Guides / For Machine Shops
Five Things to Check Before You Quote a Case Model
September 25, 2026
Nearly every first keyboard case model arrives with the same five problems. None of them are incompetence. They are the things you only learn by having a run go wrong, and this customer has not had one go wrong yet.
Find them before you quote. It takes half an hour, it lowers what you have to make, and it makes you the shop that knew something they needed.
One: no finishing allowance on the tight fits
The most common and the most expensive.
The plate pocket, the weight recess and the gasket shelf are drawn as precise fits against the mating part. Then the case gets anodised, or worse, e-coated, and the coating goes on both walls of the pocket. The pocket closes by twice the build and the plate will not go in.
The model is not wrong. The drawing is missing one line: whether dimensions apply before or after finishing.
Ask which. If they have not thought about it, tell them what the coating adds and let them decide whether to open the pocket or dimension after finishing. This conversation has saved more first runs than any other on this list.
Two: a blanket tight tolerance
A title block that says plus or minus 0.05 mm across a whole case, because tighter felt safer.
It is not safer, it is expensive. It forces slower finishing passes, more in process measurement and more scrap, on twenty features where half a millimetre would have been fine.
What they actually need tight is a short list: the gasket shelf depth, the plate pocket fit, the screw and boss pattern position, the port cutout position and plate flatness. Everything else can run at a normal block tolerance.
Offer the trade explicitly. Quote as drawn, then quote with a normal block tolerance and five callouts, and show them both numbers. They will take the second one and remember who offered it.
Three: the PCB was never measured
The case exists to fit a circuit board. Very often the board was designed by somebody else, or bought, and the case was modelled from a drawing of it rather than from the physical part.
Three things go wrong. The standoff or boss pattern does not match. The USB-C cutout is positioned for the socket but not for the moulded boot on a thick cable. And something standing up on the PCB fouls an internal wall.
Ask for the PCB outline even though you are not machining it, and ask whether the case was checked against a real board or a drawing. If the answer is a drawing, suggest they check one before you cut two hundred.
Four: the mount was chosen for sound, not for manufacture
Gasket, leaf spring and integrated plate mounts all get chosen from forum discussion about typing feel. Which is a perfectly good reason, and it has consequences the designer has not costed.
A gasket shelf controls compression, so it needs a real tolerance rather than the block default. Leaf spring arms are thin flexures that will chatter and leave visible marks if the feeds are not planned, and their thickness sets the spring rate so a thickness error is a feel error. An integrated plate puts the switch cutouts and the entire cosmetic exterior on one part, which means a mistake at the last operation scraps everything done before it.
You do not need to talk them out of any of it. You need to price the risk honestly and say why. On an integrated plate in particular, quote the overage properly, because the value at risk in each part is much higher than the geometry suggests.
Five: two alloys, one colourway
A top case in 6061 and a bottom case in 6063, usually because a supplier had the right thickness in one and not the other.
They will not match. Silicon content changes how the oxide layer forms, so the same dye produces visibly different shades, and no anodiser can correct it.
The same problem arrives in a second form: parts that have to match being anodised in different loads. Dye depletes as parts are processed and bath temperature drifts, so two loads of the same colour will not agree closely enough for halves that meet in a visible seam.
Both are free to fix at drawing stage and impossible to fix afterwards. Say it once, early: one alloy for everything that has to match, and all matching parts in the same anodising load. Put the second one on the order confirmation so it cannot be lost.
What to do with the list
Send it back as five short observations, not a lecture. Something like: here is what I would change before we cut anything, here is what each one saves, and here is the quote both ways.
These customers own their design and can act on it immediately. Most industrial customers cannot. That makes a DFM review on a keyboard case unusually productive, and it is the fastest way to turn a first enquiry into a repeat customer in a community where everybody compares notes.
If you would rather the requests arrived already checked, see what we look for in a shop.
Frequently asked questions
Should we do a free DFM review?
On a first enquiry, yes, and keep it to thirty minutes. These customers own their own design and can change it the same evening, so the review directly reduces what you have to make. It is the highest return half hour in the whole job.
What if they will not change the model?
Then quote what is in front of you, with the cost of the problem features visible as a line or a comment. A customer who declines a change having seen its price has made a decision, which is fine. One who was never told has a grievance waiting.
They sent only a STEP file. Is that enough to quote?
No. A model carries geometry but not intent: which faces are cosmetic, which tolerances matter, what the alloy is and what the finish should be. Ask for a drawing, and if they do not have one, ask the five questions in this article instead.
Need a quote for your case?
Send the files. We match you with up to 3 shops that machine keyboard cases and reply within 24 business hours.
Get matched with a keyboard case shop