Support Strategy for Thin Ring Shanks and Fine Prongs
Almost every failed thin shank we see has too many supports, not too few. The operator saw a bowed ring or a snapped prong, concluded the part needed more holding, and re-sliced denser. The next print bowed in the same place, with more scars to polish. The reasoning feels correct because it treats a DLP build as a clamping problem. It is not: it is a compliance problem, and supports are the most compliant elements in the assembly.
This note works through what a support actually carries, why adding supports to slender geometry can increase the load instead of resisting it, and which numbers decide the outcome. The reference geometry is a 1.2 mm to 1.8 mm ring shank and a 0.8 mm to 1.0 mm prong tip in a wax-filled casting resin; the ratios carry over to other sections.
What a support actually does
A support has three conflicting jobs: resist the peel force transmitted through the part as the layer separates from the film, locate the part precisely enough that successive layers register, and come off without leaving a defect that cannot be polished out of a 1.3 mm shank.
The dominant load is the peel event: as the platform lifts, the cured layer clings to the film while the part and its supports must transmit that force. Peak peel force on a small footprint in an oxygen-permeable film system typically lands in the 0.1 N to 0.5 N range. That is small in absolute terms and enormous relative to a slender part's stiffness.
Peel force scales with the projected area of everything curing in that layer, not with the mass of the part. A thin shank has a small footprint, so it generates little peel force on its own — which is exactly the fact that gets ignored when supports are added.
Supports add peel area faster than they add holding
Every support column that reaches the current layer contributes its own cross-section to the projected area being cured, and therefore to the peel force for that layer. On a thin part, the supports can easily outweigh the part itself. A worked example with a 1.2 mm by 1.5 mm shank section and 0.8 mm diameter support columns:
| Element | Cross-section at the layer plane | Count | Total projected area |
|---|---|---|---|
| Shank (part) | 1.8 mm2 | 1 | 1.8 mm2 |
| Support column, 0.8 mm dia | 0.50 mm2 | 4 | 2.0 mm2 |
| Support column, 0.8 mm dia | 0.50 mm2 | 8 | 4.0 mm2 |
Four supports already carry more peel area than the part they hold; eight triple the peel load the assembly sees, and all of it must be transmitted back through the same thin shank. The supports are not merely useless at that point — they are the load. The failure that follows gets blamed on insufficient support.
Dense arrays add a second-order effect: where two column bases sit within a fraction of a millimetre, scattered light cures the resin between them and the columns merge into a plate. The peel force becomes that of the plate, arriving as a sharp step — which is why dense arrays leave a horizontal band exactly where the array ends.
Support columns behave as springs, not clamps
The lateral stiffness of a support column loaded at its tip is a cantilever problem, and it is brutally sensitive to diameter:
k = 3E I / L3, with I = πd4 / 64
So stiffness scales with the fourth power of diameter and the inverse cube of length. At roughly 1200 MPa, columns 10 mm tall:
| Column diameter | Second moment of area | Lateral stiffness at 10 mm | Deflection under 0.2 N |
|---|---|---|---|
| 0.6 mm | 0.0064 mm4 | 0.023 N/mm | 8.7 mm |
| 0.8 mm | 0.0201 mm4 | 0.072 N/mm | 2.8 mm |
| 1.0 mm | 0.0491 mm4 | 0.177 N/mm | 1.1 mm |
| 1.2 mm | 0.1018 mm4 | 0.366 N/mm | 0.55 mm |
Read the 0.6 mm row again. Under a load well inside the normal peel range, a 10 mm column deflects further than the shank is wide. A tenth column does not help: each is an independent spring of the same low stiffness, and each adds peel area. The fix is diameter and length, not count.
This is the usual cause of prong misalignment. A tip supported 15 mm up through a 0.7 mm column drifts laterally by a fraction of a millimetre over successive layers, and the drift does not self-correct: the tip ends up short, thin on one side, or hooked. Thickening that column to 1.1 mm and adding one diagonal brace beats tripling the number of 0.7 mm columns.
Span between supports matters more than support count
Between two support contacts, the part is a beam. Deflection under a distributed load goes with the fourth power of span, so the gap between contact points dominates everything else:
δ = 5 w L4 / (384 E I)
Widening a gap from 3 mm to 5 mm raises mid-span deflection by a factor of about 7.7. This explains most bowed shanks: a designer who adds supports uniformly but leaves a 6 mm gap across the crown, because a support there would be awkward to remove, has built a 1.3 mm section spanning 6 mm. Nothing either side of the gap helps.
