The photo that arrives most often is a cut edge, not a machine. Someone has just finished a 2 m gantry router build, run the first test part, and there is a faint mark across the surface at regular intervals - every time the gantry passes a particular point on the X axis. Nine times out of ten that mark is the joint between two rack sections, and nine times out of ten the rack itself is fine. It is the alignment that is out.
What follows is how we set rack on a 2 m gantry in our own shop, in order, including the two places people skip. It assumes a steel rack - ours are S45C or 42CrMo, quenched to HRC48-60, with a tooth error held to around ±0.01 mm/m at DIN6-8 - and a pinion driven by a stepper or servo through a gearbox. If you are working with a different grade the numbers shift slightly, but the sequence does not.
You are aligning the pinion, not the rack
This is the thing that costs people a weekend. The rack is a reference surface, but the pinion is what has to sit right against it, and the pinion lives on a plate that is bolted to a moving gantry.
So the question is not "is the rack straight" - it is "does the pinion centre height stay constant as the gantry travels the full 2 m". If the gantry plate is 0.1 mm lower at one end than the other, you get exactly the same symptom as a badly joined rack: a periodic bump, and a tooth loading pattern that is heavier at one end. We have seen customers re-shim the rack three times chasing a problem that was in the gantry plate.
Check the pinion centre height at both ends of travel before the rack goes on. It takes twenty minutes with a height gauge and a straight edge, and it tells you whether the rack is even the problem.
The mounting surface is doing more work than the rack
A rack is a long, thin, hardened bar. It will follow whatever surface you bolt it to. That is convenient if the surface is flat and a disaster if it is not, because the rack does not resist - it just conforms, and then the tooth line goes out of parallel with the rail.
Two practical points here:
- Machine the rack mounting face, do not shim it flat. Shimming under a rack is fine for a 300 mm section on a small machine. Across 2 m you end up with a stack of shims of varying thickness, and the rack will bow between the fixing points even if the shim stack measures correctly at each point. If the face is out, have it machined.
- Deburr the tapped holes. A raised burr around a fixing hole holds the rack off the surface by a few hundredths at that point, which is a local pitch error. Nobody checks this. It is the single most common cause of a tight spot in the mesh.
Fixings should be on a consistent pitch and nipped up in sequence from one end, not tightened fully at one end and then worked along. On a 2 m run we normally see six to eight M6 fixings per rack section depending on section length; the exact number comes from the drawing, but the sequence matters more than the count.
Set the first section, then use it as your datum
Everything downstream is measured against the first section, so get it right and do not rush it.
Bring the gantry to one end of travel, engage the pinion lightly and let it set the pinion's position along the rack. Then check two things at that end:
- Tooth line parallel to the rail. Dial against the tooth flank - not the rack side face, which is not always a precision surface - and run the whole section. You are looking for the same reading at both ends.
- Backlash at the mesh. Plunge a dial indicator onto a pinion tooth, lock the pinion, and rock the gantry by hand. Typical target on a gantry axis is in the region of 0.05-0.15 mm at the pitch circle; the drawing for your pinion will state it. Too tight and you get noise and heat, too loose and you get reversal error that shows up as a witness mark on every direction change.
Only when the first section is settled should you drill anything for the second.
Joining the second section without a step
This is the one that fixes the mark on your test part.
Rack sections are supplied with machined end faces, and the joint is supposed to close metal to metal with no visible gap. In practice you shim the second section until it lines up with the first, and the reference you use is a dial gauge riding on a pinion tooth across the joint - not a straight edge along the back of the rack, and not a set of feeler gauges at the end face.
Run the gauge along the tooth line from well before the joint to well after it. The reading should be flat. If there is a step, move the second section toward or away from the rail, and re-check. A step of a few hundredths is enough to produce a visible bump on a cut surface at every pass - you have 2 m of travel and the joint gets hit a few hundred times a minute at typical feed rates.
A joining pin, where the drawing provides one, locates the two sections axially. It does not take out a height step. People sometimes assume it does and skip the shimming step.
The pinion side is half the mesh
Almost all the attention goes to the rack, and then the pinion is bolted on with whatever came out of the box.
The pinion has to match the rack in module and pressure angle, and it has to be sized with the drive so that the linear speed and thrust you need are actually available. A motor drive gear that is one module out from the rack will run, badly, and the failure will look like a rack problem.
There is also a hardness question. Our racks run HRC48-60. If the pinion is considerably softer, the pinion takes the wear and the mesh opens up over a few months - which is fine if you are happy replacing pinions, and not fine if the machine is in production. For a 2 m gantry fed at high rates, a hardened pinion is the cheaper choice over two years even though it costs more on the invoice.
If the module and tooth form are not settled yet, that decision comes before alignment - you cannot align a rack you have not chosen correctly. We keep a guide to choosing a rack by module, teeth and section length alongside the product range, and it is the place to start if you are ordering for a new build rather than fitting what you already have.
Lubrication, and why the joint wears first
Grease the rack teeth and the pinion every 100-200 operating hours. Lithium-based grease or a proper gear grease, applied to the tooth flanks, not sprayed at the rack from a distance in the hope that some of it lands.
Two adjustments to that interval:
- Dusty cutting shortens it. Plasma, laser and MDF all put abrasive dust into the mesh. On a plasma table we would halve the interval, and on any machine with a downdraft table we would check the mesh weekly.
- An unlubricated rack wears at the joints first. It is not obvious why until you watch it: the joint is where the tooth loading is least consistent, so if the mesh is dry, that is where the metal starts moving. By the time the wear is visible in the middle of a section, the joint has been wearing for a while.
Also grease both flanks. It is tempting to load the driving flank only, but a gantry decelerating a heavy Z axis loads the other side regularly.
Before you cut the first part
Five checks, in this order, and you will not be chasing a mark on the surface later:
- Pinion centre height constant at both ends of travel.
- Tooth line parallel to the rail along the full run.
- No step at the joint, verified with a dial gauge on a tooth, not a straight edge.
- Backlash at the mesh set to the pinion's published figure, then a hand-rock test at three positions along the axis - near, middle and far.
- Every fixing tightened, in sequence, and re-checked. A single loose fixing in the middle of a 2 m section changes the tooth line locally.
One more habit worth building: mark the joint positions on the machine frame after alignment. Rack wear shows at the joints first, and if you know where they are you can inspect those two or three places in five minutes instead of running a dial along 2 m.
If you are sizing rack and pinion for a new gantry, send the axis length, the cutting force and the feed rate you need - modules from 1.5 up to 12 are standard for us, and the matching pinion and drive come with it. You can send that through here and we normally come back with a suggestion the same working day.
Questions we get on the phone
Do I need ground rack for a gantry router?
Depends what the machine cuts. For wood, MDF and foam, a standard rack is usually enough and the money is better spent on getting the mounting face machined. For aluminium or anything where the cut edge is inspected, go ground - the difference in pitch error is the difference between a witness mark and no witness mark.
Can I use one long rack instead of joining sections?
Beyond about 1 m in a single piece, freight and handling become the problem rather than the engineering. Two sections joined properly run as well as one long section; two sections joined badly do not, which is why the joining step gets its own section above.
My rack is noisy at high feed. Is that backlash?
Usually it is too little backlash, or the pinion is running dry. Backlash that is too tight loads both flanks at once and the mesh howls. Check lubrication first, then re-measure backlash, then look at pinion centre height.
How often should I expect to replace a rack?
On a well-lubricated gantry cutting wood, we have customers running the same rack for years. On a plasma table with a dry mesh and abrasive dust, tooth wear becomes visible much sooner. The variable is maintenance, not the rack.
