Horizontal toggle clamp

A horizontal hold-down toggle clamp of the M6, 900 N class, modelled the way it is made: a formed, zinc-plated base in two mirrored L halves held by the pivot rivets, a U-bar folded from one blank, a link of two stamped plates, a handle of two stamped halves joggled together under a red PVC grip, an M6 spindle with a black rubber cap and two lock nuts, and four rivet pivots. It is a real four-bar: t drives the link from open, through the toggle line, to a stop just past it, and back; the over-centre position is proven by the knee's joint angle, and the mechanical advantage is a live output of the solved model, checked against a closed form.

What it shows

• Sheet metal. feature.sheetMetal makes each base half (a flange bent up into an ear, outline in the flat), the U-bar (a narrow web bent into two legs) and each handle half (a joggle, jogs, so the halves meet under the grip); file.export with sheetMetal writes each flat pattern as DXF. The link's plates are flat: feature.sheetMetal with no flanges, one plate of stock thickness mirrored to the other. One source, two instances, for the mirrored base halves (params, hand, and feature.mirror with when).
• Own frames. Every part is drawn with its own pivot at the origin; the assembly passes lengths and angles by params, so changing a link length or the over-centre angle rebuilds the whole mechanism.
• One closed loop. Four mate.add revolutes make the four-bar; assembly.solve reports 0 degrees of freedom; the only drive is joint.set on Link_pivot, a formula of t, with limits on that mate.
• One pivot part, two configurations. rivet is placed four times in two spans.
• Outputs and assertions from the solved pose. variable.measure of position, jointAngle and boxMax/boxMin, variable.output for the mechanical advantage, and an independent closed form of the same ratio from the drive alone.

The real thing

Reference. Sizes follow the DESTACO 217-U (flanged base, U-bar), catalogue sheet 217 Series, pages MC-HHD 9 and 10 (shop.torvigo.es/area_descargas/DESTACO/2741.pdf), read directly. The class was checked against the 215-U (890 N, M6 or 1/4; 215 Series sheet, static.raptorsupplies.com/catalog/product/datasheet/destaco-215-uss.pdf) and the 225-U (2220 N, M8 or 5/16; 225 Series spec sheet, shop.torvigo.es/area_descargas/DESTACO/2742.pdf). All three are read, not recalled.

The sheet does not list the pivot centres; they were read off its side view (A 13.7/32.8, B 20.9/44.2, C 25.6/25.8, D 31.3/11.6, ±0.5 mm, in x behind the base's front and z above the bench). The model's pivots are A 13.5/30.6, B 17.43/43.93, C 24.70/24.13, D 30.92/9.60, found by a search (an unpublished check: its script is not in this repository) that keeps A's x at the catalogue's 13.5 mm and the pivots within 3.5 mm of the drawing, puts the knee exactly 3° past the toggle line, and maximises the handle's travel at a bar opening of 91°. The pivots read off the drawing themselves stop at about bar 78° and handle 52° (dead point at 76.3° of link rotation), so the read-off cannot deliver the catalogue's stroke either. Link lengths A→B 13.9, B→C 21.1, C→D 15.8 mm.

The handle travel is 41°, not the catalogue's 61° (+10°). In this linkage the bar and the handle (A, B and C) come into line at about 89.1° of link rotation from the stop, which is the dead point: the bar opens fast there and the handle gains more, but the model must stop short of it (it stops at 88.2° of link rotation at the default, 88.9° is the hard limit on the mate). No set of pivots near the drawing was found that reaches both 91° and 61°. With 91° kept, the handle's travel is 41° at the default, and the parameter stops at 42°. The README names the gap rather than bending the bar opening to fit.

The class. The M6 sizes are 890 to 900 N (about 92 kgf); the brief's 100 to 200 kgf does not quite meet an M6 spindle. Holding capacity is not proven here. Nothing in this model carries a load; the rating is the manufacturer's.

Over-centre. 3°, by judgement: a trade article (Firgelli Automations, De-sta-co Toggle Clamp Mechanism, worked example) uses three degrees past centre and says "a stop limits further over-center travel". Secondary, and the only source read for it.

