[IC/brainstorming] phantom kits run 2

[mkawa]

i was thinking thicker foam. i've literally used craft foam from michael's (a crafts shop here) to great success

good point on the tapping

[The_Beast]

i was thinking thicker foam. i've literally used craft foam from michael's (a crafts shop here) to great success

good point on the tapping

I wouldn't use that foam is may be conductive. I've only ever seen the pink foam labeled as non-conductive

[mkawa]

it's not so much that it's conductive, but it's not clearly ESD safe. in my pure application it's good enough (doesn't touch any ICs), but you're right, we would need to get ESD safe material for a phantom application if it touched IC traces

[damorgue]

My CPU cooler definitely has a foam sheet attached to the back plate that lies against the mobo, and it is black. I wonder where one could buy that as a non-company. It is probably not sold very often to people.

[jcrouse]

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

I can offer powdercoating services

I'm sure many of you know this but if you are talking about the plate here, which is what was being discussed at this point in the thread, it would be difficult to do the plate. For a good "strong" powdercoat finish I think the recommended minimum radius on edges is .015 (1/64"). I may be smaller but the same issue arises in that it makes the manufacturing costs more. If you laser cut you won't get that on the top and bottom edges of the switch cutouts and you will risk chipping the coating. Also, it is difficult to control the thickness of powdercoating, especially around edges and cutouts. This would be too much of a tolerance variation to have uniformity and consistence for the fit of the switches in the plate. Now that you have read this ... if you were taling about a case, say aluminum, as opposed to the plate, forget what you just read.

John

PS ... How about a black oxide finish, like socket head cap screws ... boost ?

[bpiphany]

Answering things in no specific order.

• Making a few extra layout options don't take any time really. I've gotten this keyboard PCB thing under control by now =)
• I will add PCB mount holes to all switches, at least for the standard layouts. A few in the spacebar row may end up without them, and that is not because of fancy schmancy non-standard layouts.
• And the PCB mount stabilizer holes for the ISO enter really don't work well with some of the holes required for ANSI. So at least some of the stabilizer holes won't be possible to have drilled to size at manufacturing.
• I also think a Filco is a pretty price worthy candidate just for the case.. I find it hard to see how that price will be matched.
• And finally I had an epiphany yesterday.. Doing this whole thing to keep the winkey layout and just to get the programmable controller sounds kind of stupid actually. We should just do our own daughter boards to plug into the Filcos. I don't  think they have changed theirs. So it should be pretty universal. The tenkeyless and fullsize uses different ones though.

[bpiphany]

I will go with the following layout options unless anyone I care enough for shouts loud enough I totally forgot to add a Cherry spacebar row to any of the layouts, but it will of course be available. And as usual, don't get hung up on what the images actually says on the caps... And all non-overlapping options from different layouts will be combinable. But, there will only be mounting plates for the ones the mounting plate person decides. I think I will be able to squeeze in holes for PCB mount stabilizers for all the "regular" layouts as well.

• Filco ANSI
• Cherry ANSI
• Filco ISO
• Cherry ISO
• Filco Japanese
• "Realforce"
• "Symmetric"
• "Arrows"

If anyone knows the positions of the stems on the 4.50u spacebar I would be keen to know them. Or I will just pester Melissa with another question =D The "arrows" layout sports the same 4.50u spacebar.


To be supported layouts
 layouts.pdf (70.16 kB - downloaded 473 times.)

[damorgue]

• Filco ANSI
• Cherry ANSI
• Filco ISO
• Cherry ISO

What is the difference between Filco and Cherry? Winkey?

The cuttable pcb, is that of the table then? I haven't followed this one closely but I know it was discussed at some point.

Edit:
Regarding the spacebar, are you sure it isn't 4U?

4 units (76mm wide, 3 mounts, 28.5mm apart)
One center mount and two extra mounts 1.5 units (28.575mm) left and right of the center.

[bpiphany]

Yeah, with Cherry I mean winkeyless and 7.00u spacebar..

The other project to do a universal-Filco-replacement-board-to-solve-all-layout-options-once-and-for-all, will need to wait a while longer I suspect. I don't really know how interested people are to get one of those... But I think it may be a good middle way to have this board for the larger sizes and a separate one for the small footprint layouts.

This one will of course be possible to split between the navblock and the numpad as well, to fit either a tenkeyless or a fullsize.

