A wall panel forgives you. Cut it a millimetre shallow and nobody knows. A tray does not forgive anything, because the part that matters is not the carving, it is the hole. Get the pocket two millimetres too deep and you cut through the floor, and there is no sanding your way out of that.
That is also why a tray is worth making. It is the first project where you stop being someone who carves pictures and start being someone who makes objects, and it takes about an hour.
This guide follows one tray the whole way, in eight stages, from the file on the screen to olives on a serving board. Every photograph below is the same piece of walnut at a different point in its life, so you can see exactly what each pass leaves behind and judge whether yours looks right at the same moment.
What this guide covers
- What a valet tray actually is
- Before the machine: stock and the three numbers
- The eight stages, start to finish
- Feeds and speeds for four classes of machine
- The food-safe question, answered properly
- The designs
- Common questions
What a valet tray actually is
A valet tray is the dish by the front door or on the bedside table that holds what comes out of your pockets: keys, a watch, a wallet, loose change, a folding knife. The everyday carry world calls it an EDC tray. Older catalogues call it a catchall. They are the same object.
Structurally it is a shallow rectangular or oval dish with a recessed floor and a raised lip, usually with a carved scene in the bottom so there is something to look at when the tray is empty. Typical outside size is 200 to 260 mm long, 130 to 170 mm wide, and about 25 mm thick. The one in these photographs is 230 by 150 mm in 25 mm walnut, with one main pocket and two small wells at one end for keys and coins.
It is the ideal first pocketing project for three reasons: it is small enough to cut from an offcut, the geometry is forgiving everywhere except depth, and it is a genuinely good gift, which matters when you are learning on material you paid for.
Before the machine: stock and the three numbers
Everything else in this guide is detail. These decisions are the project, and all of them are made before the spindle turns.
The stock
Walnut, cherry and maple are the classic three, and they are classic for a reason: tight grain, no tear-out in a pocket corner, and they all finish beautifully with oil. Oak works but the open pore structure collects dust in the carved floor and looks coarse at this size. Pine will cut, will dent the first time somebody drops keys in it, and will not be a gift anyone keeps.
The mistake that wastes boards is not flattening the stock first. A tray is wide and shallow, so any cup or twist in the board becomes a pocket that is 11 mm deep at one end and 15 mm at the other. The machine cuts a perfectly flat pocket into a board that is not flat, and you discover it when you break through on one corner. Surface the top face before you cut, or start with stock you trust.
Run the long grain along the long axis of the tray. A tray with the grain running across its short dimension is a tray with a weak lip.
1. Stock thickness
Use 25 mm (one inch) stock. Thinner looks mean and flexes; thicker wastes material and cutting time for no visible gain. If you only have 19 mm, the tray still works, you simply lose depth in the pocket.
2. Pocket depth
12 to 15 mm on 25 mm stock. That is deep enough that keys do not slide out and shallow enough to leave a solid floor.
3. Floor thickness, which is the one that bites
Whatever is left underneath. On 25 mm stock with a 14 mm pocket you have 11 mm of floor, and that is comfortable. Never leave less than 8 mm. Below that the floor will cup as the wood moves.
If you would rather not do this arithmetic by hand, drop the file into our free STL checker. It reads the real depth out of the model, tells you the board thickness you need, and warns you if the model is lying the wrong way up. Nothing is uploaded, it runs in your browser.
Say the next part twice, because it is where people cut through: if the design has a carving in the bottom of the tray, measure from the lowest point of the carving, not the top of the pocket. A 14 mm pocket with 4 mm of relief detail sunk into the floor leaves you 7 mm of floor, not 11. The deepest point of the carving is your real floor thickness.
The eight stages, start to finish
What follows is one tray, photographed at each stage. If your workpiece looks like the picture at the same point, you are on track.
Stage 1. The file on the screen
Everything exists here and nowhere else: a model, a plank and two cutters. This is the cheapest possible moment to find a mistake, so spend a minute here rather than ten minutes with the saw later.
