Hurdy-Gurdy (4-string) From Plywood and 3D Printed Parts

The hurdy-gurdy is a unique and historical musical instrument. It uses a hand-crank to turn a wooden wheel, which continuously bows one or more strings, while a series of keys allows those strings to be fretted to play melodies. Additional strings produce a backing drone and a rhythmic buzz if required. It gives music a really medieval quality. Unfortunately even a modestly-specified instrument seems to be eye-wateringly expensive and typically on long lead-time; they don't make mass-market 'student' hurdy-gurdies like they do with ukuleles etc. This isn't ideal if you just want to mess around with one to decide if you want to learn to play it properly. As I sit squarely in this camp the only option for me was to make one.

There are several DIY gurdy designs floating around on the web, of widely varying levels of sophistication. It's fair to say that the most highly regarded one (by a wide margin) is the Nerdy-Gurdy, which is available in kit form from the Netherlands. These kits are produced in limited quantities and for me represented a fairly significant outlay, given I only wanted to 'dip my toe in' as it were. Most of the plans for the earlier iterations of this instrument are freely available, but require the would-be builder to have access to a large laser-cutter. I couldn't get access to one locally and Baltic birch plywood was extremely expensive at the time I was looking around (2023), as it was (and probably still is) subject to a trade embargo.

I finally settled on a design authored by MarcoCammozzo and published on Thingiverse. This was for a fairly large four-string instrument (1 drone, 1 trompette and two melody strings), with a two-octave keybox and a scale length of 42 cm. The description was accompanied by a video demonstration of Breton An Dro played on his machine, which made me think that this was a workable design and capable of all the right noises. He shared plans for the 3D-printed parts of his design but there was little information on the construction of the body or final assembly of the instrument, so it was clear that there would be a fair amount of work involved.

After an initial experiment where I printed his keybox parts half a side at a time in PETG, I was satisfied that I could make the design work for me once I'd worked out how to construct the body and put it all together. I chose PETG for the 3D-printed parts, because it's much tougher than PLA and doesn't soften if left in a hot environment. I also wanted to use standard plywood from a DIY store, rather than the coveted Baltic birch which was too expensive. I didn't know at the time whether either material would be tough enough to withstand the string tension (spoiler: it was fine!).

This Instructable describes my build and has all the files and drawings you need to make this hurdy-gurdy, along with the lessons I learned on the way. I have modified or completely redesigned many of the 3D printed parts in order to simplify construction and use. In retrospect it was a tonne of work but on the whole I'm pretty happy with the result. Although its appearance is somewhat 'industrial' - lacking the graceful curves and beauty (and varnish) of a luthier-built machine - it's great fun to mess around with. So now I have something to practise on, to try to get good enough to justify buying an expensive professionally made instrument.

I made this project with a standard desktop 3D printer capable of printing PETG. I also have access to a reasonable range of woodworking tools and basic workshop facilities. Building the machine took a lot of time, patience and research, doing a little bit here and there over a period of about four months. This was in large part because I wasn't following a detailed design and had to research many aspects for myself, redesigning as I went.

One final note of caution - I have found that getting my hands on a hurdy-gurdy (the topic of this Instructable) was only the first step onto what is turning out to be a huge learning curve. There is an enormous amount to learn in order to set up and play these instruments well. Fortunately there is a wealth of information on the internet and many talented and enthusiastic players and makers willing to share their skills and knowledge on YouTube and elsewhere (for example, Noelle Kristen Beaudin, Scott Gayman, Nigel Eaton, Sergio Gonzalez, Neil Brook and many more). I hope this Instructable helps you with that all-important first step!

The following supplies were needed to carry out this build. It looks like a lot (if procured in one go) but I got a long way forward from just a spool of filament and a bit of plywood, plus some consumables that I already had lying around in the workshop.

1. PETG filament - I used black because I had lots of it
2. PLA filament (optional) for wheel cover
3. 10 mm threaded rod (studding)
4. M10 bolt (hex head) between 70 mm and 75 mm in length
5. M10 nuts and nylok nuts
6. M10 washers
7. 2 off bearings 10x26x8 mm
8. Medium nylon strimmer line (~1.5 mm nylon)
9. Tenor ukulele strings (pack of four Low-G C E A - I used Aquila Nylgut)
10. Wooden edging strip (20mm x 1 m)
11. 4 mm plywood (1200 x 800 mm)
12. 12 mm plywood
13. 10 mm square section wooden beading (2 m)
14. Miscellaneous wood off-cuts
15. 4 off machine heads for tensioning strings
16. Stainless steel M3 cap-head screws (24 off 16 mm, 28 off 30 mm) and nuts (4 off)
17. Stainless steel M4 cap-head screws (3 off 16 mm, 1 off 30 mm) and nuts (4 off)
18. Self-adhesive velvet strip (30 cm)
19. 30 cm thin cord
20. 2 off guitar strap buttons
21. Insulated single-core hookup wire
22. 25 mm length of brass rod or bamboo skewer (approximately 2.5 mm diameter)
23. Polyurethane glue
24. PVA glue
25. 2-part epoxy resin glue
26. Small countersunk screws of various sizes
27. Neodymium magnets (6 mm x 1.5 mm diameter)
28. Brass pillars with matching countersunk screws (~4 mm diameter)

CAUTION - this work involved lots of sharp tools, nasty adhesives, heavy objects, hot 3D printing and twangy strings, so if you attempt to duplicate it you do so at your own risk, of course.