[bpiphany]

Edit:
Regarding the spacebar, are you sure it isn't 4U?

4 units (76mm wide, 3 mounts, 28.5mm apart)
One center mount and two extra mounts 1.5 units (28.575mm) left and right of the center.

No, I see 6 x 1.25u + 3 x 1.00u, and that leaves 4.50u for the spacebar. Also SP make 4.50u keys but no 4.00 ones...

[Acanthophis]

Will it be a TKL or a Full?
Rather would have a TKL.

[bpiphany]

Will it be a TKL or a Full?
Rather would have a TKL.

Both.

Silly people not wanting a numpad... =P

[harrison]

Will it be a TKL or a Full?
Rather would have a TKL.

Both.

Silly people not wanting a numpad... =P

having a break-away numpad is an awesome idea.  the silly people are those coming into a topic like this, and not reading the previous posts. 

[mkawa]

i think the idea is not that the numpad is literally removable, but that the same pcb run can accommodate both layouts

ofc this requires designing two versions of the chassis, but if it's a dead simple backplate, that's not such a big challenge..

[harrison]

i think the idea is not that the numpad is literally removable, but that the same pcb run can accommodate both layouts

ofc this requires designing two versions of the chassis, but if it's a dead simple backplate, that's not such a big challenge..

  yeah, i suppose me describing it as a 'break away numpad' is a bit misleading.  either way, it's a great idea.

[bpiphany]

No I am actually thinking of a numpad part on the PCB that you can either leave there to build it into a full size filco (or whatever). Or you break the numpad off, and then it is also possible to mount a microcontroller on the break-away part to create a stand-alone numpad. It all depends on what cost less though. IT may be more cost effective to run two different PCB altogether. That is something we would need to investigate further.

[mkawa]

can you actually run traces across board perforations with any of the fabs? also, i think maintaining the traces so that you both terminate in a uc OR route to the main board uc is going to be too much trouble for too little benefit, especially since it means that you can't do SMT pick-and-place to fab the numpad part of the pcb.

[bpiphany]

It would only be the second controller that can't be mounted, everything else should be ok. I was thinking to place a v-cut at the split line, and bridging it with 0 ohm smd resistors or using simple through hole jumpers.

[mkawa]

i can see where this might break, but i don't know that it will be a problem. basically, if it's a 2-layer board and you need both layers through that junction, things could get ugly quickly. my feeling is though, that unless there's a lot of desire for a standalone numpad, this isn't something we should bother with. if it there is, however, then we might as well just do it right and build both up as one separate-able pcb with a uc placed on both ends and a flex connector that can allow them to be plopped into the same chassis

[bpiphany]

I don't want to say anything definitive. But the numpad is a 4x6 matrix including an extra optional row at the top, and three diodes, plus GND. That would be 14 jumpers or so, which isn't too bad =)

[litster]

I think one PCB design that you can break off parts to make tenkeyless and a tenkey numpad separate is a great idea.  And when it is a full size keyboard, it should just use one controller.

Features like this and features like the current Phantom being able to support multiple layout and configurations is what makes Phantom unique and awesome.  There is no other custom PCB that can do this.

[mkawa]

but but 2 ucs is TWICE THE COMPUTING POWER!!!!

[IvanIvanovich]

Crazy idea... make the numpad totally separate, and have an internal usb connector on the main pcb. This way someone could either use it as a 'fullsize' with it in one case, or as a totally stand alone numpad/macro pad (since it should have it's own controller too) in it's own case solution. It would eliminate any question of routing traces, bridging issue and I don't think a lot of people will complain about having a 1 port usb hub added to main pcb to accommodate that. For those that don't wish to have a numpad, could be nice to be able to plug in the wireless dongle for the mouse or something.

[litster]

bpiphany might complain about having a 1 port USB hub added to the main PCB, because there might not be room for a USB hub chip on the main PCB.  Certainly it would make the PCB design more complicated.  You would also have to figure out where to put the extra USB connector on the main PCB so the numpad can plug into.  And where the extra USB wire between the numpad and the main PCB would go inside the case.

[dorkvader]

And finally I had an epiphany yesterday.. Doing this whole thing to keep the winkey layout and just to get the programmable controller sounds kind of stupid actually. We should just do our own daughter boards to plug into the Filcos. I don't  think they have changed theirs. So it should be pretty universal. The tenkeyless and fullsize uses different ones though.