Three things to check on the screen before you commit. First, the depth of the deepest point of the carving against your stock thickness, using the numbers above. Second, that the model is sitting the right way up, because a relief modelled to be cut from the top will make no sense flipped. Third, the overall size against your real cutting area, not the advertised bed size.
Notice the two cutters lying on the board. They are the whole project: a square-tipped 6 mm flat end mill that removes material fast and leaves steps, and a round-tipped 3 mm ball nose that removes material slowly and leaves shape. Every stage that follows is one of those two doing what it is good at, and most bad tray cuts come from asking one to do the other's job.
Stage 2. Roughing the pocket
The flat end mill is hogging out the main pocket in layers, and the picture shows exactly what that leaves: stepped terraces stacked like contour lines, and a bass you can only just recognise as a blocky mass. The shape is there. The carving is not. That is correct at this stage and it should worry you not at all.
Use a pocket toolpath cutting 3 to 5 mm per pass on a hobby machine, more if your machine is rigid. Leave 0.4 mm on the walls and floor for the finishing pass, so you are never asking the roughing bit to leave a show surface.
Offset or raster clearing both work. Offset gives a tidier floor on a curved pocket, raster clears rectangular pockets faster, and it matters far less than people argue about online.
Look at the chips. Roughing with a flat cutter throws big curled shavings, pale and springy, and they pile up on the spoilboard. If your roughing pass is producing dust instead of curls, your feed is too slow for your spindle speed and you are rubbing rather than cutting, which burns the bit and scorches the walnut.
The corner problem. Here is the thing nobody mentions until it has already happened to you: a 6 mm bit cannot cut a corner smaller than 3 mm radius. If your design has crisp internal corners, your tray will not have crisp internal corners, it will have 3 mm rounded ones. Two honest answers: design the tray with radiused corners in the first place, which looks deliberate and is what most good trays do, or run a rest machining pass with a 3 mm bit to clean out what the larger bit could not reach. Rest machining only cuts the leftover material, so it is quick.
And look at the clamps. Four low-profile step clamps, one near each corner, bolted into the T-slots with their pads on the waste margin well clear of the tray outline. The board is deliberately larger than the tray so the clamps have somewhere to sit. A board being pocketed is a board that wants to move, and tape and a prayer is not workholding.
Stage 3. The finishing pass
Now the ball nose goes in, and this is the pass that turns a stepped hole into a carving. Watch the two halves of the pocket in the photograph: smooth, fully modelled scales and fins where the cutter has already passed, and the rough terraces from stage 2 still waiting on the far side.
A 3 mm ball nose, stepover around 8 to 10 per cent of the bit diameter. Halve the feed you used for roughing. This single pass takes longer than everything else in the project combined, and that is normal. Run it after clearing and before the profile, while the workpiece is still fully supported by the waste around it.
Note that the tray is still buried in the middle of the board at this point, with uncut waste all the way around it and the clamps biting on that waste. The outside shape does not get cut until the pass after this one, and there is a good reason for that order: a part that is still part of the board cannot chatter, and chatter is what puts ripples in a finished carving.
The same pass also cuts the hewn border around the rim, which is why the frame above still shows it as a plain sunken groove while the bass beside it is already finished. That border is a texture, not a shape, so it costs you nothing extra in setup and it hides the small imperfections a rim collects during handling.
The waste changed, and that is the tell. Roughing threw curls. Finishing throws fine pale dust that settles into the carved hollows. A ball nose taking a tiny stepover is shaving, not chewing. If your finishing pass is throwing chips, your stepover is far too large and you will see every one of those steps in the finished tray.
Resist the urge to go finer than 8 per cent to "get more detail". Past that point you are adding an hour of machine time to produce ridges your sandpaper would have taken off in thirty seconds.
Stage 4. Cutting it free
The profile is cut, the machine is stopped, and the tray is still attached to the board by six small bridges of walnut. Those tabs are the only thing that stopped your finished tray becoming a projectile on the last pass.
Most people use too few and too thin. For a tray this size use six to eight tabs, 6 mm wide and 3 mm thick, spread around the perimeter with two on each long side. Put them on the flat straight sections, never on a curve or a corner, because a tab on a curve tears a crescent out of the edge when it breaks.