A large number of 3D-printed parts are needed to build this hurdy-gurdy. A few are exactly as in the original MarcoCammozzo deisgn (starting with '3d-gurdy...'). I modified some of his parts for this build and designed quite a few more from scratch as appropriate.

An initial difficulty was the original keybox sides - these were modelled as single parts which wouldn't fit onto the bed of a standard amateur 3D printer. So I had to split them length-ways then find a way to bond the two halves together securely, otherwise the whole endeavour was dead in the water. I found that PU adhesive gave a good strong bond between flat PETG surfaces, so I used a simple jig to join the parts together. This actually worked really well. Buoyed by this success, I proceeded with the rest of the build.

Many of the 3D-printed parts evolved as the build progressed; I would probably have designed some of them completely differently if working from a blank sheet. Having said that, I took photos of the build from the outset, so keep in mind that in some of the photos the parts might not be to the final design.

All the STLs needed to print the plastic parts of the machine are attached. I used a gyroid infill, with density varying between about 25% and 50%, with the denser fills reserved for load-bearing parts such as the adjustable bridge and the string anchor. All the parts are one-offs, except the tangents (28 off long and 20 off short - two per key), the wheel cover clips (2 off) and the 3d-gurdy-keybox-box-support (2 off). I printed the natural keys in black and the sharps/flats in white, which seems traditional for these instruments. Note that the wheel needs to be printed with the z-shift point randomised around the circumference (this is a slicer setting). I didn't do this the first time and ended up with a bump on the wheel which I couldn't remove even after quite aggressive sanding.

The next step was to start making the body by cutting two 4 mm plywood panels according to the drawing (in which all the dimensions are symmetrical about the horizontal centreline). The plywood needed to be marked out carefully, giving consideration to a) which way the grain would run on the finished instrument; and b) how to make best use of the available plywood, given that the side panels would need to be made later (and they would also have their own grain direction). I chose to have the grain running length-ways on the upper and lower panels, but width-ways on the side panels. I used a tenon saw to cut the plywood, which gives nice straight cuts with minimal splintering.

The dotted rectangle shows the slot through which the wheel goes. This slot was cut in only one panel, which was destined to become the upper panel of the body. To cut the slot I drilled a 6 mm hole in one corner of the marked out rectangle, then used a coping saw to cut out the shape. Plywood is a tricky material to saw cleanly with a coping saw. The cut edges of the finished panels (interior and exterior) were therefore lightly sanded as required to clean them up.

The headstock is a critical supporting component for the rest of the body so I needed to make it first. I used 12 mm plywood. The first step was to hack out the basic shape twice (see the drawing), then finish the two parts with rasps, plane and sandpaper as appropriate so they matched well. I also removed the sharp corners close to the machine heads. These two halves needed to be held in position with a third piece of 12 mm plywood of thickness 65 mm (shown in hidden detail on the drawing). This piece was held in place by countersunk screws to allow disassembly whilst I developed the design. A bench vice is very useful for ensuring that the parts are square and well-aligned prior to drilling pilot holes for the screws. Once in position, the protruding edge of this third piece was removed using a hand plane, so it sat flush with the two halves of the headstock.

At this point the headstock was disassembled and the holes for the machine heads were drilled. The machine heads could then be fitted - note the different orientation of the front and rear pairs. Each head has a small screw to prevent it spinning round in the hole when under tension. These will not be easily accessible on two of the heads once the headstock is fully assembled, so I had to make sure I was happy with the orientation.

Finally the fourth piece of 12 mm plywood was shaped to fit on the underside of the headstock. This needed chamfering on both ends prior to being glued in place with PVA. I chose glue as I didn't want large screwheads to be visible on the final instrument.

My design for the headstock was based more on speed and expediency than anything else. I wanted to make something quickly that allowed the rest of the machine to function. In retrospect I would have done several things differently. In particular:

1. Increase the height of the uppermost set of machine heads (for the melody strings) by about 15 mm - currently the strings make too great an angle around the nut, putting excessive force on it and making the wound low G string tricky to tune smoothly
2. Reduce the height of the lowermost set of machine heads (for the drone and trompette) by about 10 mm
3. Increase the external width so that the drone and trompette strings don't make such a sharp angle over their nuts (note that you can't increase the internal width without modifying the instrument body)
4. Change the orientation of the machine heads so they can all be unscrewed easily with a normal-length screwdriver (if they need replacement)

You may wish to consider a redesign of the headstock if you attempt this build.

