Your frame will be exactly as square as the thing you built it on.
That sounds obvious written down. It gets ignored constantly, because the jig is the part of the project that produces nothing you can show anyone, and the temptation to tack a frame together on a garage floor and a couple of sawhorses is strong.
Don't. The floor isn't flat. It's the reference for everything you're about to build, and it slopes toward the drain.
How flat does flat have to be?
Tighter than the tolerance you want on the frame, and that's the whole specification.
For most chassis work, a surface flat to about 1 mm over its length is plenty. Suspension pickups want better than that, and half a millimetre relative to each other is a sensible target, because error there turns into bump steer and a car that won't track straight. What I'd avoid is chasing tenths of a millimetre on a welded steel table — you can't hold it without machining the top, and the frame is going to move further than that when you unclamp it anyway. The number that matters isn't the absolute flatness of the surface; it's how well you hold the pickups relative to each other, and that comes from measuring and shimming the fixtures rather than from the table being perfect.
Check what you've got before building on it. Stretch a length of string diagonally across the table and look for daylight under it at the middle, or run a long spirit level in several positions and rotate it end-for-end each time to cancel out the level's own error. A laser level is better if you have one. A water level made from clear hose is nearly free and surprisingly good over long distances.
Three ways to get a surface
Steel plate on a frame. Thick plate — 10 mm or more — on a well-braced base. Heavy, stable, weldable, and you can clamp anywhere with magnets. It'll want machining or careful shimming to be properly flat, and it's the most expensive route by material cost.
A welded table from rectangular tube. The cheapest thing that works. Build a rectangular frame from heavy box section, brace it well, then level the top surface by shimming the legs rather than by trying to get the weldment perfect. The key is making it rigid enough not to twist when you clamp to it, which means more cross-bracing than feels necessary and feet that don't rock.
A modular fixture table. Certiflat, BuildPro, Fireball and others sell tables with a grid of holes for pins and clamps. They're expensive and they're the right answer if you're going to build more than two frames, because repeatability is free afterwards. The holes also give you a coordinate system, which is worth more than it sounds.
Whichever you pick, the surface has to be rigid, not just flat. A flat surface that twists when you clamp a tube to it is a flat surface briefly.
The coordinate system is the real deliverable
Here's the part that connects the jig to the model, and it's the bit people skip.
Pick an origin on the table and mark it permanently. Scribe two perpendicular reference lines through it, confirm they're square with a 3-4-5 check over the longest distance the table allows, and centre-punch stations along both at a known interval.
Make that origin the same origin as your model. That's the whole trick. Every coordinate out of FrameDesigner then means something physical, and setting up a fixture to hold a joint at (640, 310, 780) becomes a measurement rather than an interpretation.
If your table origin and your model origin don't match, every setup needs arithmetic, and arithmetic done forty times at a workbench produces at least one error.
Fixturing the hard points
You don't need to fixture everything. Fixture the things that have to be right and let the rest follow from the tubes.
In practice that means the suspension pickups, the engine and gearbox mounts, anything that bolts to a subframe, and the main hoop base positions on a cage.
Build each fixture to locate one point in all three axes and nothing else. A post off the table to the right height with a pin at the right position beats an elaborate cradle, because a simple fixture is easier to verify and easier to shim when you find it's 0.4 mm out.
Verify every fixture before any tube goes near it. Measure each pin position relative to the table origin, write the number down, compare it against the model. Finding a fixture error at this stage costs five minutes. Finding it after the frame is tacked costs the frame.
Think about getting the frame out
Really. People weld a frame into a jig they then can't extract it from, usually because a fixture passes through what became a closed triangle.
Before you tack, look at each fixture and ask which direction it leaves in once the structure is complete. Anything that has to come out sideways through a braced bay needs to be removable in pieces, or it needs to not be there.
Where the jig stops helping
A jig holds position while you tack. That's its job and it does it well.
It does not stop distortion. Clamping a frame rigidly and welding it out doesn't eliminate shrinkage — it converts movement into residual stress, which then comes out as spring when you unclamp. The frame moves after you thought you were done, and you have no reference left to measure against.
So tack in the jig, measure, then release the rigid clamping for finish welding and work with the frame supported rather than restrained. Keep measuring as you go. Sequence and distortion covers the welding side of this in more detail.
One more thing, and it's the cheapest insurance in the shop: before you remove the frame from the table, record every critical dimension. Diagonals, pickup-to-pickup distances, overall widths at several stations. Those numbers are your only evidence of what the frame was when it was known-good. If it moves later, they tell you by how much and where.
Comments
Loading comments…