Soft-jaw fixture and CNC process plan

The brief

An individual assignment in five parts:

• design soft jaws that hold precision bars of Ø30–50 mm vertically in a machine vice, with repeatable loading;
• design a modular fixture to machine the jaws in batches on a 700 × 700 × 50 mm aluminium base plate;
• program the machining in SolidCAM, with G-code, removal rates, a plan to halve the cutting time and the energy used;
• plan conventional and non-conventional routes for a keyed bush, and analyse sampling plans and a defect problem;
• run a life-cycle assessment of a jaw extruded and machined in Shanghai and shipped to London.

Lab tasks on pin capability, surface finish and CMM inspection completed it.

Soft jaw

The jaw is a 100 × 28 × 37 mm block of 6063-T6 aluminium weighing 241 g. A 90° V-groove, 20.08 ± 0.05 mm deep and centred 50.0 ± 0.1 mm from the end, runs along its height, so a bar stands upright and is pushed back onto the same centre plane each time it’s clamped. The bar centre sits 21.2 mm (Ø30) to 35.4 mm (Ø50) from the V’s vertex. The rear mounting face, Datum A, sits against the vice jaw, and two countersunk M6 clearance holes at 60 mm centres match the vice’s hole pattern.

On the drawing, the V-faces and Datum A are Ra 3.2 µm, with the V-faces parallel to A within 0.030 mm. All other machined faces are Ra 6.3 µm, with general tolerances of ±0.1 mm and ±0.5°.

(opens full size)

Batch fixture

The plate holds 21 blanks in a 3 × 7 layout, so the V-groove can be machined across all of them in one loading. Each blank starts as 104 × 42 × 31 mm stock. At each station, three Ø12 hardened dowel pins locate it (two on the long side, one at the end) and two low-profile M10 toe clamps push it onto them: 63 pins and 42 clamps across the plate.

The jaw takes three setups on a 3-axis machine:

1. Face the top, profile the outline and cut the V-groove.
2. Flip the part and face the bottom.
3. Turn it onto the hole face, then spot drill, drill and countersink.

A 5-axis machine could do everything except the clamped bottom face in one setup, cutting it to two.

(opens full size)

(opens full size)

CAM and G-code

I programmed seven operations in SolidCAM, in three coordinate systems with six tools: a Ø50 face mill, a Ø10 end mill, a 45° taper mill for the V, a spot drill, a Ø6.4 drill and a 90° countersink. To cut the time, I raised the profiling feed to 1,200 mm/min and the step-down from 1 mm to about 5 mm, which brought profiling down to 2 min 31 s and the whole jaw to 8 min 58 s, all machined dry. Machining one jaw removes 125 g of aluminium: about 1.9 kWh at the brief’s 15 kWh/kg, or 1.1 kWh at 7.5 kW for the cutting time.

(opens full size)

(opens full size)

Process routes for a keyed bush

The part is a Ø60 × 80 mm cylinder with a Ø30 bore and an internal keyway, turned from Ø65 bar.

Conventional route

Face, rough and finish turn to Ø60, centre drill, drill and bore to Ø30, part off and face to length, broach the keyway (or slot it), then deburr and inspect.

Feed for the finish

For Ra 3.2 µm at 2,000 rpm with a 0.02 mm nose radius, f = √(32 × Ra × r) = 0.045 mm/rev, about 90 mm/min.

Hardened tool steel

Wire or sinker EDM for the keyway. For additive manufacture, laser powder bed fusion or DMLS gives a near-net blank, which still needs heat treatment and finish boring, grinding or EDM.

Quality

Sampling plans

I drew operating-characteristic curves for samples of 10, 20 and 30 from lots of 1,000, accepting up to one defective. The n = 10 plan suits the producer, accepting 95% of lots up to 3.7% defective. The n = 30 plan protects the customer: lots that are 12.4% defective are accepted only 10% of the time.