Cut the profile in layers with the same 6 mm end mill, going about 0.5 mm past the bottom of the stock so you get a clean edge rather than a fuzzy fringe.
Free the tray with a flush trim saw or a sharp chisel. Do not twist the part out. Twisting is how a tab takes a chunk of your lip with it, and that chunk is always on the side that will face the room.
Stage 5. Sanded raw
This is the stage that makes people think they have ruined it. Off the machine, sanded, unfinished: the walnut is pale, dry, dusty and flat, and it looks nothing like the wood you bought. Keep going. This is the dullest it will ever look, and the next stage is the one that pays you back.
The whole appeal of a machined tray is the crisp step where the floor meets the wall, and that step is exactly what careless sanding destroys. Sand the top face and the outside profile normally, 120 then 180 then 220. Inside the pocket, wrap the paper around a flat block for the floor and keep the block away from the walls. For the walls themselves, fold the paper and run it vertically down the wall, not in circles.
The carved floor is the exception. Use a soft pad or a brass brush and accept that you are cleaning fuzz rather than smoothing. Over-sanding a relief is how the detail you paid for disappears, which we cover in finishing CNC relief carvings.
Blow or brush the dust out of the carving before you oil. Oil turns trapped sanding dust into a pale paste that sets in the recesses, and picking that out afterwards is a miserable evening.
Stage 6. The oil
This is the photograph worth taking every time. Half the tray is oiled and has gone deep chocolate brown with the grain blazing and the carving suddenly reading as depth rather than pattern. The other half is still the pale dry board from stage 5. The line between them runs straight across the tray.
Nothing about the wood changed. The oil filled the surface and stopped it scattering light, so you finally see what was always in the board.
Apply it thin, let it sit for fifteen minutes, then wipe off everything that has not soaked in. The most common finishing mistake is leaving excess oil to dry on the surface, where it stays tacky for weeks and collects dust. Two thin coats beat one thick one every time, and on a carved floor use a brush or a cloth over your fingertip to work oil into the recesses rather than flooding them.
Which oil depends entirely on whether food will ever touch the tray, and that question has a proper answer further down.
Stage 7. In use
A watch, a set of keys, a folding knife, dropped in rather than arranged. This is the test the tray was built for and the reason the depth number mattered: at 14 mm the keys stay where you put them, and at 8 mm they walk out onto the table overnight.
One practical last step. Glue a thin felt or cork base to the underside. It stops the tray marking a polished dresser, it stops it sliding when somebody drops keys in from height, and it hides a patched breakthrough if stage 2 went badly.
Stage 8. Where it leads
Once you have cut one pocket, cutting five is the same job with different numbers. That is where the serving trays come in, and they are quietly our second best selling line: the Highland Cow tray and the Octopus tray with eight tentacle wells have between them sold more than seventy times this year.
The differences from a valet tray are small but real:
- Wall thickness between compartments. Never below 8 mm, or the divider snaps the first time somebody leans on it. 10 to 12 mm looks right and survives.
- Equal depth everywhere. Compartments of different depths look like a mistake even when they are deliberate. Pick one depth and keep it.
- Bigger stock. A serving tray wants 300 to 400 mm of length, which is where people discover their machine's real work area is smaller than the specification.
- Smooth floors where food sits. A carved relief floor holds moisture and crumbs. Carve the surface between the wells instead, exactly as the tray in the photograph does.
- Finish before fitting any handles, because oil under a fitting always shows.
The Alligator tray with four compartments is the easiest of these to cut well, because the compartments are large and the dividers are generous.
Feeds and speeds for four classes of machine
These are starting points for hardwood with a 6 mm two-flute end mill. Your machine, bit and bearing wear all move these. If the cut sounds like screaming, the feed is too slow for the spindle speed, not too fast.
- Hobby, belt-driven (3018, small rail machines): 10,000 rpm, 800 mm/min, 1.5 to 2 mm depth of cut. Be patient, take more passes.
- Mid hobby (Shapeoko, X-Carve, Onefinity): 16,000 rpm, 1,800 mm/min, 3 mm depth of cut.
