Fixing Signature Plastics SHIFT key legends digitally

[7bit]

For sure, the alternative would have been to scale at import by 1:1000,
instead they use a scale factor of 13.88:1000, or 1:72.04610951008645533141

I confess I don't understand any of the above statement. The vector paths I gave SP require no scaling by anyone. The only thing I left to their judgment was the width of the cutter, which I told them to match to their existing Gorton Modified legend width. I think leaving as little to chance and (mis-)interpretation as possible is the key here.

This proves that you don't understand mathematics at all!!!!


Your SHIFT legend consists of curves with a start and end points, and these points have a distance from the 0-point and have also distances between each other.

Sure it is possible to scale the vector graphic to any size, but then, my larger graphic is as good as yours, except that Signature Plastics scale factor by pure chance happens to be a totally odd 1:72 .

If they knew how to do it, they would take my information (1:1000) and scale it accordingly to get what they need.

[zslane]

Sure, but it is possible, and I should think preferable, to give them a graphic that is already "actual size" so that you don't have to rely on them scaling it "correctly" at all. There would be less guesswork involved that way. After all, the most reliable way to insure that the milled legend is the right size, is to scale it yourself before you send it to them.

[7bit]

Thanks a lot!


The SVG you sent me really does the trick!


Now, I can prepare all of those odd Space Cadet legends.

[zslane]

Oh my. And here I thought they were already done and ready to go!

Doh!

[Muirium]

Think of 7bit as the opposite of Scotty on Star Trek.

Scotty under promises and over delivers. 7bit, not so much!

[Oobly]

These are now confirmed by SP:

1. Legends are cut in "negative", so you're cutting away the material of the legend itself, not "positive" where you cut away the surrounding area and leave the legend itself standing.
2. There are two ways that SP uses to cut legends.
3. The first is using a simple single-line toolpath with a cutter the width of the desired line. For this you only need the centre line of the legend and the line width.
4. The second is done using a smaller cutter and uses the outside closed path to define areas that are cut away by a complex toolpath created by whatever software you use for generating your G-code or whatever code your CNC machine uses. The cutter size determines the smallest radius possible on the legend.
5. The first method is quicker and easier to do.
6. The second method takes more setup and more time to do and due to using both a smaller cutter and more complex path could result in a broken cutter unless the toolpath is carefully adjusted so it doesn't go over an already cut area.
7. The cost is currently $50 regardless of type.
8. The same legends usable for all row profiles as long as they're the same size. Larger / smaller caps require their own legend cut.
9. The legends are profile specific, they will not fit other profile caps. For instance, SA legends will not fit the curvature of DSA holders / molds.

[zslane]

Missing from the above list is the issue of minimum distance between legend elements, presumably dictated by material flow characteristics as you've mentioned before, Oobly. Any official statement with regard to that? There seems to exist a conflict between SP's guidelines and your own experiences.

Also, by way of clarification, I interpret item 9 to mean that even though the same legend vector file might work unaltered for both DSA and SA caps, the need to cut a separate legend plate for each keycap family means you can't save money ($50) by "reusing" a plate milled for DSA to produce SA keycaps (or vice versa).

BTW, it wasn't that long ago that the price for new legend plates was only $25, yes?

[Oobly]

Since I posted about the Row 4 SHIFTS here first, I will leave this here:

"We are just finishing polishing the 2.25 tooling and should have it ready to
test by the end of the week. The 2.75 is still in 'rough' core block stage
and needs to be sent to our tool maker to burn the cavity. This will
hopefully take about 2 weeks (though I will have a firm date tomorrow when
our engineer talks to the tool maker). When he has finished burning the
cavity, we need another 2 weeks in house to polish and fit the cavity so
that it is ready for production.

Melissa Petersen
Signature Plastics LLC"

I would request this isn't shouted from the rooftops yet, since neither mold is actually ready for production yet, but it looks like it may be only a month away if all goes well. I have requested updates at each stage, so we can track the actual status of these accurately. Once they get closer to production-ready I will post in the News section.

