CNC Machining of Hand-Board Models

• 1
  What a hand-board model actually is
• 2
  From hand board to a machinable solid
• 3
  How the part is held and cut
• 4
  What accuracy you can realistically hold
• 5
  Material choices that behave like the hand board
• 6
  Where this route stops making sense
• 7
  Step by step: from mockup to first article
• 8
  Which route fits which hand-board job

In appliance, instrument and consumer-electronics work, the hand board is the first object you can hold. It is usually made by a model maker from sheet plastic, foam, wood or urethane, cut and glued by hand. Its job is to settle shape, grip, button position and assembly order before anyone pays for a steel mold. That is why the hand board carries decisions that a screen cannot carry: reach, weight, flashing, how a 2 mm gap feels in the palm.

The hand board is not a drawing. It is a solid with real surfaces, and that changes what you can do with it. A hand sketch has no scale you can trust. A hand board has a scale you can measure. Once you have measured geometry, you can digitize it and cut copies in metal or engineering plastic.

CNC machining of hand-board models is the step where that handmade object stops being a single item and becomes a repeatable part. It is a translation job, not a copy job. The model maker's 1.5 mm glue line and sanded corner become a fillet radius and a wall thickness in CAD, and those numbers have to survive machining.

This matters most in small runs. If you need 5 to 50 units for fit checks, field trials or a trade show, machining the digitized hand board is often faster and cheaper than cutting a tool. If you need 100,000 units in ABS, the hand board was never the right end product.

• 1
  Hand boardHand-cut mockup for shape and feel. One-off, low dimensional accuracy.
• 2
  Digitized modelScanned or measured solid with defined walls, fillets and datums.
• 3
  Machined partCut from solid stock to the digitized model, repeatable unit to unit.

The first real work is capture. A structured-light or laser scanner gives you a point cloud in minutes, but a point cloud is not a model. Someone still has to fit planes, cylinders and fillets to it, then rebuild the sharp edges the scanner rounded off. Hand-sanded surfaces come back as wavy patches, and if you machine those waves directly you get a part that looks hand-made, because it is.

For flat-sided hand boards, calipers and a height gauge are often better than a scanner. Measure the datums, model from the datums, and rebuild the freeform surfaces only where the shape actually matters. This keeps the CAD tree clean and makes the drawing easy to tolerance. On a 300 mm enclosure, measuring six planes beats rebuilding 400,000 scan points.

Decide early which surfaces are functional and which are cosmetic. A mounting boss position is functional; the sweep of a handle is usually cosmetic. Functional surfaces get a real tolerance. Cosmetic surfaces get a surface finish callout and a generous profile tolerance. Mixing the two is the most common reason a first article fails.

Then close the solid. Hand boards are often hollow shells glued from panels, so the CAD model needs walls that can actually be cut. A 0.8 mm wall on a 200 mm part will chatter and bow. We normally ask for 1.5 mm minimum on plastics and 2 mm on aluminium for parts this size, unless the geometry gives us support.

Finally, add machining allowance where you can. If the hand board is 1 mm undersized because of sanding, machine to the CAD value and let the part be right. Never machine to a worn physical master.

Most hand-board geometry is thin, curved and has no flat back. That is a fixturing problem before it is a cutting problem. The usual answer is a soft jaw or a machined pocket that matches the part's outer surface, plus a small amount of stock left for the second setup. If you skip the pocket, the part moves, and a moving part gives you a taper you cannot see until inspection.

For a one-sided part, 3-axis milling is enough. The tool comes down from one direction; pockets, holes and the outer profile all cut from the top. Cycle times are short and programming is cheap. This covers a large share of hand-board work, especially flat control panels and cover plates.

When the part has features on four sides, a 4-axis machine with a rotary table saves setups. When it has features on five sides, or undercuts on a curved surface, a simultaneous 5-axis center is the honest answer. The tool stays normal to the surface, which also lets you use a shorter cutter. Short cutters deflect less.

The trade-off is programming time. Five-axis toolpaths take longer to prove out, and the first part usually needs a trial cut in a soft material. On a part that measures 60 mm, that overhead may not pay back. On a 400 mm housing with a curved split line, it usually does.