The wheel and crank are mounted on a ~25 cm length of M10 studding. This was cut from a longer length and it was important to make sure that it was very straight. This was tested by checking that it would roll freely on a flat table. The length isn't too critical but leave it too long and it takes a long time to remove the wheel, as the wheel locking washer needs to be unwound first. A lot of M10 hardware is used in this assembly. The bearing needs to be press-fitted into the rear bearing housing + peg.

The two nuts that engage with the centre of the wheel need to be left loose so that their flats can align with the hexagonal bore. There is a very small amount of longitudinal clearance in this bore, but any residual slack will be taken up very quickly once the wheel is turned and the nuts tighten down towards the crank. In principle reversing the direction of crank rotation would briefly cause the nuts to release but I haven't found this to be a problem (because I never do it!). Note the orientation of the wheel (its hub is larger on one side). You will need to get this correct when fitting the wheel assembly into the body of the instrument.

Finally assemble the crank with a ~75 mm M10 bolt, per the photograph. Make sure the handle can turn freely on this bolt once it is all assembled. I have access to a lathe and I found it helpful to reduce the height of the bolt head by a few millimeters to make it look a bit nicer.

The first step in constructing the body is to press the second bearing into the receptacle on the front bearing strut. Next, glue the 3d-printed front and rear bearing struts to the inside face of the upper body panel as shown (see note below). Note the positions of the bearings, relatively close to the panel. The flat surface of the rear bearing strut sits right against the edge of the body panel, whereas the position of the front bearing strut is less important. I positioned it about 20 mm away from the wheel slot. When planning the placement of parts I first marked a centre-line down the length of the panel (on the inside face) then measured from both sides of this to ensure symmetry before making faint alignment marks in pencil.

When using the PU glue, a light spray of water on the plywood side of the bond is recommended. Apply the glue sparingly to the PETG part. This glue doesn't set instantly so there is time to ensure that the alignment (particularly along the centre-line of the body) is exactly right. Once aligned, the join needs some pressure to prevent the glue foaming and shifting the alignment. I used either a g-clamp with a wooden spreader or that handy block of steel again. I only attempted to glue one part at a time, to make sure that everything was right. After a few minutes the glue starts to foam around the edges of the bond. I used a flat-bladed screwdriver to scrape this away, although it doesn't matter too much for internal bonds which won't be visible. The glue should be dry within half an hour but I typically left it as long as possible, just to make sure.

At this point it is worth taking the wheel assembly (with wheel removed to keep things simple) and testing the fit into the two bearing struts. If this has been done properly it will fit straight through both and turn easily.

The next step is to attach the headstock to upper body panel with PVA glue. There needs to be a 4 mm gap between the headstock and the end of the body panel to accommodate the drone and trompette nuts. During assembly I just used a plywood offcut as shown. The body panel should overhang the headstock by 4 mm on each side. Clamp or weight this part whilst the glue dries (I allowed 24 hours for PVA bonds).

Now the upper body panel needs some wooden beading around its edges so that the side panels can be fitted. The important thing here is to ensure that the beading is exactly 1 thickness of plywood from the edge of the upper and lower panels (i.e. 4 mm). To simplify this, I printed a couple of gluing guides on the 3D printer which would clip over the edges fo the body panels and overhang to a distance of 4 mm, allowing me to easily ensure that the beading was parallel to the edge of the body panel without using a ruler. The beading length wasn't critical, but it was important to ensure that the beading didn't get too close to the corners of the body panels. Having to shape the beading ends at the corners would just create extra work for little practical benefit. The beading was bonded to the body panels with PVA glue then clamped. This process is then repeated for the lower body panel.

Once the beading has been fixed to the upper and lower body panels, these parts can now be positioned and glued. This is quite tricky and needs a very flat surface and a couple of squares, to ensure that the upper and lower parts are aligned exactly. It's worth practicing a couple of times to get the process right before applying glue to the headstock and bearing struts (PVA and PU glues respectively). Once the glue is applied, the three contact points are put under pressure with weights to ensure a good bond. Remember to remove any excess PU glue from the rear bearing strut while it dries.

Note - after some months I realised that the string pressure on the bridge was causing the upper body panel to deform slightly on the crank side of the wheel. This caused a faint but annoying buzz when the drone string was vibrating. I designed a retro-fittable bridge support (basically an H-frame) which could simply be pushed into place inside the body to stiffen up this panel as shown. This solved the problem. You should consider adding this part prior to full assembly of the body and perhaps glue it into place.

The next step is to cut and fit the side panels. These are 10 cm wide. I cut a long length of plywood (with the surface grain crossing the cut line) to this width. This long length was offered up to the body, with one end located in the correct position. A square was used to mark out the position of the required cut at the other end. The cut was then made and the two ends of the resulting plywood rectangle were chamfered where necessary to ensure a close fit with adjacent panels. The rectangular panel was then glued into place against the wooden beading, using weights to apply pressure to the bond.

