Sinking and raising a dome

With a blank chosen, the dome is made in two kinds of work. Sinking hammers the middle of the blank down into something that gives, a sandbag or a hollow in a block, so that the metal stretches and thins. Raising works the edge over a stake from the outside, so that the metal is pushed together and the circumference shrinks. Most domes use both: sink first, while the metal is soft, to get depth and stretch the radial lines; raise afterwards to bring the rim in to size and the profile to the template.

This chapter turns HF-01’s blank into a sequence of stages with times, measures each raising course by what it does to the rim, places the anneals, and shows how the template caught a forming habit the 1918 trade text listed in the references warns about: striking too hard at the centre.

The tools

The traditional tools for each operation are few and simple.

• For sinking: a leather sandbag filled with sand or shot, or a hardwood block with shallow hollows of different sizes turned or carved in its end grain; a bossing mallet or a round-faced hammer. The old texts also mention cast lead blocks. Lead is poisonous, marks aluminium with embedded particles, and has no place in a modern workshop; polyurethane blocks and sandbags do the same job.
• For raising: a raising stake or a mushroom-head stake, held in a vice or a bench plate; a raising hammer, with a narrow, rounded face, or a wooden or nylon mallet for soft aluminium.
• For planishing: a planishing hammer with a polished, slightly domed face, and a polished stake or dolly whose curvature is close to the dome’s 200 mm radius.
• For checking: the template from Chapter 3, a straightedge and rule for depth, dividers and a flexible rule for the rim.

The sequence

This is an illustration. The workshop’s plan for each of HF-01’s domes, from its records of similar work, is set out below. Times are per dome and are what the 5.0-hour estimate in Chapter 1 was built from.

Figure 1Forming sequence for one HF-01 dome

1. 1Prepare, 0.25 hmeasure blank thickness, mark rolling direction, deburr the edge, mark centre and guide circles
2. 2Sink, 0.75 hon the sandbag, in spiral courses from the centre out, to about 40 mm deep
3. 3Anneal, 0.25 hto the supplier's practice, then cool
4. 4Raise courses 1 and 2, 0.8 hon the mushroom stake, working from the shoulder out to the edge
5. 5Anneal, 0.25 has before
6. 6Raise courses 3 and 4, 0.8 hthe same, finishing to the template
7. 7Planish, 1.5 hlight, overlapping blows over the polished stake until no marks remain
8. 8Trim and check, 0.4 hscribe the rim line, trim, file, gauge thickness, check template both ways

The stages add up to 5.0 hours. Two things in the sequence are worth explaining. Sinking comes first because it stretches, and soft metal stretches best; once the metal has been hardened by raising, further sinking is harder and thins it unevenly. And planishing, at 1.5 hours, is the single longest stage, because the finish the customer sees is made there, not in the forming.

Sinking

The blank is laid on the sandbag and struck with the bossing mallet, starting at the centre and working outwards in a spiral, turning the blank a little after each blow so that the blows overlap evenly. Each blow pushes a small area of metal into the bag and stretches it. Several spiral passes take the blank to depth.

Two things go wrong in sinking. Blows concentrated in one place stretch that place more than its neighbours, producing a local thin spot or the ring-shaped thin band that Chapter 2’s exercise found. And the blank’s edge, which is not struck, starts to wrinkle as the middle sinks and pulls it inwards; light wrinkles are normal and are removed by raising, but deep ones should be eased out as they form, by turning the blank over and tapping them flat on a flat stake.

This is an illustration. The worker sinks HF-01’s blank until a straightedge laid across its edge stands about 40 mm above the bottom of the hollow. At that point, the edge, which started at 352 mm, measures about 344 mm across, because the sinking has drawn it in.