- Prosumer (Avid, ball-screw machines): 18,000 rpm, 3,000 mm/min, 5 mm depth of cut.
- Industrial (ATC, vacuum bed): 18,000 rpm, 5,000 mm/min or more, full 12 mm depth in one pass with the right bit.
For the 3 mm ball nose on the carved floor, halve the feed and expect the pass to take longer than everything else combined.
The food-safe question, answered properly
This matters the moment somebody puts cheese on your tray, and most answers online are wrong in one direction or the other.
The rule that actually applies: almost every common finish is food safe once fully cured. The problem is not the dried film, it is the solvents while it cures, and cure is not the same as dry to the touch. A finish that is dry in four hours may take three weeks to cure.
For a tray that food will touch, the honest short list:
- Food-grade mineral oil. Cheap, safe immediately, zero cure time, and it needs redoing every few months. The traditional chopping-board answer.
- Mineral oil and beeswax paste. The same thing with a wax layer that resists moisture better and feels far nicer. This is what I would use, and it is what went on the tray in stage 6.
- Pure tung oil or raw linseed oil. Genuinely food safe once cured, but cure takes two to four weeks and you must not rush it. Note that boiled linseed oil contains metallic driers, so read the tin.
- A hardwax oil such as Rubio Monocoat or Osmo. Durable, beautiful, certified food safe when cured, and more expensive.
What to avoid on a food surface: anything with added driers you cannot identify, any spray lacquer not sold as food contact, and polyurethane, which is fine when cured but chips into flakes at a pocket edge and looks wrong on a tray.
One thing that is not about finish at all: do not put wet food directly in a carved relief floor. The recesses hold moisture, and a carved tray is for dry food, service and display. If you want a tray that takes olives and oil, use a design with plain smooth compartments.
The designs
Every one of these is a single STL with the pocket, the floor relief and the outside profile already modelled, so your only job is the toolpaths.
Valet trays, for the bedside or the hall table:
- Largemouth Bass and Fishing Lure valet tray, the design cut in the photographs above
- Leaping Brown Trout fly fishing tray
- Mallard Duck marsh scene tray
Serving trays with compartments, for food and the ones that actually sell:
- Highland Cow serving tray with snack compartments
- Octopus serving tray with eight tentacle wells
- Alligator serving tray with four compartments
- Hummingbird serving tray
- Sea Turtle serving tray
Common questions
I cut through the floor. Can it be saved?
If it is a small breakthrough in a flat area, glue a thin patch of the same wood underneath, sand flush and put a felt base on the tray. Nobody will know. If it is through the carved relief, start again and add 3 mm to your floor next time.
My roughing pass left steps I cannot sand out. What went wrong?
Nothing, if you stopped there. Those steps are what a flat end mill leaves and they are meant to be removed by the ball nose in stage 3, not by sandpaper. If you have already run the finishing pass and the steps are still visible, your stepover was too large or you left more than 0.4 mm of roughing stock for the finisher to clear.
Can I cut a tray with only a 3 mm bit?
Yes, and it will take three times as long. A 3 mm bit clearing a 14 mm pocket is a lot of passes and a lot of heat. If it is all you have, reduce depth of cut and let it work.
What size tray fits my machine?
Measure your real cutting area, not the advertised bed size, and subtract the clamps. A 300 by 300 mm machine cuts a comfortable valet tray and struggles with a full serving tray. Scale the design down rather than forcing it, and remember that scaling a relief changes its depth too, which we cover in scaling STL files without losing detail.
Should the pocket floor be flat or carved?
Flat is more useful and carved is more beautiful. For a valet tray that holds keys, carved is fine because the items are dry and hard. For anything that touches food, choose flat compartments.
How long does one take?
On a mid hobby machine, roughly 12 minutes to clear the pocket, 25 to 40 minutes for the carved floor, 6 minutes for the profile. Call it an hour of machine time and half an hour of sanding and oiling.
Can I sell trays I cut from these files?
Yes. Every design here includes a commercial licence for the physical items you cut. You may not resell the file itself.