3D-Printed Membrane-Reed Crumhorn V1.2

# 3D-Printed Membrane-Reed Crumhorn (in F)

A cylindrical-bore wind instrument driven by a **membrane reed** rather

than a fipple or a cane reed. A thin sheet of latex is stretched over

the mouthpiece and flutters as a self-exciting valve, the same principle

Nicolas Bras uses in his membrane instruments. It gives a buzzy, reedy,

double-reed sort of voice — somewhere between a duduk and a crumhorn.

Seven STLs, eight prints — BranchBend_scaled.stl gets printed **twice**

if you want the curved crumhorn shape. No fasteners beyond zip ties, and

no glue required.

---

## What you get

| File | What it is |

|---|---|

| segment4_v7.stl | Foot segment. Lowest tone hole, plus an 18mm stub for the bell |

| segment3_v7.stl | Middle segment, three tone holes |

| segment2_v7.stl | Upper segment, three tone holes, socket at the top for the mouthpiece |

| BrasMouthpiece_fused_tip15_v6.stl | The mouthpiece — inner pipe and plenum housing in one piece |

| MembraneRingSplit_id30_0.stl | Split clamping ring, closed with a zip tie, holds the membrane |

| WindcapNozzle_bendsocket.stl | Blowing nozzle. Optional, but it steadies the note noticeably |

| BranchBend_scaled.stl | 90° bend. **PRINT THIS TWICE** — one for the blow branch, one for the bell |

| crumhorn_G3_key.scad | Optional key for the lowest hole, plus its anchor ring. Source only — the touchpiece position wants setting to your own hand |

| crumhorn_tonehole_insert.scad | Glued into the lowest hole, gives the key's pad a flat seat. Only needed if you fit the key |

Assembled length about 380mm.

**Print BranchBend_scaled.stl twice.** One goes on the mouthpiece's

side branch to bring the blowing angle round to something comfortable —

you want that one regardless. The second slips onto the stub at the foot

and gives the instrument its curved crumhorn silhouette. Without it you

have a straight pipe that plays perfectly well but doesn't look the

part.

---

## Printing

PLA, 0.2mm layers, 3 perimeters. Tested on a Creality K1.

**Print the tube segments diagonally, tone holes facing up.** Printed

upright they're slow and the sockets suffer; printed flat the holes sag.

Diagonal is the compromise that works.

The small zip-tie ears on each segment are useful as support anchors.

If one breaks off in post-processing it doesn't matter much — but see

the alignment note below before removing them all.

The split ring prints standing on its flanged end. The flange is part of

the model, not a brim; trim it off with a knife.

---

## Assembly

**1. Tube.** Push segment 4 into segment 3, then segment 3 into segment

2. The joints are friction fits; they should need firm hand pressure and

no glue. If yours are loose, a wrap of PTFE tape will not work — two

of the tone holes sit inside the joints and vent through windows cut in

the tenon, and tape blocks them. Glue instead, or reprint the tenons

0.1mm larger.

**2. Rotation matters.** Line the tube up so every tone hole faces the

same way. Two of the holes only work when the joint is clocked

correctly — you can see straight through the hole and its matching

window when it's right, so hold it to the light and check.

**3. Membrane.** Cut a square of latex about 50mm across. Lay it over

the mouthpiece's front rim, keep it flat and free of wrinkles, draw it

snug but not drum-tight, and hold it with an elastic band while you

work. Slide the split ring on over it and cinch a zip tie around the

ring's groove until it grips. Trim the excess latex.

**4. Mouthpiece.** Push the mouthpiece tenon into segment 2's top

socket. Fit a bend onto the mouthpiece's side branch to bring the

blowing angle round to something comfortable, and the nozzle onto the

end of that bend.

**5. Bell.** Push the second bend onto the 18mm stub at the foot. This

is the "crumhorn" part of a crumhorn — see the section below, including

the vent holes needed to keep the pitch where it was.

