[IC/brainstorming] phantom kits run 2

[fartq]

Folks,

I don't want to say anything concrete without lit and bpiphany weighing in, but if there is a second run of phantom parts, i think it would be productive to weigh in on what changes should be made to the kits to make them better/easier to construct, etc. here are my ideas:

1) every kit should come with a case. it has been far too hard to source cases for the design, despite filco compatibility. first off, filco boards are not cheap, and there's no inexpensive secondary source of filco cases, since they're not cheap. my idea for this was to simply have thick aluminum or steel rectangular bottom plates cut, finished, drilled and tapped for standard pcb spacers (ie, motherboard spacers). this should be extremely inexpensive, and is a perfectly adequate case with the addition of either small printed or milled feet.

2) the board design should be revised to SMT so that it can be stuffed with diodes, LEDs, controller, etc. everything but switches at the factory. this will make sourcing the kit significantly easier, not to mention construction.

3) tolerances on the plates has been an issue. we need to work directly with a competent shop this time that can give us high precision lasercutting and good guarantees on uniformity of thickness and flatness of each plate. treble is usually my go-to guy for machining and materials sourcing, but his time has been an issue, so we might have to find another shop (possibly remote) that can give us this. again, finishing is an issue as well. in the previous run, finishing ended up being the key to getting thickness uniformity, so that may have to happen again.

4) switch sourcing: we won't be ordering from germany again. fortunately, demik has found a reliable domestic source of switches. unfortunately, due to the current switch shortage, they're all pcb mount, so the pcb will need to be revised if possible to accept such parts, and we may not get the most amazing choice of switches. however, cost and lead times should be much better this time (switch sourcing was by far the long pole in the previous run).

[boost]

Acrylic cases would be cheaper than the aluminum...

[fartq]

the lasercutting costs significantly more than the materials, but it would eliminate the finishing issues. something to try.

[boost]

the lasercutting costs significantly more than the materials, but it would eliminate the finishing issues. something to try.

I can offer powdercoating services

[fartq]

ooo, good to know

[boost]

ooo, good to know

Yep, got a ****ton of colors to chose from and very competitive on the prices

[Acanthophis]

Sounds interesting.
Not for my wallet, though :/

[The_Beast]

I know a guy with a laser cutter that can do 1/2" plate aluminum

[mkawa]

what are his tolerances though? and pricing?

[The_Beast]

what are his tolerances though? and pricing?

I'm pretty sure 1/32+/-  &  idk depends on material+how complicated it is.

[mkawa]

+/- 1mm is too much

[jcrouse]

+ or - 1mm would be total of .078 in English units. The will need to be + or - .002 or .003 max. The specs for the plate are on cherry's website. I do remember that the corners of the cutout square called for a maximum radius of .005, which is pretty difficult to obtain. An undercut could be required.

[The_Beast]

I looked at some of the parts we've sent him in the past. One was +0, -1/32, would that be acceptable?

[jcrouse]

See above.

[The_Beast]

See above.

I was talking about using this guy for the case, not the plate ^__^


Also, I've never asked specifically about tolerances. +/- 1/32 is the tightest tolerance I've seen sent to him, me might be able to get that even lower.

[jcrouse]

Sorry bout that. On that note, what technology was used to cut the plates, laser, waterjet, edm or something else?

Thanks,
John

[mkawa]

oh! for the case i was thinking that all we would want to do is drill and tap

i guess we could round the corners, but it's literally just a metal plate with some standoffs on it

[alaricljs]

Mmmm, bare bones

[The_Beast]

I'll email him about his tolerances, how thick does the plate need to be?

[mkawa]

i'd say around 6mm so that it doesn't dent and still has a heft to it? beyond that the sky is really the limit.

[The_Beast]

So it could be 1/4" thick (I would think it would be easier to get than 6mm)? Also, do you have a dwg file with locations, dimensions, tolerances...... of the plate exactly how you want it(imperial measurements preferred)?


I could get a quote on it

[harrison]

i was somewhat anxious to see this thread, as always, the buyer's regret is that ver2 is always better than the first.  fortunately, for me anyhow, building is part of the fun so through hole vs SMT isn't a factor.  That said, I'd like to know there's going to be a good opportunity cost made available to those that participated in the first round for any cases or plates.

[Djuzuh]

http://www.vendio.com/stores/E-sports-Gaming-equipments/item/keyboard-accessory/colorful-replace-case-filco-an/lid=31051008

did anybody try this item for the phantom?

[mkawa]

So it could be 1/4" thick (I would think it would be easier to get than 6mm)? Also, do you have a dwg file with locations, dimensions, tolerances...... of the plate exactly how you want it(imperial measurements preferred)?


I could get a quote on it

literally, rectangular bulk material t6061 or some other alloy with low internal defects at the dimensions of the phantom pcb (don't remember offhand), attachment mechanism for 4-5 2-3mm x 1-2mm standoffs (either drilled and tapped holes or something permanent).

basically, cheap bulk material. bulk of cost will probably be finishing

[harrison]

also, as an approximation when talking to vendors about cutting, what's the typical square footage (or linear inch) of cutting that needs to be budgeted for a 3 layer (bottom, 1/2" middle, and 1/4 or 1/8" top) laminated acrylic case?