Raising

Raising works from the outside. The partly formed dome is held on the mushroom stake, top down, with the stake’s head touching the inside of the dome a little inside the region to be worked. The worker strikes the outside of the dome just beyond the point of contact, so that the metal between stake and blow is pushed down onto the stake and, being on a curve, is compressed around the circumference. Working round the dome in one full circuit at one distance from the top is a course. Each course moves outwards from the shoulder towards the edge.

The 1918 text, describing raising a cover on a block, gives the timeless advice: work in courses, turn the work a little with each blow, and do not strike too hard at the centre. The raising itself shrinks the rim, and the rim’s diameter after each course is the easiest measure of progress.

Figure 2Diameter of the formed edge after sinking and each raising course

Cut blank352 mm

After sinking344 mm

After course 1339 mm

After course 2334 mm

After course 3330 mm

After course 4327 mm

The target of 327 mm comes from the geometry. The trim allowance continues the sphere 6 mm beyond the rim along the surface, which at a radius of 200 mm is an extra angle of 6 / 200 radians beyond the rim’s 53.1°, putting the edge at a radius of 200 × sin(54.85°) = 163.5 mm, a diameter of 327 mm. From the 352 mm cut blank, the edge’s circumference shrinks by 7.1% in total, a little more than the net blank’s 5.9% because the trim ring is shrunk too.

Choosing the stake

The stake’s head should be more sharply curved than the dome, so that it touches the inside of the dome along a narrow band and the metal between that band and the hammer blow has somewhere to go. For HF-01’s 200 mm sphere radius, the worker used a mushroom stake with a head radius of about 75 mm for raising, and changed to a polished head close to 200 mm for planishing, where the aim is the opposite: as much contact as possible, so that each blow flattens rather than reshapes. Using the planishing stake for raising makes the courses slow and ineffective; using the raising stake for planishing leaves a dimpled surface that follows the stake rather than the sphere.

Wrinkles and laps

Every course compresses the edge, and compressed sheet wants to wrinkle. A good course keeps each wrinkle small and closes it on the stake before moving on, so that the compression goes into thickness. If a wrinkle is allowed to grow and is then struck flat, it folds over into a lap, two layers pressed together with a seam between them. A lap is a defect that cannot be planished out; in the trim allowance it can be cut away, but in the dome itself it means a scrapped part. The rule is simple: never strike a wrinkle that has folded over; ease it back open first, or stop and anneal.

Annealing

Soft aluminium hardens as it is worked. After sinking and after two raising courses, the dome rings noticeably under the mallet and springs back from the stake, and further work risks cracks at the edge. The workshop anneals at those two points.

Annealing practice must come from the metal supplier’s data for the alloy and temper, which gives a temperature and a holding time. Small workshops without an oven often heat the part evenly with a broad, soft flame and use a temperature-indicating crayon rated for the supplier’s temperature, marked on the part, to know when to stop. An older trade method blackens the part with soot from an acetylene-only flame and heats it until the soot burns off; it is a rough indicator at best and is no substitute for the supplier’s figure. Overheating some aluminium alloys damages them, and hot aluminium shows no colour change, so a part that looks cool can burn a hand. The anneal goes on the job record with the method used.

Planishing

Sinking and raising leave the dome’s surface covered in small facets and dents. Planishing smooths them with many light, overlapping blows of a polished hammer on the outside of the dome, held over a polished stake whose curvature closely matches the dome’s. Each blow flattens a facet against the stake. Planishing stretches the metal very slightly at each blow, so it also refines the shape; a worker can lower or raise a region a little by planishing more or less there.

This is an illustration. Planishing takes 1.5 hours per HF-01 dome, worked in rings from the top outwards, then repeated in a crossing pattern. It is checked under raking light, a lamp held low to the surface so that any remaining facet casts a shadow. Chapter 8 discusses the acceptance of finish.