And now for a bit of speculation: from this I reckon they're being made by EDM, which is an expensive and extended process requiring making a number of positive electrode blanks (often in either copper or graphite) and finishing the surfaces manually after the actual EDM process, but results in very tough, durable molds. These should be good when they're done

Anyhoo, carry on. As you were and all that.

[zslane]

I will have the results of the digital SHIFT legend experiment around the same time their R4 molds are finished. If the results are satisfactory, we will need to decide what to do next. I will have some number <= 25 Dolch DSA 2.75u shift keys available for folks who want them. It follows that the 2.25u shift keys probably ought to be made up next to yield matched pairs. And then, of course, there are all the other DSA sets that folks might want corrected legends for...

Next question: when to begin the process of combining the new legend with the new R4 molds and start up a GB for existing non-Honey SA sets (under the assumption that fixed Honey keys could come from Round 5a rather than a separate group buy just for shift keys). Of course, even without the R4 molds, all the other non-standard sized shift keys that can be made from existing R2/R4 molds would benefit from the new legend regardless of the state of SP's new 2.25/2.75 molds, so there's those to consider as well.

Does anyone have any good organizing principle to guide how we proceed forward when November rolls around?

(Edited to remove stupidity on my part.)

[7bit]

You should run a s/(SHIF)([[:space:]]+)(T)/$1$3/g group buy in all except HONEYB colors.

[zslane]

You should run a s/(SHIF)([[:space:]]+)(T)/$1$3/g group buy in all except HONEYB colors.

I just do graphics. I don't run group buys. That's someone else's job.

[Khers]

Best way of getting your will – run a group buy! Maybe time to reconsider?

Totally understand your point of view though. I wouldn't either. My limited experience tells me that GBs can, and will, turn into serious time sinks and I just don't have the time to spare...

[Oobly]

Hi guys!

I'm going to add some more controversy... While going through the various specifications for legend cutting I came across this line:

"On any 2-shot legend, there needs to be a minimum space of 0.020" to 0.035", depending on the
cutter thickness (line thickness) between characters and lines and also on islands - i.e. the center of
the B or P. Note that this does not mean there will be 0.020" of space between the characters when molded. "

For cutting line fonts, you provide the centre line of the characters only. The distance between these lines must be at least 0.020" for the smallest cutter and 0.035" for the largest. The largest line width specified is 0.025", which leaves 0.01" distance between molded characters. The smallest line width specified is 0.012". That leaves 0.008" between molded characters...

The SHIFT legends on SA caps are 0.1" height, which means they're cut with a 0.015" width cutter. Which equates to between 0.008" and 0.009" space between molded characters.

Do with that what you will...

-Oobly

P.S. - I must say, these documents are full of misinterpretations and errors.. such as "0.187" (18.7pt)" I'm pretty sure that's not correct according to any definition of point size (72.272 points per inch or just 72 for some cases).

[7bit]

Huge thanks for this!

You obviously got partly an answer to questions Ive already asked twice during the last few weeks!

One question has been answered direclty to me, yesterday:
The lines can exceed that 0.38"x0.38" frame!


I wondered how the specification does not collide with the fact that some legends (like DOWN) seem to be wider than 0.38".


The answer:
Width for "island" parts, ie sections enclosesd by lines must be within the 0.38"x0.38" frame, but the lines themselves can exceed that limit, by their line width.


This means that using the 0.025" cutter (thanks to Oobly for posting this answer to my question, I've asked SP), we have 0.43"=0.38"+2*0.025" for the whole legend!

[7bit]

I've added kerning to my script. Now, I can typeset any legend, including this one:

SHIFT_black.png (3.5 KiB) Viewed 5459 times

ALT_GR.png (3.97 KiB) Viewed 5458 times

Thanks again, zslane!

[Muirium]

That SHIFT needs serious work. The white space around the I is minuscule, but you could park a truck between F and T.