---

## The membrane — read this part

**Use latex.** Cut it from a party balloon, or a latex/nitrile glove.

Thin plastic film (a sandwich bag) tunes better — it is a stiffer,

lighter termination and gives a more even scale — but in practice it is

maddening. It tears, it pinholes on the printed edges, it needs remounting

constantly, and getting it to speak at all is hit and miss. Latex is

tougher, survives being clamped and re-clamped, and speaks reliably.

The trade is that latex is heavier and softer, which flattens the

instrument and squeezes the intervals together. That is a real cost and

it is described honestly below.

**Membrane tension dominates everything.** This is the single most

important thing to know about the instrument. In testing, simply

knocking a mounted membrane shifted the pitch more than three separate

redesigns of the mouthpiece did. Two mountings of the same material on

the same mouthpiece will not give the same pitch. If you remount it, it

will sound different.

Blow steadily and fairly gently. It needs less pressure than you expect.

---

## What it actually sounds like

Measured with a tuner, latex membrane, all seven parts assembled. Hole 1

is nearest the mouthpiece, hole 7 is all holes closed:

| Fingering | Measured |

|---|---|

| all closed | 197 Hz |

| 6 | 202 Hz |

| 5 | 226 Hz |

| 4 | 242 Hz |

| 3 | 268 Hz |

| 2 | 305 Hz |

| 1 | 343 Hz |

That is roughly G3 up to F4 — about seven and a half semitones spread

over seven fingerings, where an even diatesserate scale would want

eleven. **The intervals are compressed and approximate.** The membrane

is a soft termination and it takes some of the pitch control away from

the tube.

In plain terms: it plays a rising scale of seven notes with a strong,

reedy voice. It is not tempered and it will not play in tune with other

instruments. It is a solo instrument, and a good one for drones, modal

noodling and folk tunes that don't stray far.

Your numbers will differ from these. The membrane sees to that.

---

## Fingering and hand span

Seven holes are drilled but the span is wide. Most people will want to

**plug the topmost hole** (the one nearest the mouthpiece) with tape or

a plug and play it as a six-hole instrument. That keeps the scale

contiguous and brings the reach down to something like a low D whistle.

Try it both ways.

The other awkward reach is the lowest hole. There is an optional key for

that one — see below.

---

## Tuning it

Holes can only be made bigger, and a bigger hole sounds sharper. So:

- **Note too flat** → enlarge the hole, or file its mouthpiece-side edge

- **Note too sharp** → partially fill the hole's mouthpiece-side edge

Work one hole at a time with a round file and a tuner, and expect to

redo it if you change membranes.

---

## The bell

This is why you print BranchBend_scaled.stl twice.

segment4_v7.stl has an 18mm stub protruding from its foot, and the

second bend slips straight onto it — same joint as the one on the blow

branch. That upturned curve is the whole reason the instrument is called

a crumhorn (from the German krumm, crooked), so it is worth the extra

half hour of printing even though it is acoustically decorative.

The bend adds about 78mm of tube, which drops the lowest note. To get it

back, drill a pair of opposed vent holes through segment 4's wall about

6mm up from the foot — there are shallow dimples moulded in to centre

the drill. **Start at 6mm and open up in steps**, checking the pitch as

you go: enlarging them sharpens the note, so you approach the target from

below. Two opposed holes work far better than one.

With the vents open the bell becomes acoustically decorative — the tube

terminates at the vents and the curve just hangs below it.

---

## The bottom-hole key — optional

The seven holes are a wide span, and the lowest one is the awkward

reach. There is a key for it: a pin-hinged lever clamped to the tube

that carries a pad over the bottom hole, held open by a rubber band and

closed by the little finger. Print crumhorn_G3_key.scad for the key

itself and the anchor ring, and crumhorn_tonehole_insert.scad for the

seat that goes in the hole — the key needs it to seal properly, so

treat them as one job rather than two.

It rests **open** and closes when pressed, which is what a finger does —

that hole is open for every note except the fundamental.

**The band anchors on the bell.** The anchor ring clamps around the

socket collar of the bend at the foot, about 10mm below the instrument.

