortion. Don't run a continuous bead around a joint. Work in opposing quarters: weld a section, then the section opposite it, then the two that split the difference. Let each pass cool before the next one. It takes longer and it's the difference between a manifold that sits flat and one that doesn't.
The flange is where most builds go wrong. A head flange is thick, the pipe is thin, and the heat needed to fuse the pipe is more than enough to pull the flange into a banana. Two things prevent it: bolt the flange to something thick and flat while you weld, whether that's a spare head or a piece of 20mm plate drilled to the same pattern, and stitch weld it in a staggered sequence instead of running around it. Check it with a straightedge and feeler gauge before you start and again when you're done, and if it moved, grind or machine it flat before it goes near the head. A warped flange torqued down flat is a manifold living under permanent preload, and preload plus thermal cycling is exactly how you get a crack at the runner-to-flange weld.
Purging, for stainless. Every joint in 304 needs argon on the back side. Plug the ends, fill the manifold at a low flow rate (you're displacing air, not blasting it out), give it time to purge properly before you strike an arc, and keep it flowing until the weld has cooled. Purge dams or soluble paper let you do one section at a time on a large manifold instead of flooding the whole assembly. You're after a clean, lightly coloured root. Grey and crystalline means it sugared.
Filler and heat input. 308L for 304 stainless, ER70S-6 for mild steel, and 309L any time the joint has one of each. Keep interpass temperature down on stainless and keep travel speed up: stainless holds heat, so a pace that feels normal on mild steel is overcooking it. Straw and blue tint on the outside is fine. Grey and chalky means you put too much heat in.
Full penetration on the structural joints. The head flange, the turbo flange and the collector joints carry load. A tidy cap over a cold root holds for a while and then lets go. If you can't get penetration from one side, bevel the edges and build it up.
What Actually Makes a Manifold Crack
Most failures we see come back to the same short list, and almost none of them are the material's fault.
- Nothing supporting the turbo. A turbo hanging off the end of a manifold is a heavy mass on a lever arm, vibrating at engine frequency and heat cycling at the same time. A support bracket from the turbo or collector back to the block or head is the highest-value thing you can add to the build, and it's usually an afternoon's work.
- A rigid downpipe. The engine moves on its mounts, the exhaust is bolted to the chassis, and the manifold is the thing in between. A flex bellow in the downpipe close to the turbo absorbs that movement. Without one, every bit of engine rock loads the manifold welds directly.
- A preloaded flange. Worth repeating, because it's the most common single cause. A flange that wasn't flat when you torqued it never stops being stressed.
- Cold roots and undercut. Both are stress risers, and cracks start at defects rather than in sound weld.
- Mixing materials with the wrong filler. Stainless pipe on a steel flange is a perfectly good combination, but it has to be welded with 309L. Run it with 308L and dilution from the carbon steel pulls the weld metal below the chromium and nickel it needs to stay austenitic, leaving martensite at the fusion line: hard, brittle, and sitting exactly where the load is. Match your studs and nuts to the job while you're at it.
- No compliance anywhere in the design. A fully triangulated manifold with no give has to absorb expansion somewhere, and that somewhere is a weld. A slightly longer runner with a gentle bend in it has more give than a short straight shot between two fixed points.
- Going straight to full boost on the first drive. Heat cycle it gently a few times first: up to temperature, fully cool, repeat. It's not a substitute for proper stress relief, but it beats shock-loading fresh welds.
A steam pipe manifold isn't difficult to build, it's just unforgiving about the steps that feel optional. Get the flange flat, support the turbo, purge your stainless, and the rest is fitting and patience. Everything the build needs is on the shelf: SCH10 stainless and SCH40 steel steam pipe with matching elbows in both materials, head flanges cut to port dimensions for most common engines, turbo and wastegate flanges to finish both ends, and studs and nuts in steel or titanium to bolt it up. Not sure what your layout needs? Send us the engine, the turbo and the flange you're working to, and we'll help you spec the pipe and elbow count before you order.