This is an excellent idea. Given their popularity, I would also recommend seeing if they are compatible with CMstorm (also has a removable controller).

[harrison]

And finally I had an epiphany yesterday.. Doing this whole thing to keep the winkey layout and just to get the programmable controller sounds kind of stupid actually. We should just do our own daughter boards to plug into the Filcos. I don't  think they have changed theirs. So it should be pretty universal. The tenkeyless and fullsize uses different ones though.

This is an excellent idea. Given their popularity, I would also recommend seeing if they are compatible with CMstorm (also has a removable controller).

that is a fantastic idea.  i'm not sure that it's a replacement for the phantom (since it allows for custom layouts, but with a optional replacement plate and a custom controller, that would be a game-changer for the filco for sure.

[bpiphany]

And finally I had an epiphany yesterday.. Doing this whole thing to keep the winkey layout and just to get the programmable controller sounds kind of stupid actually. We should just do our own daughter boards to plug into the Filcos. I don't  think they have changed theirs. So it should be pretty universal. The tenkeyless and fullsize uses different ones though.

This is an excellent idea. Given their popularity, I would also recommend seeing if they are compatible with CMstorm (also has a removable controller).

I've started working on this a little. I reverse engineered the matrices of the Filco boards. Probably not for the first time, but this time I did them the easy way...


Both the fullsize and the tenkeyless uses a 8 x 18 matrix, plus 2 or 3 lines for the LEDs. That is a couple of IO-pins too many to fit the Teensy ATmega32u4. The AT90USB1286 of the Teensy++ on the other hand has plenty, but that chip is a lot larger. A TQFP 32u2 barely fits the daughter board of the tenkeyless. Using a QFN package the 1286 fits as well. But the pads on that thing... =P I drew out a footprint. The pads are 0.25x0.40mm, and under the die. They sort of stick out at the sides and I think people actually solder them by hand every now and then, but that might just be out of my league.


The 32u2 can be used together with an expander, like on the ergodox, of course. Or a simple BCD decoder, but in that case they better not change which pins are the rows...

[Parak]

Both the fullsize and the tenkeyless uses a 8 x 18 matrix, plus 2 or 3 lines for the LEDs. That is a couple of IO-pins too many to fit the Teensy ATmega32u4. The AT90USB1286 of the Teensy++ on the other hand has plenty, but that chip is a lot larger. A TQFP 32u2 barely fits the daughter board of the tenkeyless. Using a QFN package the 1286 fits as well. But the pads on that thing... =P I drew out a footprint. The pads are 0.25x0.40mm, and under the die. They sort of stick out at the sides and I think people actually solder them by hand every now and then, but that might just be out of my league.

I'd think that people would be turned off by soldering even TQFPs (which are not bad, honestly), so one might as well just go for the QFNs with whatever fab doing assembly.

Also, I think the QFN footprint needs thermal vias on the central pad. I did a QFN36 footprint a while back for a USB hub that never happened, and putting vias on it was not entirely pleasant via kicad. Maybe it's actually supported nowadays, though. Now that I think about it, solder paste layer coverage was something that needed oversight as well, as it has to be a percentage of the total area of the central pad.

Take a read through http://blog.screamingcircuits.com/qfn-and-dfn/ (some of the more interesting bits are on second page) when you have a chance, as that's where I got most of the info from.

[bpiphany]

KiCAD can do both percentage or an absolute difference between pads and solder mask (negative as well as positive). It would be nifty to be able to define the solder mask separately though. Since the thermal vias are supposed to be there I would probably include them in the footprint from start. You can name several pads the same, "VIA" for example =) If I were to use a QFN package I would probably take the opportunity to try mounting them with solder paste and hot air... But I think TQFP will be just fine, and I  can solder them myself. I don't think a daughter board would be that very popular...

[TheProfosist]

Just so everyone knows the current phantom fits perfectly into a PLU so its a cheap option for the current phantom.

I like the idea of doing a fullsize board for the second one especially if you can just lop off the numberpad and make it a tenkeyless.

Also please make a 7bit or 7bit esk option as since this keyboard is programmable i would like to be able to layer and the extra keys are needed to have Fn key or 2. Also the extra keys are just usefull especially once we get the programming behind what were doing. So i would like to see 4 additional keys above the numberpad and the numberpad have the option to be all one unit keys, as well as what was done with the phantom (i.e. keys between home cluster ans arrows,condensed function row, condensed number row, and small right and left shift.)