Support layout is therefore a span-control exercise. On a 1.2 mm shank in a wax-filled resin, keep unsupported runs under 3 mm; under 2.5 mm on prongs. Do it with a few stiff, short columns, not a carpet of thin ones.
Placement rules that hold up on the bench
- Support the moment, not the surface. For a ring, peak bending moment sits at the crown and the two shoulders where the shank meets the head. Those three zones get the stiff supports; the straight runs between them need only enough contact to keep span short.
- Cap contact diameter at roughly 40 % of local section width. On a 1.2 mm shank that means 0.45 mm to 0.5 mm, not 0.3 mm. A small tip barely reduces the scar but concentrates stress into a crack initiation site after casting. Wide, shallow contacts on a hidden face polish out far more easily.
- Prefer leaning columns to vertical ones. A column leaning 30 degrees from vertical has a shorter effective bending length for a lateral load and behaves several times stiffer in the direction that matters. On prongs, a pair of opposed diagonals is worth more than five verticals.
- Brace long runs. Any column taller than about 12 mm gets a mid-height tie to a neighbour or the raft; otherwise it is a long cantilever and the calculated deflection applies directly.
- Keep supports off sealing faces and off the sprue junction. A support scar on the surface that meets the investment is a leak path. A scar at the sprue root is a turbulence source and a hot spot.
- Never support the inside of a closed ring band. The support cannot be reached and will be broken off through the part. If the geometry seems to require it, the orientation is wrong.
Orientation changes the load by an order of magnitude
A ring printed with its axis vertical, flat in the build plane, cures a full annular band in every layer. For an 18 mm diameter ring with a 1.2 mm band width, that is roughly 68 mm2 of fresh area per layer, arriving all at once, every layer. The same ring tilted so that the band rises through the layer plane at a steep angle cures a short arc of a few square millimetres per layer. Peak peel force drops by close to ten times, with no change to the support layout at all.
Choose orientation before drawing supports. Tilt the ring axis 15 to 30 degrees off the build plane, crown high, shank rising as a ramp. Layer count and underside finishing cost more; a scrapped bowed ring costs far more.
Fine prongs get a separate rule: never let a prong tip be the highest point of a tall column. Angle the ring so the prong tip is reached from the side, supported along its length rather than from directly beneath, so the last few layers add area gradually instead of terminating a slender tower.
Reading the failure
| Symptom | Mechanism | First intervention |
|---|---|---|
| Shank bowed in the same plane every print, worst at mid-span | Excessive unsupported span; beam deflection under peel load | Shorten span to under 3 mm at the crown, not more supports elsewhere |
| Layer lines visibly offset in one direction, sawtooth pattern | Lateral drift of a compliant tall column; each layer registers against a displaced datum | Increase column diameter and add a mid-height brace |
| Horizontal band or step at one height | Merged support bases forming a plate, releasing peel load as a single step | Space column bases at least one diameter apart; thin out the array |
| Prong tip short or hooked | Tip supported from directly below by a long thin column that drifts and deflects | Reorient so the prong is approached from the side; support along its length |
| Contact craters too deep to polish | Tip too small, penetration too deep, or contact placed on a visible face | Increase tip diameter, reduce penetration, relocate to hidden face |
| Clean break at a support, part intact around it | Contact area too small for the local load; the joint failed before the part | Increase tip diameter; check that this column is not carrying a span alone |
| Ring oval after casting, not before | Not a support problem: shrinkage or investment restraint | Measure the printed pattern first before touching the support layout |
The last row matters most. Many "support problems" are dimensional problems visible only after casting. Measure the printed pattern on a flat plate with a profile projector before blaming supports. If the pattern is round and the casting is oval, support layout is not your variable.
What to change first
- Count the supports that reach the thinnest section and compare their total cross-section to the part's. If supports exceed the part, remove supports.
- Measure the longest unsupported run on the shank. Cap it at 3 mm.
- Check the diameter and free length of every column taller than 12 mm. Thicken before densifying.
- Re-examine orientation. If the ring axis is within 15 degrees of vertical, tilt it and re-slice before changing anything else.
- Print a span ladder: the same 1.2 mm bar supported at 2, 3, 4, 5 and 6 mm intervals, all on one plate. The height at which each span starts to bow gives the working limit for that resin and that section, measured rather than assumed.
Thin jewelry geometry punishes the intuition that stiffness comes from quantity. On a 1.2 mm shank, stiffness comes from diameter, from short spans, and from orientation. Support count mostly adds load.
Ready to discuss your project?
Tell us what you need — our engineering team replies within 24 hours.