[DanGWanG]

It would be nice to have an all-in-one purchase cost.  Similar to how KBD/OTD runs groupbuys, all you typically need is switches and some soldering skill/tools.

[bpiphany]

I am totally up for designing a new board =D

• I think we should do a full size board, with the possibility to cut it down to tenkeyless size. To have at least something new.
• SMD is the true path.. (holes for through hole diodes could still  be included).
• Building the controller from scratch directly on the main PCB.
• Ditching the Teensy and all little diode legs in the process =D
• PCB mount holes may be possible, at least on most switch locations.
• PCB mount stabilizer holes will probably be possible to mark by 1mm pilot holes, to make custom case builders happy =) They overlap too much to have all options drilled to size from start. The desired holes will need to be drilled, but this is easy done with the pilot holes.
• PLU cases? I still haven't heard if they are compatible.
• Leopold cases? They are not compatible with the phantom due to different mounting hole locations. But those hole could possibly be included in a new design. If the rest of the layout is equal. Same distances between different block of keys. (Having the controller SMD will help making room for the extra mounting holes).
• I don't know even know where to start looking for building services. Or what more steps in the design phase that need to be taken...

The ErgoDox design came and took all my time for a while, but before that I started a thread on DT asking for requests for features to include on a new design. Mainly spacebar row setups, since that can be done in a thousand different ways... I don't think I will go completely crazy though. The standard ones like

• Filco and others: 1.25, 1.25, 1.25, 6.25, 1.25, 1.25, 1.25, 1.25
• Cherry:           1.50, 1.00, 1.50, 7.00, 1.50, 1.00, 1.50
• Realforce:        1.50, 1.00, 1.50, 6.00, 1.50, 1.00, 1.00, 1.50

will be included, and a few extras of the best suggestions (that I find reasonable).

[Djuzuh]

where does one find a cherry MX 6 spacebar?

[bpiphany]

where does one find a cherry MX 6 spacebar?

I, for one,  have some...

[WRXChris]

I'm definitely very excited to see a streamlined round 2 in the works!  Controller built into the board sounds awesome, I like where your head is at Prins!

Sounds like the most pressing matter is figuring out the case situation.  Swede is back at work on his aluminum filco/phantom case at DT, so hopefully it will come to fruition sooner than later.  Treble has expressed interest in making an alu TKL case, but we all know he is very busy.  And the fact is both of these options will likely be in the $300 range for the case alone, which would limit this project to very few gh'ers. 

What about designing a case and having it manufactured like Dox did?  Another decent option would be to just buy a filco and use it's case and switches (at least for people building with mainstream switches), but the required desoldering would probably scare as many people as the pricetag on an alu case..

[mkawa]

I am totally up for designing a new board =D

• I think we should do a full size board, with the possibility to cut it down to tenkeyless size. To have at least something new.

sure, a perforation through the board at the tenkey pad and a flex connector linking the tenkey and the tkl part shouldn't add too much cost.

SMD is the true path.. (holes for through hole diodes could still  be included).

no, pick and place construction will make the through-hole stuff a waste of space and time

• Building the controller from scratch directly on the main PCB.
• Ditching the Teensy and all little diode legs in the process =D
• PCB mount holes may be possible, at least on most switch locations.
• PCB mount stabilizer holes will probably be possible to mark by 1mm pilot holes, to make custom case builders happy =) They overlap too much to have all options drilled to size from start. The desired holes will need to be drilled, but this is easy done with the pilot holes.
• PLU cases? I still haven't heard if they are compatible.
• Leopold cases? They are not compatible with the phantom due to different mounting hole locations. But those hole could possibly be included in a new design. If the rest of the layout is equal. Same distances between different block of keys. (Having the controller SMD will help making room for the extra mounting holes).
• I don't know even know where to start looking for building services. Or what more steps in the design phase that need to be taken...

someone pointed out an HK supplier that advertises via ebay recently. our chinese contacts will be able to get this done too.

The ErgoDox design came and took all my time for a while, but before that I started a thread on DT asking for requests for features to include on a new design. Mainly spacebar row setups, since that can be done in a thousand different ways... I don't think I will go completely crazy though. The standard ones like

• Filco and others: 1.25, 1.25, 1.25, 6.25, 1.25, 1.25, 1.25, 1.25
• Cherry:           1.50, 1.00, 1.50, 7.00, 1.50, 1.00, 1.50
• Realforce:        1.50, 1.00, 1.50, 6.00, 1.50, 1.00, 1.00, 1.50

will be included, and a few extras of the best suggestions (that I find reasonable).

sure, the option can be there, but we're only going to press out two types of plate en masse: winkey and winkeyless. litster and bini had a heck of a time with plate production last time, so planning ahead of time for this is going to be key.

