Got some updates from the manufacturer! The prototypes will be shipped out to me tomorrow, and will probably arrive in roughly a week from then.
The manufacturer mentioned there some minor fitting issues on this prototype which aren't critical but are kinda lame, because my tolerances are too tight and because physics (maybe?). I'll still be accepting this first prototype to check out if it's otherwise good, and I'll be making a second prototype which loosens up the tolerances before moving to GB, which should hopefully be perfect.
There were some issues with fitting the parts together after applying the surface finishes, which I was pretty surprised about, since I left about a 0.15mm gap between any parts that needed to fit together, which I thought would be plenty more than enough space given what I looked up about PVD and E-Coat thicknesses. So why is there interference? Here's what I think.
The anodized finished part did not have any issues with fitting, and there were no fitting issues with the parts before surface treatment, so this interference is certainly from the coatings. My first though was that perhaps they weren't actually using an E-Coat, and perhaps a thicker powder coating (which I thought still would fit). However, I gave them a call with a translator who said that the word they used for electrophoresis in Chinese was distinctly that, and wouldn't have been simply confused with something else.
The interference is pretty minor, as they said the parts can still be assembled together, but it might result in some scratching (only on interior parts anyway) and it might be hard to disassemble. They offered to polish down the PVD parts to make it fit better, but it might damage the finish - I agreed to have one of them polished down as a sacrifice, and have the others shipped as normal (the anodized had no problems anyway).
They ended up polishing it and actually didn't have any issues with the finish. This next Imgur album shows the parts that they polished (this is after it has been buffed down). Interestingly, the areas they marked that had interference were all at the corners of the plate, and also they verbally mentioned that a few of the slots where the rear weight fins fit into were also areas of interference. I double checked my design, and the gap at these areas were uniform with the other gaps (0.15mm) - so if the design is sound, does that mean that the coating tends to be thicker in these areas?
https://imgur.com/a/IUDBCi5I think that is actually the case, and kind of makes sense.
Sputter deposition is a common method of applying PVD, which as far as I understand involves electrically charging the object being coated. And as we know, E-Coat also electrically charges the target object for deposition. My theory is that these methods which involves electrically charging the keyboard might be the culprit.
As we have observed in other parts of the world, we know that
electric charges tend to concentrate around sharp edges. Similar to a lightning rod, sharp edges on the keyboard may hold a higher charge than other parts, and of course there can be some additional irregularities due to factors such as where the electrodes were attached, etc. My theory is that this results in these types of coatings being thicker at the corners, which is why there is interference at the intersection of corners.
If anyone who has experience with this has any ideas on if this seems legit, please let me know what you think.
In any case, the bottom line is that my tolerances are too tight. I will still be accepting this round of prototypes to see if the quality and finish looks good, and also to see if the keyboard generally feels good (weight distribution, etc.). Of course, I don't want any surprises or potential fitting issues with the production run - so I'm planning on ordering another prototype which will have revised tolerances, and perhaps design changes if the first prototypes reveal the need to make any. This probably won't take as long to produce as the first prototype run, since the manufacturer now has the workholdings needed to make the keyboard, and the surface treatment figured out.