This process was repeated around the body for all the panels except the one on the left hand side of the wheel (i.e. facing the player). The panel was cut and chamfered as before, but it was fitted to the beading with four brass countersunk screws. This allowed it to be used as an access panel to the interior of the instrument, to allow the wheel to be removed if necessary and any other maintenance to be more easily carried out. To fit the screws, the panel was first fitted in place then I drilled four small pilot holes in the corners, through the panel and into the beading behind. The panel was then removed and its holes were opened up to the diameter of the screw shank, then countersunk to allow the heads to sit flush with the surface. The panel was then put in place and screwed to test for fit, then removed and set aside for the rest of the build.

There's no reason the access panel couldn't go on the opposite side from the player if preferred - I felt it was less conspicuous to an audience on the left-hand side.

In this step everything except the keybox will be fitted to the upper body panel. When I was putting this together, I had no idea whether it was all going to work as I wanted, so I prioritised getting the drone and trompette strings operating first. In reality the order is probably not too important once one knows it will work properly. Before gluing, I sought the best positions of all the bridges etc. by clamping them in place and testing the sound. This was a time-consuming process but those following this build won't need to repeat it unless they really want to.

I first chose to fit the wheel (temporarily). One of the nice aspects of this design is that the wheel assembly can be completely removed at any time for replacement / maintenance. The free end of the wheel assembly shaft (with no wheel or retaining washer / nut fitted) is first passed through the rear bearing strut. The wheel is then inserted into the slot in the upper body panel and the shaft of the wheel assembly is passed through the centre of the wheel. The wheel should sit towards the rear of the slot. Thread the retaining washer and nut onto the free end of the shaft, then pass the axle through the second bearing on the front bearing strut. The rear bearing housing should fit into the rebate on the rear bearing support when the shaft is fully inserted. Make sure the two loosely fitted M10 nuts engage in the centre of the wheel and don't stand proud. Crank the handle a few times to make sure these two nuts move towards the rear of the instrument until stopped by the wheel support. The wheel should turn freely without rubbing on the side of the slot. If it doesn't, the wheel support may need moving slightly by changing the position of the lock-nuts which hold it in place on the M10 axle. Once everything is in a good position tighten the retaining nut to hold the wheel firmly on the axle.

Drop the nut into place between the two sides of the headstock and secure it with two countersunk woodscrews.

To fit the drone and trompette string anchors, the positions of the fixing screws first need to be marked carefully on the rear part of the body. If these are located incorrectly, the strings won't exert the correct amount of pressure on the wheel. Make faint pencil lines at a distance of 17 mm from the edge of the rear bearing strut, on each side. The photograph should make this clear. The large anchor (for the trompette) goes to the left of the crank and the small anchor (for the drone) goes to the right hand side. The anchors are fitted by offering them up to the body so the pencil lines are visible through the fixing holes, then making indentations with a bradawl. Then 2 mm pilot holes were drilled through the rear panel and the wooden beading behind it, before fitting the parts to the body with countersunk screws of at least 12 mm length.

Refer to the drawing and the photos for the attachment positions and orientations of the three bridges. I found it best to make a small pencil mark to show the position of the parts prior to gluing with PU glue. It's probably best to assemble these parts (using screws and nuts as shown) prior to gluing as it will probably be quite fiddly afterwards, especially if the hexagonal retaining holes for nuts need clearing out. I used a combination of weights, levers (i.e. steel bars) and clamps to hold these in position as they were attached (one at a time). The trompette hook and wheel cover clips can be attached in the same way, also using PU glue, paying careful attention to the orientation (as shown in the photos). Don't forget to scrape off as much excess PU glue as possible from around the join before it cures.

The PU bond between PETG and plywood is very strong, but if you make a mistake it is fairly easily rectified. I stripped a thread on one of my earlier bridge designs so it had to be changed. I found a multitool with a narrow cutter was able to remove the offending part from the plywood with very little damage to the latter, so long as the vibrating blade was held flat against it so that it didn't dig in. Hopefully this won't be necessary as most of the trial and error has already been done.

Normally a hurdy-gurdy wheel has a wood veneer bonded to its circumference. MarcoCammozzo's build didn't appear to use this, judging by his video clip. As a half-way measure, I did some experimentation using high-quality wood-loaded PLA for the wheel. This generates models that look and smell like they have been machined straight from MDF and it was possible to get a very good finish on the outer surface. Unfortunately I ran out of the filament before getting to a design I was happy with and it had doubled in price when I next looked. So I was stuck with PETG and doing things the harder way.

Once the wheel has been printed, check that its circumference is nice and smooth. This shouldn't be a problem provided you randomize your z-step on your slicer prior to printing. Then take a strip of wood veneer which is about 500 mm long (this is available from auction sides, typically with adhesive on one side already which won't be used). Put a single diagonal cut on one end to create a sharp point. Next, wrap the veneer around the wheel so that the ends overlap, with the diagonal cut on top. You can then mark a line where the second diagonal cut should be made to mate exactly with the first, once the veneer is glued.

