Mana Press — screw press for Magic cards: flatten a bowed deck, brick a freshly sleeved one

A screw press for Magic cards, printed whole — there is not a single screw, magnet or insert in it, because the two posts are the screws.

It does two jobs. The everyday one is flattening: a stack of bowed cards goes in, comes out flat, and a freshly sleeved deck that no longer fits its box goes in and comes out as a brick that drops straight back in. The second is experimental and optional — an arc pair that tries to straighten a curled foil. More on that below, including why it is flagged experimental.

How it works

Two posts rise out of a tray, threaded over their top 62 mm. A yoke drops over both and slides freely down; two big thumb nuts run down the threads and push the yoke down. Whatever is in the tray is squeezed between the yoke and the tray floor.

Nothing about the yoke is adjustable — it simply rests on whatever height the stack happens to be. That is deliberate. Gross height is set by stacking 12 mm pressure plates until the top of the stack lands in the yoke's working band, and the nuts take up the last few millimetres. The plates are the height adjustment and the pressure face, which is why there are five of them and why they carry the mana symbols: you were going to print spacers anyway, so they may as well be the WUBRG rainbow.

Four details are doing real work:

The tray is deep — 20 mm. The plate is a piston in that pocket with 0.6 mm clearance per side. A tall stack under load wants to skew sideways and the wall will not let it, so a plate cannot tip and dig a corner into your cards.

The thread is coarse — 10 mm per turn. One turn travels a whole centimetre, so the press opens and closes fast instead of making you wind for a minute. The 8.2° lead angle that comes with a pitch that coarse is still well under PLA's ~20° friction angle, so it self-locks: let go and it stays put.

The yoke is not threaded. Its bores are Ø23.4 mm over a Ø22 mm crest, so it just slides. Nothing has to stay in phase and you never turn two screws at once — each nut pushes down independently on the yoke's top face, exactly like the two screws of a bookbinder's press.

The yoke has a rib down its middle because that is where it bends. The nuts load it at the ends, the stack pushes back in the centre, so the bending moment peaks mid-span and is zero under each nut. Stiffness goes as depth³, so the rib takes that span from 18 mm to 24 mm — 2.4× stiffer where it matters for about 15 % more plastic — while the ends stay thin so the Ø48 nuts land on flat pads.

What it does not do

It is a press, not a vice. Finger tight is the whole specification. Cards are paper: the flattening comes from holding them flat for hours, not from force. Over-torquing marks the top card and bows the yoke, and neither buys speed.

What to print

The minimum useful press is 5 prints: base, yoke, both nuts, two plates. Everything else is optional — more plates for taller stacks and for the rainbow, the arcs only if you want to try the foil experiment.

A working press is 623 g solid, about 340–350 g sliced. All eleven parts together are 1036 g solid, roughly 560–580 g sliced. The base is a third of it on its own and is the long print.

No hardware of any kind. No screws, magnets, bearings, elastic or glue.

Printing

Every STL is exported in its print orientation — place them as-is and do not rotate. No supports anywhere, no brim needed.

• Posts, yoke and nuts are structural: 4 walls, 25 % infill or more.

• Plates and arcs are stiff slabs: 3 walls, 15 % is plenty.

• PETG is worth it for the yoke and nuts if you leave cards under load for days — those two carry sustained stress and PLA creeps at room temperature under a constant load. PLA works fine for overnight presses.

• arch_upper is exported already flipped, concave side up. In its use orientation that face points down and would be a 6 mm unsupported dome. Flipped, it is a bowl with zero overhang. Rotating it "the right way up" costs you the print.

• Do not scale the model. The thread's 1.1 mm tooth depth is sized so consecutive turns cannot merge into a solid collar on a 0.4 mm nozzle, and scaling breaks that margin in both directions.

Multi-colour (optional). Each plate ships a mana_press_plate_<x>_2c.stl holding two named solids at locked coordinates. Import it, right-click → Split to Parts, assign a filament to each. Do not import body and glyph separately and join with "Add Part" — that re-arranges coordinates and the glyph lands beside the plate instead of in its recess. The symbol is a flush inlay in the plate's top face, three layers of second colour, so it is about 1 g of purge per plate. Print the plates together on one plate: their colour bands all sit at the same height, so five cost about what one does.