Mechanical advantage. The output is a virtual-work ratio of the solved pose (below). At t = 30/72 (knee 7.1° short of the toggle line) it is 7.66:1; at the stop it is 21.03:1 at the 28.6 mm reach; the assertions pin the second number, and an independently computed closed form of the drive, tightly over the whole motion. The catalogue's 7:1 and 3:1 are given without an angle (its MC-TEC-4 page was not found), so they are a comparison here, not a proof. The ratio scales as 1/reach: at the sheet's X2 = 66.8 mm the stop value would be 9.0:1; the slot's closed end limits the reach to 66 mm here.

Sheet metal. Low-carbon steel, 2.5 mm: the sheet shows 2.5 mm at the base's flange, and this model uses that stock for every formed part (the nearest listed stock from a US sheet supplier, 12 Ga., is 2.54 mm: oshcut.com/materialdetails/steel-a1008-cr). Inner bend radius = the thickness (2.5 mm): "keep the bend radius greater than or equal to the material thickness", Sheet Metal DFM Design Guidelines (fiveflute.com/guide/sheet-metal-dfm-design-guidelines-for-formed-and-punched-parts), which also gives the hole rule and the relief rule used below. K-factor from DIN 6935's correction factor k = 0.65 + 0.5 lg(r/s), K = k / 2 = 0.325 at r = s (as quoted on madegooddesigns.com/tools/bend-allowance-calculator, a secondary source; DIN 6935 itself was not available).

Rivets, spindle, nuts, cap and grip are not sheet metal. The rivet shanks are Ø4 in Ø4.3 bores (0.15 mm radial, a stated running fit, not a standard fit class); the rivets have domed heads. ISO 4032 M6 nuts (width across flats 10 mm, height 5.2 mm, read from ISO 4032:1979 table, iTeh preview); the thread pitch is 3DCraft's ISO 261 table for M6, as feature.thread reports it.

Layers along y, from the centre outward: the spindle slot (±3.2), the handle halves inside it behind the spindle (0.6 to 3.1 each side, meeting at y = 0 under the grip), the bar's legs and the link's plates (3.2 to 5.7), the base's ears (5.8 to 8.3). The bar and the link share a layer and clear each other in the plane; 0.1 mm of side clearance between layers. Rivet heads clear the outer plate by 0.2 mm.

Against the process rules, stated:

• Holes ≥ 2.5 t + r from a bend (8.75 mm here): not met in the base. The catalogue's mounting pattern puts the slots against the ear (the slot's inner end is 0.40 mm from the bend's tangent line), and the D pivot bore is 2.45 mm above the bend's tangent line: moving D from the drawing's z 11.6 to 9.6 mm in the search costs those 2 mm. The DXF cuts all of them in the flat, and the distortion at that thin land by the bend is accepted; the bores are meant to be reamed in line across the two halves after forming. The U-bar's and the handle's holes meet the rule (asserted).
• Reliefs: every bend runs the full length of its edge, so no bend ends inside an edge and no relief is needed.
• Inner radius R = t is a stated limit. The sources read say at least the thickness; a smaller radius needs a source, and none was found for this steel. With R = t the bend's tangent line is at y = 10.8 mm, 3.4 mm from the slot's centre, so a screw head or washer over Ø6.8 mm rides up the bend radius, and the slot is held to 5.5 × 6.0 mm (the sheet shows about 8.5) to stay on the flat.
• Edges are as-cut, the sheared edge of the blank (a tumbled part breaks them slightly; a 0.4 mm chamfer on a sheet part's rim would only hold while its outline's corner arcs stay exactly tangent, and these are tangent at the defaults, not at every point of the parts' ranges, so none is drawn on the sheet parts).

Simplified or left out. The rubber cap is not soft (passing the toggle line it pushes 0.045 mm into the workpiece, the preload a real cap gives, and is listed as a contact). The bar's strength, the rivets' forming, the plating thickness and the flanged washers are not modelled. The workpiece is a prop.

