Centerless Grinding Delrin Guide Rods: OD Tolerance in Microns

Most shops will happily turn you a plastic rod. Very few will promise you one ground to a micron-class tolerance, and fewer still can mill features onto that ground surface without leaving a mark. The guide rod that Dylan Mercer ordered in May 2025 asked for all three: a slender black Delrin body, 180 mm long, centerless ground to Ø16.00 mm ±0.005 mm, with a milled eyelet end carrying an Ø8 mm transverse bore and a milled longitudinal flat along the body. His company in Australia builds finishing equipment for print media, where these rods guide a carriage that must glide with no measurable play — which is why the diameter tolerance, not the material, was the real specification.

Why not just turn it?

A good turning pass on acetal will hold ±0.02 mm and give you a surface that feels smooth. This part needed more for two reasons:

• The rod is long and slender (an 11:1 length-to-diameter ratio), and a slender part deflecting in the lathe chuck shows up as taper and lobing that no tool path can fully remove.
• The carriage bearings in Dylan's design run directly on the rod surface, so roundness and diameter consistency along the full length matter more than average size.

Centerless grinding removes both problems — the part is supported along its full length between the grinding wheel, the control wheel, and the workrest blade, with no chuck to introduce deflection.

Stage one — turning the blank

Each rod started as black acetal homopolymer bar stock, turned to Ø16.20 mm with the eyelet-end profile roughed in and both faces squared. The turning stage did the heavy lifting on geometry: the Ø8 mm eyelet bore was drilled and reamed while the blank was still held on the main axis, so the bore and the outer diameter shared one datum before grinding ever started. Tooling was sharp two-flute carbide with positive rake, and an air blast kept the chips from re-welding onto the surface — acetal chips carry static and will stick to the part if the air is not there.

Stage two — what centerless grinding added

The ground stage took the rods from Ø16.20 mm to Ø16.00 mm — only 0.20 mm of stock, but removed in a way that turning cannot match:

• Throughfeed grinding on a centerless machine with a 60-grit silicon carbide wheel, running under flood coolant.
• Stock removed in two light passes rather than one aggressive pass; plastics do not tolerate heat, and a single deep cut will melt and smear the surface instead of cutting it.
• The wheel was dressed every 400 pieces, and the first article after each dressing was checked for roundness and size before the batch continued.

The measured result across 2,000 rods: Ø16.000 ±0.005 mm held throughout, roundness 0.001–0.002 mm, surface finish Ra 0.4 µm — a surface that a bearing can ride on for years.

Stage three — milling on a ground OD without marking it

The milled features — the eyelet profile, the longitudinal flat, and the radial holes — all had to be cut after grinding, because grinding needs a full round surface to support the part. The challenge flipped: how do you hold a part with a finished ground surface without damaging the very surface you just paid to create?

• The rods were clamped in a precision V-block lined with a soft film, sized so that the clamp force was just enough to resist the cut — never enough to indent the OD.
• The flat was milled along the body in a single light climb pass with a sharp end mill, no re-cutting, to avoid smearing the acetal.
• All radial holes were drilled through the same fixture positions, so hole-to-flat relationships stayed inside ±0.1 mm.

Every tenth rod was checked for clamp marks under magnification. The answer was zero — the ground OD left the milling stage exactly as it entered it.

The tolerance chain (and the temperature trap)

One number nearly sank this job before it started, and it was not on the drawing: acetal expands about 100 µm per meter per degree Celsius — roughly five times the rate of aluminum. A 2 °C swing between the grinding room and the inspection room moves a 16 mm diameter by about 3 µm, which is more than half our total tolerance. Every measurement on this order was taken in a temperature-controlled room at 23 °C, after the parts soaked for 24 hours at that temperature. Grinding a part to microns is pointless if you then measure it in a room that disagrees with the grinding machine.

The problem we hit on the first fifty

The 50-piece trial run exposed a classic centerless artifact: the rods came off the throughfeed slightly tapered — about 0.006 mm smaller at the entry end — because the control wheel angle was steeper than the workrest geometry wanted. Nothing about the material caused it; the setup did. We reset the control wheel angle to 0.25 degrees, re-set the workrest blade height, and re-ran the trial: taper measured at 0.001 mm, and the full production run never moved off that number.

What the customer did next

All 2,000 rods passed Dylan's incoming check, and the carriage glide test on the first assembled unit measured zero detectable play. His note at order close: "The rods slide like they are riding on air, and the eyelet bores line up first time. We did not believe a plastic could be ground this consistently." A follow-on order for 5,000 rods, in two additional lengths, followed in December 2025.

Notes for anyone grinding plastic

• Grind in light passes with flood coolant — plastics melt before they burn, and a smeared surface will not be saved by polishing.
• Measure in a temperature-controlled room and let the parts soak; thermal expansion is the hidden variable on every plastic tolerance.
• Never clamp a ground OD in a bare steel jaw — film-lined fixtures or soft V-blocks are not optional on a finished surface.
• If a supplier quotes micron tolerances on plastic, ask what temperature they measure at. The answer tells you whether they understand the material or just the machine.

A plastic rod ground to microns sounds like over-engineering — until the part it guides has to move smoothly for a decade in a machine that gets serviced by feel.