Height: how many plates

The yoke's underside has to land between 60 mm and 76 mm above the tray floor. Below 60 it fouls the shoulder where the smooth column ends; above 76 the nuts run out of thread. Each plate is 12 mm:

Assembly — flatten mode

Step 1 — Base. Set it down. It prints as one piece, posts and all.

Step 2 — Load the tray. Cards flat, long edge front-to-back. A sleeved deck goes in as a block, loose cards as a squared-up stack. The 20 mm wall keeps a tall stack from skewing as it is squeezed.

Step 3 — First pressure plate. Symbol side up. The flat underside is the working face; the recessed top is where the next plate lands.

Step 4 — Second plate. On the rim of the first. The glyph sits 0.6 mm below the rim, so plates seat face-to-rim and never rock on the artwork.

Step 5 — Third plate, and any more you need. Keep adding until the top of the stack is roughly 60–76 mm up. It does not have to be exact; the nuts take up the slack.

Step 6 — Yoke. Lower it over both posts, ribbed side up. It slides down until it meets the stack or stops on the shoulder — both are fine.

Step 7 — Both nuts. Thread one onto each post and run them down, alternating a quarter to a half turn at a time. Driving one ahead of the other cocks the yoke and it binds. The bore is 18.0 mm long with 1.4 mm of diametral slack, so the yoke can cock 0.078 rad before it jams two-point on the posts — across the 94.8 mm post spacing that is 7.4 mm of height difference, and the pitch is 10 mm per turn, so one full turn on one nut is already past the jam point. Half a turn is inside it, a quarter turn is comfortable. If it locks, back the leading nut off; do not force it.

Finger tight. Leave it for hours — overnight is better.

Foil mode is EXPERIMENTAL

arch_lower and arch_upper are a separate, optional, unproven job. Everything else here is a straightforward screw press and behaves like one. The arcs are a theory about foil cards, printed.

The theory. A foil bows because it is a bimetallic strip — the foil layer and the cardstock behind it take up humidity at different rates, so the card acquires a permanent set with its face convex. That is why pressing a foil flat between the plates does nothing: flat clamping only loads a spring, and it springs back the moment you release it. To move a permanent set you have to yield the material past it, which means bending the card beyond straight the other way and letting it relax to flat. Same as straightening bent wire: you do not press it flat, you over-bend it.

What is unverified. The over-bend depth — a 6.0 mm crown over the card's 92.4 mm long axis, radius about 181 mm — is reasoned from typical curls of 3–5 mm, not measured against a real card. Plausible ways it goes wrong: the card comes out curled the other way (lower the crown and repeat — the set moves gradually, so this is recoverable); a stubborn card springs most of the way back (raise it, or run a second pass); too much clamp force concentrates load along the crown line and marks the card. Do not experiment on a card you care about. Try a bulk foil first.

The steps replace 2–4 above, then rejoin at the first plate with three plates on top. Shown as a front section — the crown is only 6 mm and reads as a flat slab from outside.

F2 — arch_lower, flat side down, crown up. It drops into the tray pocket like a plate does.

F3 — the card, FACE DOWN. Face down is the whole trick: the card's face has to be the concave side while it is clamped. Face up and you press a worse curl into it. The instruction is engraved on arch_upper for that reason. The card goes in flat, so at this point it only touches the crown along its centre line — closing the press is what bends it.

F4 — arch_upper on top. The two arcs are complementary to 0.01 mm, so they close on the card evenly. Then three plates, the yoke and the nuts as above.

Making it yours

The .scad is included and every dimension is a parameter at the top of it. With OpenSCAD, -D overrides any of them from the command line — no editing required:

openscad -o base.stl  -D 'part="base"'                   mana_press_standalone.scad openscad -o plate.stl -D 'part="plate"' -D 'sym="r"'     mana_press_standalone.scad openscad -o arch.stl  -D 'part="arch_lower"' -D 'arc_crown=4.5' mana_press_standalone.scad

Two numbers to leave alone unless you read the THREAD MATH comment first: th_pitch and th_root. Tooth depth must stay under (pitch/2 − 2·slop − 0.3) / (2·√3), or consecutive turns merge into a solid collar and no nut will ever thread on.