Parts and assembly order

To build it: spot weld the handle's halves at the tang and mould or dip the grip over them. Set the link's plates on either side of the handle at the knee and rivet C. Pass the bar's legs between the ears and rivet A through ear, bar, bar and ear; rivet B through the bar's legs and the handle's halves; rivet D through the two base halves and the link. Then fit the spindle: thread the lower nut on the rod, pass the rod up through the slot from below, thread the upper nut on from above, and set the height with the two nuts. The rod has to go in from below since the slot is 6.4 mm wide and the nuts are 10 mm across flats.

The motion

t plays one working cycle, not a repeat: open at t = 0, clamped at t = 0.5, open again at t = 1. The joint Link_pivot is driven by link_drive_angle = open_link_angle (1 + cos(360° t)) / 2, from open_link_angle (at the default handle opening) down to 0, the stop.

The bar's rotation has its extreme at the toggle line, a little past the clamped pose: the rubber cap goes 0.045 mm below the block's top there, the preload, listed as a contact. The stored pose is t = 0.5, clamped. The motion is proven at the samples of motion.check (73), not between them.

Limits. Link_pivot is limited to 0 (the stop) and 88.9°, just short of the point where the bar and the handle line up (about 89.1° from the stop).

Parameters

The workpiece height comes from the model's stack: the bar's underside at 23.4 mm (the sheet's), the rubber cap (10 mm) and the lower nut (5.2 mm).

Checks

At the stored pose (clamped):

• Knee is past the toggle line at the clamped pose: the joint Knee is 3° within 0.05°.
• Rubber cap meets the workpiece at the clamped pose: the measured top of the block against the measured bottom of the cap, within 0.02 mm.
• Link rests on the bar's stop pad: 0.02 to 0.12 mm between Bar and Link (0.05 mm by design).
• The ten … has its running fit clearances: at most 0.18 mm (the 0.15 mm radial clearance; an upper bound, so a loosened bore fails), and at least 0.02 mm (the lower bound is what the measure proves, below).
• Advantage at the stop, independently computed: the output × reach equals 601.57 mm within 0.3 mm (21.03 at 28.6 mm), a value computed outside the model by finite differences of the kinematics (an unpublished check: its script is not in this repository).

Over the motion (motion.check, 73 samples, at the defaults and the corners):

• Spindle to grip ratio matches the closed form of the drive: 1 / the output, against a closed form of the four-bar from the drive alone, within 0.001. This is the solved pose against the formula, independent of the solver's frames.
• Bar and link clear each other over the motion (at least 0.02 mm).
• Bar opening matches the closed form at the open pose and Handle opening at the open pose is the parameter, to 0.02°.
• Knee is short of the toggle line while closing (at t = 30/72) and Bar opens wide enough to load a workpiece (the spindle tip at least 30 mm above the block at t = 0).
• No collisions except the listed contacts.

The check over the public ranges (tools/assemblies.py check) covers these at the corners.

Mutation tests in a scratch copy: moving the bar's handle pivot 1 mm fails the knee (7.51° measured), the cap (−2.15 mm), the stop pad and both advantage checks; a Ø5.5 mm bore fails the fit assertions (0.75 mm measured against the 0.18 mm bound); a block 2 mm taller or 3 mm shorter fails the cap (+2 and −3 mm); the grip force point moved 20 mm fails both advantage checks (762.0 against 601.57 mm; 0.0375 against 0.0475); an over-centre angle of −1° fails the knee and the stop advantage; the stop pad moved 1 mm fails the stop clearance (1.05 mm).

Not proven by the samples: nothing between samples; the toggle crossing is shown by the sign change of the knee's joint angle. The fit lower bound is 0.02 mm because the clearance measure's proven bound is weak on a pin in a bore (Platform notes); the upper bound is what catches a loose fit. The holding capacity is not checked.

The README in the model's repository has the rest (Files, Rebuild, Difficulty, Platform notes, Licence): the whole README on GitHub.