[damorgue]

sure, the option can be there, but we're only going to press out two types of plate en masse: winkey and winkeyless. litster and bini had a heck of a time with plate production last time, so planning ahead of time for this is going to be key.

ISO and ANSI?

[mkawa]

oh right, i forgot about you guys. sure.

none of this 7bit LAYOUT nonsense though.

basically, layout choices should be:

ISOWKLESS
ISOWK
ANSIWK
ANSIWKLESS

that only requires us to source 4 plates (if we even source plates). if you want your own crazy layout, source your own plate, but you will have to buy one of those four plates form us with your kit...

[bpiphany]

I will just pack as many options as possible onto the PCB, there is no extra cost in that =) If there is room for through hole diodes I will add that option as well. As long as that doesn't push against any limit in the number of holes that won't add any cost either (pcbwing do 1000 holes for "free"). I think it will be possible to connect the parts with traces even doing perforations.

Plate choices and manufacturing will be up to someone else to decide on.. Cramming more options onto the PCB facilitates for home tinkerers prepared to do their own case/plate/PCB mount.

[mkawa]

yep. afaic the major revision that needs to be made to the pcb is the ability to pick and place most if not all but switch components. also, keep in mind that we will probably not go with pcbwing again, as they won't do pick-and-place afaik.

manufacturing and parts sourcing concerns are otherwise most paramount for this second run, as the goal is to minimize sourcing and manufacturing time.

[bpiphany]

I think that usually you need to supply reels to the pick and placers. I don't know if the controllers come on any reels or such or if they need to be placed by hand or some other means. Also some of the support components for the controller won't reach  very high numbers, and buying full reels may be somewhat pricey. Diodes on the other hand would be both cheap and nice to have pre-assembled... I don't know what costs we are talking about when outsourcing the component placing. We are not going to reach any tremendous amount of units =P And really it is not that hard soldering the SMD diodes. Controller and supporting components may be a bit tighter.

But if someone with experience in pick-and-place can help out with all such things, that would be great.

[mkawa]

But if someone with experience in pick-and-place can help out with all such things, that would be great.

yep, this would be ideal

[mkawa]

aha, found the link again

http://www.sitopway.com/en/index.php

they have an ebay store too

[agor]

Although many prefer cheap cases, I think there should be some kind of aluminum option like the kbd/otd keyboards. Really diggin the look :/

[ekw808]

I don't mind having a baller out option either =)

[IvanIvanovich]

What I would do:
Make pcb mount switch so plate is not necessary at all, option. No need to spend all sorts of time for a bunch of crazy esoteric layouts that very few people might use, just make the same pcb accommodate ansi/iso and winkeyless and winkey version with both step and center capslock option. Have the pcb factory assemble everything but the switches, including a controller. Controller should be programmable still though if at all possible, else just use something standard like a holtek or whatever.
For case, make one design, maybe like 'The Cheat' design and have it made in both acrylic and metal option.
Just keep it simple, and base kit (say pcb and acrylic case only) as cheap as possible of course.

[mkawa]

you know this pcb-mount switch option is not a bad idea, especially backed by a solid piece of metal

[IvanIvanovich]

Exactly, and say you go with a 'The Cheat' style case option, where the bottom piece of the case is probably like 4mm thick, I think you won't experience flex and you still don't need some plate, even if it's plexi.

[damorgue]

With enough screws to attach the PCB to the case between every second switch or so, it would probably be quite sturdy.

[agor]

Probably wouldn't bottom out as loud as plate mounted either. But I sure like the look of a plate beneath the keys :-D

[mkawa]

and it's much much cheaper and easier to machine a solid backing plate than a cherry style switch support plate

[damorgue]

and it's much much cheaper and easier to machine a solid backing plate than a cherry style switch support plate

Milling tends to be more expensive than water or lasercutting by my experience though.

I wonder if one could make a simple watercut plate with holes, then bolt it to the pcb at ~50 places with washers in between as to not short out the solder joints from the switches to the plate. All those scews would only have to be done once. The ease of switching stems would last forever.

[mkawa]

it can be a solid plate that's drilled and tapped. it doesn't have to be milled. the only thing you need is a dielectric pad between the plate and the board. i have a pure which operates on this principle, and is rock solid.

oh, i just read the second sentence of your post. lol. yah, that's the same idea. the only issue is that if we have to mount ICs on the bottom we might need something like 2-3mm thick washers, which would not be as solid as a dieletric pad

[damorgue]

Since tapping is manual labour, it tends to cost a lot, and be priced per hole. I would suggest screws and nuts since nuts are cheaper.

The insulating pad sounds interesting. I have never used that in something I have made myself. Were you thinking of that thicker foam type insulation or more like a thin film of some sorts? I suppose it could even dampen the cavity resonance as well if the foam type were to be used.

[alaricljs]

60% KBs is difficult to design 100% top-side devices.  Anything bigger should be cake.

Don't tap, drill and use PEM nuts or nutserts or whatever.