I used PU glue to bond the veneer to the wheel. This is a messy process, particularly as the veneer is likely to be wider than the wheel thickness. Apply PU glue to the circumference of the wheel, give the veneer a light spray of water on its inside face then fit it around the wheel so that the diagonal ends are flush. After much trial and error I found that wrapping the veneer with LOTS of thin elastic bands was the most effective way to apply uniform pressure over the whole circumference (thanks to The Gentle Flamigo - https://www.youtube.com/watch?v=zXOwQPcvIEI). Make sure that you press the wheel (with veneer) down onto a flat benchtop to align one edge, prior to the PU glue going off. Once aligned, lift the wheel away from the bench and support it by its centre somewhere unless you want it glued to the bench!

Once the glue has cured, the elastic bands can be removed. Then use a chisel to remove the excess veneer and PU foam from both sides of the wheel. Check for any lumps and bumps in the veneer - in extreme cases you may need to make another wheel and try again.

Note that I went for a thinner wheel than in the original MarcoCammozzo design. This was because I found that the thick wheel, whilst generating a nice loud sound, didn't work well on the high notes as the string got lifted away from the front edge of the wheel. Making the wheel thinner (i.e. effectively moving its keybox-facing edge closer to the bridge) seemed to reduce this problem, although I think it was also due to the relatively large separation between the melody strings on this design.

The next step is to fit the wheel back into the body and tighten up the nuts on the shaft. Then secure the rear bearing housing + peg onto the rear bearing strut with 4 off M4 screws and nuts. The nuts should fit into retaining holes on the inner face of the rear bearing strut. You shouldn't need to remove the wheel again unless you make a mess of the next part, which is dressing the wheel surface.

Firstly, attach some masking tape round the edge of the wheel slot so that no bare plywood is visible near the circumference of the wheel. Then take a new Stanley-knife blade and rest it on the masking tape with the blade angled slightly downwards and lightly touching the veneer on the circumference of the wheel. The blade's edge should be exactly parallel to the axis of rotation of the wheel. Press the blade against the masking tape to stop it moving, then start to crank the wheel rapidly. The blade should scrape away any high-spots on the veneer. This is an iterative process but should finish up with small heaps of fine sawdust (both inside and outside the body) and a very smooth wheel. Don't be tempted to take off too much material in one go, as you may dig the blade into the wheel. When this is done, remove the masking tape and clean off any dust which has collected. Here is a very useful video from Jimi Hellinga (aka Elektrovolt) which explains the process in more detail (along with a lot of other useful information for later).

https://www.youtube.com/watch?v=JkPwHBvdx8w

It's quite hard to carry out a build like this in a simple linear fashion, especially when a lot of modifications are being made on the fly and adjustments need to be made to get the best sound. Therefore at this point I put some rosin on the wheel, to allow me to get the string tuning approximately correct. I took a block of rosin and applied light pressure to the wooden surface of the wheel, whilst cranking the handle rapidly for ~5-10 turns. Then I took a thin cloth and firmly pressed a single thickness against the wheel surface with my finger, and cranked vigorously until I could feel my finger getting very warm. The idea of this step is to remove loose rosin and to melt and smooth the remaining rosin so it forms a uniform layer over the wheel. The video link above also details this process.

The keybox is probably the most fiddly aspect of this build. If I were doing this project again, I would have designed the keybox from scratch to make it easier to assemble and to add additional features such as hinge mounting points for the keybox lid. There are lots of parts which have to go together with very careful alignment if the keys are to work properly. I did most of the assembly on a flat bench before gluing to the body, because if I made a mess of things I didn't have such a major rework on my hands.

The first step is to join the two halves of each side of the keybox with PU glue. For this I made a make-shift jig comprising a length of 50 mm aluminium angle with a wooden block fixed to the base with a cable-tie to locate the two halves of each keybox side together. A large block of steel was used to provide a compressive force against the adhesive. The whole thing was leaned at a 45 degree angle from the ground to keep the parts aligned in both planes and pressed together. One of the key holes is very close to the join so care must be taken to avoid excess glue in this area. Once cured, any excess adhesive / foam was removed with a chisel.

Next, I roughly assembled the three keybox support pieces and the two keybox sides to get familiar with how they should fit together. Note the orientation of the support piece which goes closest to the headstock in the finished instrument. The keybox sides are curved on one long edge; this edge should sit neatly into the supports, with the flat end of the keybox sides pointing towards the headstock. The support pieces should be adjusted so that they are 10 mm in from each end of the keybox, with the middle support directly under the join between the two halves of the keybox sides. Once alignment is satisfactory, the keybox can be glued with PU glue. A very flat bench is a pre-requisite for this process. Assembly is probably best done by starting with the front keybox support, as it has a raised inner section which can be used to clamp the sides against. Downward force is also necessary prior to tightening the clamp, to ensure that the sides are fully bedded down on the support piece. The keybox sides need to be aligned very carefully using a square (or two) well before the PU glue dries, so that the holes for the keys line up - otherwise it will look terrible and the keys won't fit.

At this point the second and third keybox supports were glued into place. This was much easier as the keybox sides were now fairly rigid, having been glued onto the front support. No clamps were required but a weight was used to ensure a good bond.