Keep the datum scheme simple. Three planes and two holes are enough for most parts. If your datums sit on a hand-sanded surface, they will move between setups, and no amount of machine accuracy will fix that.

There is a gap between what a machine can do and what a hand-board part can accept. Our machines hold ±0.005 mm on rigid, well-fixtured work. A thin digitized shell with 1.5 mm walls will not hold that, because the part deflects under the cutter and moves when you unclamp it. The machine is not the limit. The part is.

A workable split for a 200 mm to 400 mm hand-board part: functional holes and mating faces at ±0.02 mm, profile of the outer shell at ±0.15 mm, and cosmetic surfaces free within ±0.3 mm. Those numbers are honest and they are cheap to inspect.

Surface finish follows the same logic. As-machined aluminium sits around Ra 1.6–3.2 μm. Fine finishing reaches Ra 0.2–0.8 μm, but every step costs cycle time. On a textured housing, a bead-blasted Ra 1.6 μm surface hides tool marks better than a polished one, and it is far cheaper.

Do not tolerance a hand-board part tighter than the hand board itself was made. If the original mockup was sanded to ±0.5 mm, a ±0.02 mm callout on the cosmetic surface only creates scrap and argument. Put the tight tolerance where a bearing, a screw boss or a connector sits.

• 1
  Datum facesMachine first, from stock. Everything else references them.
• 2
  Mating holesDrill and ream after the profile so the wall does not spring.
• 3
  Cosmetic shellRough, semi-finish, then a single light finishing pass.

The hand board's material is usually chosen for how easy it is to cut by hand, not for how the product will be made. Usually you should not copy it. If the production part is ABS, machining ABS gives you the closest feel, but ABS machines poorly: it gums, it burns, and thin walls flex. POM or PC machines far better and gives a stiffer part.

For appearance models, aluminium 6061 is the default. It cuts cleanly, takes anodizing in clear or colour, and holds a sharp edge. If the part must be light, 7075 gives higher strength at similar weight but costs more and machines slower. For a housing that will be handled, 5052 and 5083 bend and dent less than 6061.

For functional testing under load, 17-4PH stainless in the H900 condition or 4140 steel are the usual picks. Both machine well and both can be heat treated later. Titanium TC4 (Ti-6Al-4V) is used when weight and corrosion matter together, but expect longer cycle times and more tool wear.

If the part will sit near electronics, beryllium copper and C110 copper give you conductivity, but they are gummy and need sharp tooling. For medical or cleanroom mockups, 316L and PEEK survive repeated cleaning. We keep 6061, 7075, 304, 316L, 17-4PH, POM, PC, ABS and PEEK in regular stock.

One warning about plating. Electroless nickel and hardcoat anodizing add 10 μm to 50 μm per surface. On a 2 mm wall that changes the fit. Tell us the finish before we set the dimensions, not after.

CNC machining of hand-board models is a small-batch tool. It gets you a few dozen accurate parts fast, with no tooling cost and no minimum order quantity. What it does not do is scale. If you need 5,000 identical housings, the per-part cost of milling will stay high while injection molding drops sharply once the tool is amortized.

It also struggles with hollow shells. A hand board is often a shell with 1 mm walls and internal ribs. Machining that from solid means removing most of the material, which takes time and risks distortion. Vacuum casting or 3D printing handles hollow geometry better. For a shell under 3 mm wall, we usually suggest casting the outer skin and machining only the interface points.

Deep narrow pockets are another limit. A pocket deeper than four times its width needs a long, thin cutter that deflects. You can reach it with 5-axis and a bull-nose cutter, but the floor may need a separate finishing pass, and the corner radius is set by the tool, not the drawing. If your design needs a 0.5 mm internal corner at 30 mm depth, no milling process will give it to you at reasonable cost.

Very large parts change the math too. Our maximum processing size is 4,000 mm, and the large travel is 4,000 × 400 × 150 mm. Beyond that, you are looking at a different process. Within it, size mostly costs cycle time, not accuracy, as long as the part is supported.

Finally, if the hand board is still being changed every week, do not machine it yet. Cut the changes in the mockup first. Machining a moving target produces expensive scrap.