The long topped F is dumb for doubleshot anyway. Shorten it and improve everything!

[Oobly]

That SHIFT needs serious work. The white space around the I is minuscule, but you could park a truck between F and T.

The long topped F is dumb for doubleshot anyway. Shorten it and improve everything!

Shorten it and it doesn't match all the other font characters... I agree, the kerning needs tweaking to make it pleasing to the eye, but the character sizes should remain as they are. Look at the thread title in different views for an example of different font kerning of SHIFT and notice how it's not much different from the actual molded SHIFT legends. The F and T can be closer together and the H I F further apart, just a little. That will "fix" it just fine. I really think this is being blown way out of proportion, though.

The actual minimum gap between parts of a legend is 0.008", but it depends on the position of the gap and the curvature of the cap at that point. SP tends to quote 0.020" to give them some leeway. This is due to the way they actually cut the legend, since it could result in less than 0.008" gap further away from the surface and this will block the plastic flow. The material needs 0.008" gap to flow. Since that part of the legend is before the upward curvature of the ends of the cap and it's only very slightly upwards of centre I would say it could go close to 0.01" and be acceptable to mold.

Legends can actually extend to 0.48" as long as you don't have islands / counters on the area outside of 0.38"...

The T is too close to the L in the ALT GR. Just look at this text to see how most font handling systems produce a result very similar to the molded one.

TLDR: I don't think SA and DSA profile legends need "fixing". I do think SP needs a second font for SA profile, though, which has better default character spacing and gives an alternative to Gorton modified.

[zslane]

From today's e-mail from Melissa Petersen:

It worked! These will be ready to ship on Monday...

As soon as I receive them, I will post a scan of one.

I think we'll want some 2.25u keys next.

[Muirium]

This does not surprise me. It's not so much they don't have faith in their abilities, as they just plain don't know what works.

[Oobly]

zslane, do you mind posting the file you sent to SP? Does it match any of the OP images? What was the final gap between the F and T?

I'm trying to nail down the real gap limitations and SP are giving me the runaround on it. They state 0.015" now as the absolute minimum, but from what I can tell, that's yet another "safe" value given by the engineers, since we have actual physical caps with smaller gaps than that.

[zslane]

The file I sent to SP had a gap of ~0.007" between the F and the T, and the OP image that corresponds to it is labeled Digital legend with 25% glyph separation (gs = 0.25).

How would I "post" the Illustrator file that I sent to SP?

[Oobly]

I see. Thank you. To be honest, I don't see that one as being considered "best" by general community concensus. More likely the 50% one. Actually even the 75% one looks good to me, but that's just my personal view and I think the 50% one would be most likely to be accepted as "best" (possibly with slightly more separation around the "I" too). I think we should start a poll somewhere on it that has high visibilty so we can get a view of what most people (of those who actually care about the legends) would prefer before moving on to getting them made for the SA Shifts. By the way, the legends will be usable across rows, so we only need to make 1 for each size (1.25, 1.75, 2.25, 2.75) and they'll be usable for both Row 3 and Row 4 caps.

I consider 0.008" to be a good general "absolute minimum" gap in legends where you have divergent geometry, very similar to the 0.007" gap you specified, but let's see if SP allow me to put that in the document.

[Oobly]

Okay... officially, 0.015" is the minimum, with 0.020" being considered "safe". It may be that the actual molded cap has slightly smaller gaps than 0.015" due to distortion and remelting during second shot molding. This seems to be what happened to the SA Nuka Cola legends. The actual SVG has 0.015"-0.020" gaps and the actual cap has 0.009" gaps in those places.

I wonder if SP are using your design or their own for the "updated" Shift legends? I suspect they've increased the separation a bit from your design in order to make it a bit more tolerant.

Please post pics of your DSA caps when they arrive!