Next the keybox assembly was glued to the upper body panel. Note the orientation - chamfered support piece towards the headstock. The flat end of the keybox was butted right up against the nut and care was taken to ensure that the keybox sat exactly on the centreline of the instrument. PU glue and weights were used to make the bond.

The keybox wheel end part was then glued in. The curved lower edge of this part faces outwards (towards the wheel) to match the rest of the keybox and the upper edge (with the string slots / hooks) had to line up exactly with the top of the keybox sides. This part has nothing to locate it in place, so very careful positioning and clamping was required during gluing.

Once the structure of the keybox was complete, the keys could then be fitted. Black keys went into the bottom row of holes and white keys went into the top row of holes, with the largest keys going towards the headstock. I used Freecad to generate the STL files; this doesn't support multiple parts so there is a very thin join between each key. This needs to be cut away prior to fitting. Every key was individually checked for rough edges prior to insertion and care was taken to ensure a sliding fit in the corresponding keybox hole. The black keys closest to the wheel needed to be installed as a group, as the keys overlap the adjacent shafts. Orientation is also important - the keys should be installed as printed, or they may not all fit properly. It's probably worth getting all the keys into position before fitting any of the tangents. One of the high-pitched white keys also has a slot in its shaft to allow the tangents from the black key underneath to slide past. Refer to the photograph of the completed keybox.

I populated the keybox from the headstock and moved down towards the wheel. Each key needs two tangents fitting. The short tangents are for the white keys and the long ones are for the black keys. These were screwed onto the key shafts using M3 screws (16 mm for the white and 30 mm for the black). It's worth putting the screws all the way through the tangents (a t-bar hex key is recommended as there is some effort involved here) and making sure they turn fairly freely prior to fixing to the key shafts. The 3 mm holes in the key shafts were sufficiently undersized to allow the screws to bite firmly, but it would be easy to strip the resultant thread in the PETG if the screw were over-tightened. The narrow edge of the tangent (i.e. the bit which will fret the string) should be centrally located on the key shaft. Note that a few of the tangents are very thin - these are fitted to the last four key shafts close to the wheel end. The thin tangents on the highest white key will need to be twisted slightly towards the slot in its shaft to ensure the correct tuning. This is just visible in the photograph. I kept myself motivated during this part of the build by having the melody strings in place so that I could test each key as it was installed.

A thin strip of self-adhesive velvet was attached to the inner right hand side of the keybox, above the white keys. This was to damp the noise made by the tangents as they drop against the side of the keybox during playing. Don't allow this to extend beyond the outermost tangents, in order to leave space for the keybox lid locating pillars as described below.

While I was making the gurdy I didn't really give much thought to a keybox lid, particularly as MarcoCammozzo seemed not to need one. I quickly found that it was essential, however. Unfortunately there was no provision in the 3D-printed parts for hinges or other fixings. I first made a lid from 4 mm plywood, shaped to match the contours of the keybox, then inlaid six neodymium magnets around its edge. I also needed to put matching magnets into the top of the keybox. The inlaying was done using a 6 mm end-mill, fitted into a pillar drill chuck. The depth-stop on the chuck needed to be set to just less than the depth of the magnets (about 1.5 mm). Careful measurement of position was needed, as well as ensuring that the north-south orientation of the mating pairs of magnets was correct (so they would stick together). I glued these into place with two-part epoxy resin. The magnets worked fairly well but wouldn't prevent the lid sliding during play, so I fitted four very small brass pillars into the corners of the lid. These just fitted inside the keybox mechanism without fouling the tangents and completely prevented any lateral movement. So while I got away with it in the end, a redesigned keybox would have been a better approach.

Note that when I was designing the keys themselves, I didn't pay too much attention to the physical length of each key or its absolute position (both longitudinally or laterally), preferring to adopt a layout that looked roughly right. Fortunately it seems there are no hard and fast rules about this. The nice thing about this particular machine is that it's a easy matter (albeit time-consuming) to redesign, reprint and replace all the keys so that they suit the player better.

The next step is to fit the strings. If you are anything like me you will have tried this much earlier in the build, just to get a sneak preview of the sound. A guitar tuner or an app which does the same thing is very useful here.

The first step was to hook the tailpiece over the hook on the rear bearing housing, where the crank emerges from the body. The tailpiece is fairly loose-fitting so will rattle a bit until the strings are fitted and tensioned.

The figure shows which strings are fitted where on the hurdy-gurdy, along with their target note and approximate frequency. I used the bottom three strings from a Low G Tenor Ukulele set, plus a length of nylon strimmer wire. All the strings were anchored using the holes located in the tailpiece or the string anchors at the crank-end of the instrument. A standard classical guitar knot was used to retain them (see the internet for some options).