That is far enough below the pivot for the band to pull the key open

rather than shut; anywhere higher up the body and the geometry flips

sense and the band holds the pad down. There is no room on the body

itself — the stub, the vent holes and the key's own clamp ring take up

everything below the bottom hole.

Two consequences. The band's pull seats the bell onto its stub rather

than working it loose, which is a small bonus. And the bell stops being

optional: take it off and the key goes limp. Refitting it changes the

tension and the pull angle, so expect to re-tune the band's position.

Tighten the anchor ring hand-snug and no more. It is squeezing a 2mm

wall around a socket that already has a tenon inside it.

### Sealing the hole — fit the insert

The bottom hole is about 11mm across on a 19mm tube, so its rim spans a

quarter of the circumference. That rim is not a circle in space: it is a

**saddle**, falling about 3.5mm below the crown at its sides. A pad

pressed onto a saddle seals differently depending on which edge lands

first and how hard you lean on it, and that is the whole difficulty.

The answer is a small glued-in insert that gives the hole a flat rigid

rim. Print crumhorn_tonehole_insert.scad. A short tapered spigot drops

into the hole, a saddled flange sits down on the tube around it, and the

flat top of that flange is what the pad meets. One soft pad against a

hard flat seat — which is what every real woodwind does, and what makes

the seal repeatable instead of a matter of technique.

**Print it flat top down.** The saddle is then a concave recess facing

upward, so there is no overhang and no support.

**Dry-fit before gluing.** It should drop in and sit flat with no

rocking, unforced. If it binds, sand the taper back — a spigot that jams

while the flange stands proud will set that way, and a permanently

tilted seat is worse than no insert at all.

**Then clamp it hard while the glue sets**, with the key off. A zip tie

round the tube works. Sight along the flange from two directions before

the adhesive grabs: 0.3mm proud on one side is permanent.

The spigot stands about 1.5mm into the bore. That is intended, and its

inner end is chamfered so it is not a sharp step in the air column.

**It costs about a third of a semitone**, flattening the lowest note.

The hole gets smaller and gains a chimney, and both flatten. If that

bothers you, ream the insert's bore — never the tube. The insert is

the sacrificial part and it is the one that should take any

irreversible cutting.

### Two things that don't work, so you can skip them

**A boss glued over the hole.** A ring sitting on the outside, giving

the hole a flat rim without entering it. It needs two things exactly

right at once: the curved underside must match the tube, and it must

be glued square over the hole by eye while the adhesive grabs. Nothing

locates it. Either one slightly out and there is no way back. The insert

works precisely because it cannot be glued crooked — the thing it aligns

to is the thing it sits in.

**Foam on the tube and rubber on the pad.** Soft against soft. This does

work, and it is a reasonable fallback if you would rather not glue

anything permanent into the hole: thin craft foam glued flat around the

hole, cut through afterwards using the hole as the guide, with an

oversized rubber pad on the key. It seals well enough to play. But it

never seals tightly, the closed position is mushy because both faces

give, and the foam leaves its own chimney on the hole anyway. The insert

is better in every respect except reversibility.

### If it still leaks

Press the pad closed with your **thumb, straight down over it**, instead

of with the little finger on the touchpiece. If it seals that way and

not the other, nothing is wrong with the seat — the arm is twisting

under an off-axis load, and the touchpiece offset wants adjusting to

your hand. That offset is a parameter at the top of the key file, and it

is the one dimension that has to be set by holding the instrument rather

than by measuring anything.

If it seals only when pressed hard, the band is too tight. Slacken it or

move the anchor ring round.

---

## Credit

The mouthpiece is a **remix of Nicolas Bras's membrane-reed design** —

the T-junction housing, annular plenum and front-rim membrane are his

idea, not mine. The tone hole positions were sanity-checked against

Fabien T.'s published membrane clarinet, whose lowest hole sits at 0.877

of the tube length from the mouthpiece against this design's 0.873.

Everything else — the segmented parametric body, the joint scheme, the

split clamping ring, the vented bell — was worked out here.

Designed in OpenSCAD. Licensed CC BY-NC-SA.