[zslane]

Here is a scan of the new 2.75u SHIFT key based on my legend vectors that just arrived from SP:

DSA Dolch 2.75u SHIFT key with custom legend

dolch_custom1_scan.jpg (77.54 KiB) Viewed 5237 times

Comparing with my various measurements, I would say that the gap there is slightly larger than the 0.0075" gap in my "25% gap" vector file, but is less than 0.0135" which was what I was calling a 50% gap. I'd estimate it is somewhere around 0.009" or 0.010". Maybe we call it a 33% gap.

Oh, and none of the 25 keycaps I received have any apparent manufacturing flaws, and they all appear consistent with each other.

[zslane]

And here are some measurements:

DSA-Dolch-275U-Shift-Key-New-Measurements.png (110.86 KiB) Viewed 5219 times

The gap between the F and T is 0.010". The line thickness is 0.014 which is basically 1pt. If I were to do this again, I would increase the line thickness by 15% to 0.016" (or 1.15pt).

The line strokes on these keycaps are a little thinner than the original Dolch legend, but that's because I sent SP a vector file with a 1pt stroke weight on the center lines. I wasn't sure which was more accurate, 1pt or 1.1pt, and the 0.1pt difference seemed to me within the margin of error of my scanning and tracing methods. So I erred on the side of thinner (images of those early attempts at the hashbang somewhat fresh in my mind) and I went with 1pt center lines, but based on these samples, I feel 1.15pt would match the original Gorton Modified font thickness better.

[Oobly]

Thanks for the pics, that looks pretty good, but a little too close to each other for my taste. Great to know they can do it if pushed Interested to see what their own version looks like.

The line thickness should be specified in inches, not points:

0.100" and 0.125" Upper Case Characters: 0.015"
0.100" and 0.125" Lower Case Characters: 0.012"
0.150" and 0.187" Upper and Lower Case Characters: 0.020"
0.230" Upper and Lower Case Characters: 0.025"

This is in my specifications document which is almost ready for posting to the wiki (just adding some legend area specs for the larger keys)

[zslane]

Clearly tastes vary. Muirium probably feels that 0.010" still puts the F and T too far apart. I suspect that 7bit's strategy of shortening the top F bar is the only way to fix this further, but then the shape of the F in SHIFT won't match any of the other Fs (all those function keys, for instance). I guess it is a matter of where you want to compromise.

[Muirium]

I'd go for a whole new set of legends, using a better designed font. Gorton "Modified" has many problems. SHIF T is the most glaring one. But I really dislike the OWERTY effect, too. And all you need to do is look closely at an SA navigation block to get the creeping realisation that the font is far and away the weakest thing about these keys.

But if merely patching, I'd absolutely equalise the F's bars and leave the inconsistency. (In fact, I suggested it.) There are too many problems with this bastardised font to worry about minutiae like that. SHIF T is an abomination.

[zslane]

Looking at the SA nav cluster of my Round 4 SPH set, I see that someone decided to try and select fonts that would fill all the available top surface area of each key. Consequently, legends like INSERT and DELETE use the narrow Gorton Modified font, while HOME uses the extra-wide Gorton Modified font. Giving priority to filled surface area over consistent letter shapes is a dubious choice, IMO. I would have chosen the narrow font--and a single, uniform size--for all modifier keys.

I have no objections to Gorton Modified apart from some of the kerning issues intrinsic to SP's metal blocks. I don't even mind their uppercase Q. I'm really not that much of a typeface snob; just so long as I don't end up with legends like those on the Galaxy Class set. However, what I do like is consistency across a keyboard. A reasonable consistency is quite achievable with Gorton Modified as long as you, as a keyset designer, are willing to select one font for all keys of a given category and allow whatever "extra space" that yields on those keys with fewer characters in their legends.

[Muirium]

For an example of legends that SP tried but couldn't pull off, 7bit just posted a good one:


http://deskthority.net/group-buys-f50/s ... 12047.html

Put that trashcan in the trash!

I actually happen to have seen some of these in person — ssh! can't say any more — and the gap between the lid and the can really is an open gap. That's what SP were trying to warn us against with the FT ligature. While they went ahead with the trash cap because… hmm…