Next, the strings were passed over the bridges, the wheel and the nut at the headstock end. They were then secured on the machine heads - again, see the diagram for details. There are numerous on-line resources on how to attach a string to a machine head so I won't go into details here. If you are like me, you will like your machine heads to all turn clock-wise for an increase in pitch, so I suggest passing the melody strings under the spindles of the corresponding machine heads and passing the drone & trompette strings over theirs. Careful reference to the photos shows that in my build there is probably more stress on the strings than is necessary; this could be rectified with a headstock redesign, but I'm willing to tolerate it for now. Adjust the machine heads to remove any slack in the strings, without applying any significant tension.

When the melody strings are fitted and tensioned there is a significant force exerted on the main bridge which tends to push it into the wheel. To counter this a loop of thin wire is installed which threads round the top of the bridge and through a pair of holes in the tailpiece. The ends need to be twisted together on the top of tailpiece to form a closed loop. A short length of rod (wood, metal, whatever) is placed inside the loop between the bridge and the tailpiece and rotated several times to create twists in the loop on either side of the rod. At the same time, the rod will be gripped slightly by the wire. Each turn of the rod will apply an increasing force between tbe bridge and tailpiece, so don't overdo it prior to tensioning the melody strings. The rod can be removed when no longer needed, if the wire is sufficiently stiff not to unwind over time.

I applied tension to the drone string first as it needs to vibrate at the lowest frequency, thus the applied force would be lowest of all the strings. The drone bridge is a roller with two slots in it. One is for allowing the string to be bowed by the wheel and the other is for lifting it clear of the wheel, thus muting it. The position of the roller can be adjusted by rotating the M4 cap-head screw which holds the roller in place on the bridge via an embedded captive nut.

The next string to be fitted was the trompette, which requires significantly more tension than the drone. Prior to tensioning the string, the 'dog' must be installed under between the trompette bridge and the string. The string sits in the v-groove on the top of the dog, and the rear of the dog sits loosely in the trompette slider. The slider can be adjusted back and forth using an M4 cap head screw with captive nut. Although there's a lot of tension on the trompette, the bridges and anchors that support it are fairly robust so I wasn't too concerned about this aspect.

An important part of the trompette set-up is the fitting of the tirant, which is a string and peg arrangement which imparts a lateral force onto the trompette string, to change the amount of pressure against the wheel. This helps tune the characteristic buzz generated by the dog when the wheel is accelerated sharply. I used approximately 20 cm of very thin cord which was previously used to seal the top of a rice bag, so nothing special. This was folded in half and tied in a cow hitch around the the trompette string, leaving the two free ends loose. The tirant peg was inserted into the top of the tailpiece and the free ends of the tirant cord were passed under the tailpiece and through the slot in the end of the tirant peg. A few twists of the peg were sufficient to grip the cord firmly and allow variable tension to be applied to the trompette string.

If the wheel has been rosined, it should be possible to get both the drone and trompette to vibrate when in contact with the moving wheel. I slowly applied tension to both until they were tuned to C3 and C4 respectively (approximately 131 Hz and 262 Hz). Tuning doesn't need to be perfect at this stage. Accelerating the wheel sharply should cause the dog to buzz - if not, adjust the tirant peg slightly. It may also be necessary to adjust the v-groove in the top of the dog to get the best sound - I used a triangular file the remove a little bit of material and deepen the groove slightly.

The next step is to apply tension to the melody strings, to reach G3 and G4 (196 Hz and 392 Hz). The wire-wound string is the lower pitch of the two. This part was quite scary as a significant amount of tension was required across entirely 3D-printed parts, in particular the small hook which retains the rear end of the tailpiece. I needed two iterations of the hook design before I got something robust enough not to break or creep over time. The current iteration is still going strong after 18 months. I was also worried that the plywood body would just fold in half under the tension, but so far I've had no problems at all.

The bridge height needs to be carefully adjusted using the two set-screws such that the string passes almost flat over the edge of the wheel. Clearly a modicum of pressure is required on the wheel for the strings to play properly, but much fiddling is required to get it right. If the bridge is too high, it either won't play at all or the low G string will get pushed off the wheel when attempting high notes. It's not uncommon to see accomplished players using instruments with bits of paper stuffed between strings and bridge to shim them to the right position.

Getting a good 'coup' sound (this is when the dog / buzzing bridge vibrates when the wheel is accelerated) was quite tricky. The dog in this build is made from PETG, which is slightly flexible. I think I may eventually replace it and the support structure with hardwood, in order to make a sharper sound.

At this stage, provided everything has gone smoothly, the gurdy is probably capable of making some horrid noises and sweet notes in equal measure. Improving the sound is discussed in the next step.

At this point I finally had a working hurdy-gurdy to start messing around with. If you follow these steps, you will probably now have a fairly unpleasant-sounding cat-scarer on your hands, as I did.

The sound can be improved considerably by applying a small amount of cotton fibres around the contact point between the string and the wheel. This should only be done when the string is approximately in tune, as the cotton will move out of position if the string tension is changed much. I used cotton wick material which I obtained from an E-cigarette shop. This had nice long fibres and was reasonably cheap. Cotton wool is not apparently advised for this job as the fibres are too short and won't wrap round the strings easily.