A new material

C-charts for a new “super plastic” against the standard one showed 93 cracks against 41 over 20 days, 127% more: stable, but at a worse level. I recommended not approving full production until brittleness, moulding temperature, cooling rate and residual stress had been investigated.

Lab measurements

Taken in a group session and analysed myself:

Pins

Diameter Cp 1.22 and Cpk 1.03; length Cp 0.11 and Cpk −0.08. Not ready for mass production.

Surface finish

Doubling the face-milling chip load from 0.05 to 0.10 mm/tooth raised Ra from 0.30 µm to 1.21 µm.

CMM

45 dimensions on 17 features: 33 in tolerance and 12 out, so the block was rejected.

(opens full size)

(opens full size)

Sustainability

My OpenLCA model (CML method, cut-off) gave 4.23 kg CO₂e per delivered jaw. I recommended recycled aluminium, a smaller stock allowance, more efficient toolpaths and recycling the swarf. The next section explains why I no longer trust that total.

Looking back

I answered every task with the working shown, and the numbers reproduce. Cutting the V-groove and holes together from the clamping-face side, all referenced to Datum A, was the right call in the CAM. But reviewing it now, the sections don’t always agree with each other, and some results aren’t as solid as they look.

• The LCA total isn’t reliable. More than half of it (2.41 kg CO₂e) came from a container-glass dataset that had been linked into the aluminium and electricity processes by mistake. I also used UK electricity, although the brief put the extrusion and machining in Shanghai, and a 1 MJ placeholder for aluminium extraction. The lesson: read the contribution tree before trusting the total.
• The fixture and the CAM conflict. The Setup 1 profiling runs round all four side faces, exactly where the dowel pins and toe clamps bear. My fixture section put the V-groove in Setup 1, while the CAM cut it in the hole-drilling orientation. In Setup 2 a toe clamp sits over the finished V, and the through-drill breaks 3 mm into the plate with no relief hole. Drilling the jaw’s own mounting holes first, then bolting it down through them, would free all four sides for profiling.
• The batch isn’t in the program yet. The G-code machines a single jaw; running all 21 stations needs a work offset for each one.
• The biggest time saving was elsewhere. I didn’t derive speeds and feeds from the brief’s cutting-parameter table, and I left no 0.5 mm finishing allowance on the toleranced faces. The V-groove takes 3 min 58 s, 44% of the cycle, because its second pass repeats all 21 depth levels. Removing that repeat would save about 2 minutes, more than my profiling change.
• The GD&T doesn’t say what I meant. Read literally, a parallelism tolerance on faces at 45° to Datum A can’t be met. An angularity or profile tolerance with a secondary datum would express the intent: keep the groove square so the bar stands vertical. My critical tolerances were also about IT10–12, against the brief’s IT7–8.
• The statistics needed more care. I compared crack counts only; across all five defect types the increase was 25% (241 against 193), with cracks driving it. For Cp and Cpk I used the overall standard deviation on a process whose range chart was already out of control, and four short pins dominated the length result when they should have been re-measured. Two of the largest CMM “failures” look like errors in the drawing’s nominal positions rather than in the part.
• The feed formula was outside its range. Ra = f²/32r assumes the feed is less than twice the nose radius, but here 0.045 mm is more than 0.04 mm. That’s the brief’s nose radius; a realistic 0.4 mm insert gives about 0.20 mm/rev by the same formula.

What I’d do differently

• Bolt the jaw down through its own pre-drilled holes for the profiling and V-groove, and keep all the Datum A features in one setup.
• Leave a 0.5 mm finishing pass on Datum A and the V-faces, derive speeds and feeds from the cutting table, and remove duplicated passes before tuning feeds.
• Program the batch with a work offset for each station.
• Specify angularity or profile with an A|B datum frame, at IT7–8 on the V’s depth and position.
• Re-run the LCA with the right datasets and a Chinese grid mix, and add a sensitivity case for recycled versus primary aluminium.
• Look at every defect type before focusing on one, check stability before calculating capability, and re-measure outliers.