The error: an over-sunk centre

This is an illustration. On the first dome cut from a 352 mm blank, after the scrapped 340 mm one of Chapter 3, the worker sank the blank with extra blows at the centre, wanting to be sure of the depth. After raising courses 1 and 2, the template, laid across the dome, rocked on the top: the dome stood about 88 mm high at its centre against the template’s 80 mm, while the shoulder at 80 mm from the axis was close to the template’s 63.3 mm. The top was over-sunk, too peaked for the sphere.

This is the mistake the 1918 text warns of, the curve coming out deeper than required from striking too hard at the centre. It matters for two reasons. The shape is wrong, outside the ±2 mm tolerance. And the peak has had more stretching than the plan allowed, so its thickness is lower; the worker’s ultrasonic gauge read 1.34 mm at the top, still above the 1.3 mm limit but with almost no margin left.

An over-curved region has too much area for its surroundings. It cannot be pushed back down without somewhere for that area to go. The worker’s correction was to leave the top alone and work the shoulder: light planishing over a stake of slightly smaller radius in the band between about 60 and 120 mm from the axis, which stretched that band a little and let the top settle as the shoulder came up to meet the template. After the remaining raising courses and planishing, the profile checked within ±2 mm across both diameters, and the top’s thickness was unchanged at 1.34 mm. The correction cost about 40 minutes.

The workshop added two things to its sequence. The template is now checked after sinking, before any raising, when an over-sunk centre is easiest to correct. And the sinking instruction now names a depth, about 40 mm, measured with a straightedge, rather than leaving the worker to judge depth by eye.

Checks between stages

The over-sunk centre was found because the template was used before the dome was finished. The workshop’s stage checks are short enough to do every time, and each one answers a question that is cheaper to answer early than late.

Figure 3Stage checks for each HF-01 dome

• Avoid: After preparingblank thickness by micrometer, cut diameter, rolling direction marked
• Avoid: After sinkingdepth about 40 mm by straightedge; template laid across to look for a peak
• Avoid: After each raising courseedge diameter against the planned value; no folded wrinkles
• Avoid: After course 4template within ±2 mm across two diameters; edge about 327 mm
• Avoid: After planishingno facets under raking light; template rechecked
• Ask: After trimmingthickness at the top and four mid points; rim continuous; record filed

The checks also produce the records that make the next job’s plan better. After a few domes, the workshop knows how much its sinking draws in the edge, how much each course takes off, and how much the edge wanders, for its own workers and tools. Those numbers, not a textbook’s, set the next blank and the next estimate.

Two domes, one plan

HF-01 needs two domes that match each other as well as the template. Working them in parallel, stage by stage, helps: both are sunk, then both annealed, then both raised, and so on, so that the worker’s hand is consistent and any problem found on the first is avoided on the second. The 1918 text notes that light covers can even be raised two at a time, nested together. For 1.6 mm sheet that is too heavy to be practical, but the principle, keep the two jobs at the same stage, holds.

Exercise: a larger cut blank

A worker suggests cutting the HF-01 blank at 360 mm instead of 352 mm, to be sure of the trim.

1. Taking the edge’s final diameter as 327 mm, how much must its circumference shrink from a 360 mm blank?
2. At 4 to 5 mm of diameter per course, roughly how many raising courses would that need after sinking takes the edge to about 352 mm, and what does the larger blank cost?

Worked answer.

1. From 360 to 327 mm is a reduction of 33 mm in diameter, a circumferential shrink of 33 / 360 = 9.2%, against 7.1% from the 352 mm blank.
2. If sinking draws the edge in by about the same 8 mm, to 352 mm, the raising must take 25 mm off the diameter, roughly five or six courses at 4 to 5 mm each against four, adding perhaps 0.5 to 0.8 hours and possibly a third anneal. It also adds about 0.02 kg of aluminium per dome, which is trivial. The larger blank buys trim margin at the cost of raising time and wrinkle risk. Since the workshop’s records show the edge wandering about ±4 mm, the 6 mm allowance already covers it; the better response to a worry about trim is to check the edge diameter against the bar chart after each course, as the plan already requires.