Rather than try to explain it, I'll refer you to the previous Elektrovolt video once again, which also describes how to apply cotton to the strings.:

https://www.youtube.com/watch?v=JkPwHBvdx8w

I found that getting the amount of rosin and cotton just right makes a huge difference to the sound quality, as does the height of the bridge. I found it useful to look at the vibrating strings inside the keybox whilst setting this up. They should vibrate at their fundamental frequencies (i.e. with the maximum amplitude in the middle of the string); get something wrong and one or both might vibrate sporadically at their first overtones (null in the middle, with a peak on either side).

I didn't find it necessary to tweak the tangent positions - eyeballing them to lie along the centres of the key shafts was sufficient for me to be happy with the sound. I checked the original keybox design against an on-line fret calculator and it was spot on. One can do fine-tuning with a guitar tuner or an app. I found that the strings held their tuning very well except for the drone. The drone almost always needs a small tweak after any significant inactivity. This is hardly a surprise as it was obtained from a piece of garden machinery rather than from a music store! I experimented with electric bass guitar strings but the drone was then so loud that it drowned out the other strings.

I strongly advise that you refer to the on-line hurdy-gurdy community to really understand the set-up process. I found it a bit like setting up a 3D-printer: lots of interdependent variables, all critically important to the overall outcome. Patience and persistence are key to finding the sweet spot; I'm reasonably happy with how my machine sounds (given my very poor playing skill) but I know I could probably get it optimised, given time and effort.

In order to smarten up the overall appearance of the instrument I applied some home-made beeswax polish to the wooden surfaces of the body, headstock and keybox. This really brought out the grain of the plywood. Ideally any painting or decoration should be carried out during assembly, but as I was on a journey of discovery I couldn't be sure a) whether it was worth the effort; b) exactly when to do the decoration. It's difficult to paint or polish the instrument before all the 3D printed parts are glued on (because they might not stick properly otherwise), but obviously it's harder to decorate neatly if you have to work round all the fittings. So I plumped for a bit of beeswax at the end and I was happy enough with the outcome.

I printed the wheel cover from some low-quality 'wood' PLA. This appeared to have negligible wood content, so I decided that it needed covering. I had some veneer left over from making the wheel, so I modified the wheel cover to have a width exactly equal to two widths of the veneer, such that I could cover it without making any additional cuts. I then glued the veneer strips with PU glue and used clothes pegs to hold it in place. I did one side at a time. Given the lack of cutting, it was possible to get a very good join between the two pieces of veneer. The PLA had just enough resilience to snap in and out of the wheel cover clips with the application of modest pressure.

I found it quite tricky to play the hurdy-gurdy at all in a sitting position unless I wedged it somehow between my legs and midriff. Otherwise the body of the instrument just shifted as I cranked the handle. So a strap of some sort to go round my waist was obviously required. Proper hurdy-gurdy straps are more like a three-point harness (pretty much essential for playing while standing up). I wanted to keep things simple in the first instance, so decided to fit a standard guitar strap. I bought a couple of standard off-the-shelf guitar-strap buttons. The first was fixed centrally to the forward face of the headstock. The second was fitted to one of the rear body panels, on the left hand side of the crank. Here the plywood is only 4 mm thick so some reinforcement was required. I cut a small length of wood slightly under 10 cm in length and removed a couple of the corners as shown in the photo. This was glued into position inside the instrument body (using PVA) at the point where I wanted to mount the button. I could then drill a pilot hole through the body wall and the extra piece of wood, in order to provide a good anchor for the guitar button screw.

Finally all that remained to do was attach a thick guitar strap and screw the access panel into place with neat brass countersunk screws to complete the build.

In the preceding steps I've described the process I went through to create a working hurdy-gurdy for a modest capital outlay. I was very happy with how it turned out - it put me in a position to start learning to play without breaking the bank. Unfortunately life is very busy and my innate musical ability is negligible, so progress is glacial at present. This is obvious from the attached video - after many attempts I gave up trying to get those few bars of Breton An Dro without fumbling something. I'll get there in the end, I'm sure...

I acknowledge once again the work of MarcoCammozzo. His design gave me the initial leg-up and showed me what was possible with a little bit of effort and a 3D printer. I also acknowledge all the information shared by the on-line community of hurdy-gurdy makers and players, who made my task much easier than it could have been.

Having completed this machine, I would probably make the following changes if I were to do it all again:

1. Reduce the lateral separation of the melody strings slightly
2. Add provision for keybox lid hinges
3. Make keybox easier to assemble
4. Provide levers to lift strings from wheels
5. Provide capos for drone and trompette
6. Rework buzzing bridge with harder material (i.e. wood)
7. Modified headstock to improve string angles against nut, access for machine heads etc.
8. Incorporate some stiffening in the upper body panel, near the bridge
9. Find a better way to cut plywood, to avoid ragged edges
10. Decoration (now I understand the build order)

On that note, I'll bring this instructable to a close. I hope you found it interesting and thanks for reading it. Please let me know if you build one of these and do post a picture! All the best...

dr_phil