Result. I drew the Raccoon Soaker's solenoid valve four times, and each time the model had something wrong. What finally fixed it wasn't better measuring. It was a measuring sheet: a draw.io drawing of the model with every dimension a lettered line. I typed my numbers over the question marks, dragged the shapes to match the real thing, and the model read the numbers back.
Four wrong valves
The enclosure needs the valve's shape to hold it, so the model needs that shape. I gave it numbers, and each time something was still guessed:
- "73.66 long, 34.75 wide, 56.67 tall." My three measurements were right, but they don't say what shape the thing is. The model filled in the rest: a square block in the middle and a coil on top, centred.
- "Spigot 20.5 long, Ø25.49." I meant the threaded ends. The model briefly took it for a stub under the body. The middle came out 32.66 long once that was cleared up.
- Four mounting bosses. I found four screw bosses on the valve, perfect for mounting it, and measured them 24.64 apart along the flow and 23.91 across. The model put them on the side. They're on the bottom.
- The coil. The model guessed it 17 mm thick. It's 30.42.
None of these was a bad measurement. Each was a measurement of the wrong thing, or of the right thing that the model then put somewhere I didn't mean. A list of names ("wide", "tall", "spigot") can't carry where a dimension starts and ends.
The measuring sheet
So I asked for the model as a drawing I could mark up. The first version labelled each dimension in words, with a guess beside it: "K = thread centre line → bosses' tips ? (guess 17.38)".
Two problems. It still had the bosses on the side (the drawing made that obvious in a way the 3D view hadn't). And the labels were hard to read: it's tough to guess what a sentence means when it sits on a line between two shapes. So the second round went letters only:
- every measurement is one dimension line with one letter:
K = ? - a key on the side says what each letter is
- the art is to scale (5 px to a mm), so if a shape's wrong, I drag it to match
- three views: from the bottom, from the side, from the outlet end
I filled it in and also added two letters it didn't have: Y, a terminal block on the coil's face I hadn't known about (5.61 deep), and Z. The terminals turned out to come off that block, not straight off the coil.
What the numbers changed
| Guessed | Measured | |
|---|---|---|
| Bosses: where | on the side | on the bottom |
| Bosses' tips below the thread's centre line (K) | 17.4 | 13.53 |
| Coil thickness (L) | 17 | 30.42 |
| Coil, along the flow (I) | 27.5 | 26.33 |
| Middle section across (W) | 30 | 34.40 |
| Terminals | 12 straight off the coil | a 5.61 block, then 10.37 tabs (P), 14.41 apart (R) |
Because the bosses are on the bottom, the valve now lies on its side under the box, screwed to an L-bracket, with the coil pointing sideways. The bigger coil made it hang lower, so the tilt pivot had to rise from 136 to 144 mm. And with the real coil in the model, a new check found it sitting right where one of the cable glands hangs under the box. That gland moved.
Still open
Two numbers don't agree yet: the overall height I measured (56.65) and the sum of the two halves on the sheet (K + J = 60.01). The model uses the halves for now. That's the thing about a sheet: the disagreement is right there in black and orange, not buried in a chat.
What I learned
- Measure onto a drawing, not into a list. A number needs to know where it starts and ends.
- One letter per dimension. Sentences on drawings get misread. Letters don't.
- Let the drawing be wrong on purpose. Draw what the model believes, so the person holding the part can see the mistake and drag it.
- Measure the mounting points first. The four bosses decided everything else about how the valve mounts.
Comments (0)