[damorgue]

If the design for the Phantom is copied and extended into a fullsize, all the layouts of the Phantom would remain and still be available. It all just depends on the plates then.

[TheProfosist]

If the design for the Phantom is copied and extended into a fullsize, all the layouts of the Phantom would remain and still be available. It all just depends on the plates then.

I personally liked this part about the Phantom alot and dont see why anyone wouldnt really...

[nullstring]

Sorry if this has been previously discussed.. but..
How feasible would it be to get the phantom to be compatible with both the Filco TKL and the CM quickfire rapid?

Getting it working with the CM quickfire would make sourcing one of these from scratch much cheaper.

[bpiphany]

The quickfire looks like it could be a Filco clone. The important thing is that the distance between the different block of keys is the same. Differing mounting hole locations in the PCB is easier to work around. Do you have any pictures of the inside?

I'm going to drop the compressed function row and the 7bit spacebar row. Extra keys above the arrow block and numpad will still be there though. I'm aiming at having all the switches placed correctly later today.

[alaricljs]

I did a review of this a while ago and don't remember where I found the pics.  The post that on filco is centered under F7/F8 is off to the left and the one under the A is a little different as well.  Other than that they are identical.

[The_Beast]

Still waiting for a plate dwg to send to two machine shops. One has a lazer and one has a water jet

[nullstring]

I actually don't know exactly why the CM quickfire rapid isn't compatible. Just that it isn't.
I will try to do more research and reply to this thread unless you guys beat me to it. (Won't be able to do this until at least tomorrow.)

[damorgue]

Still waiting for a plate dwg to send to two machine shops. One has a lazer and one has a water jet

Waterjets won't cut it (pun)
I have yet to see a waterjet which can pull of these tolerances. Most laser cutters can.

[bpiphany]

Still waiting for a plate dwg to send to two machine shops. One has a lazer and one has a water jet

Waterjets won't cut it (pun)
I have yet to see a waterjet which can pull of these tolerances. Most laser cutters can.

It's not really that sensitive.. The switches and Cherry stabilizers slipped in quite nicely on my plates cut with a 0.8mm water jet. The Costar stabilizers require tighter tolerances though in my experience. There used to be dxf files around. They were lost in the "fire" i presume. These are the old tenkeyless phantom plates.

[litster]

Still waiting for a plate dwg to send to two machine shops. One has a lazer and one has a water jet

Waterjets won't cut it (pun)
I have yet to see a waterjet which can pull of these tolerances. Most laser cutters can.

True that.

The first phantom prototype plate was waterjet cut.  the jet was too big in that all the inside corners were round.  That made it especially hard to install stabs without filing.  The actual shipping Phantom plates are laser cut.  They are much, much better.  No filing necessary.  In addition to that, the thinnest plates we can find is 1/16" thick.  It is just a little too thick.  But BiNiaRiS found a place to sand them down a bit before anodizing, and now they are just perfect.

[jdcarpe]

Are the Phantom plate drawings and PCB drawing copyrighted? I managed to save copies before the Big One of '12, but I am hesitant to repost them without permission.

[alaricljs]

Technically they are, but bpiphany was PrinsValium and he's a nice guy and will undoubtedly reply this afternoon.

[bpiphany]

I don't know much about copyright technicalities =) But I think that stuff you make you kind of automatically achieve some sort of copyright over. The phantom plates are free for anyone to use though, and they are included in my attachment four posts up =D

[jdcarpe]

So they are. When you said "old," I thought you meant there were newer versions which had been lost.

[TheProfosist]

Was there a way to build the phantom so it was my modified 7bit layout but had an actual spacebar in the bottom row? Would this be possible with just a modified plate?

[alaricljs]

Not everyone knows what your layout was.  There were 3 plates produced:  ANSI, ISO and 7bit.  The ANSI and ISO were plain vanilla except for 7bit switch spots in the nav cluster.

[damorgue]

Not everyone knows what your layout was.  There were 3 plates produced:  ANSI, ISO and 7bit.  The ANSI and ISO were plain vanilla except for 7bit switch spots in the nav cluster.

Your forgot they came with/without winkey, no? So 5 in total.

[alaricljs]

oh yeah... 1.25 and 1.5, but that's nothing to do with the 7bit stuff

[TheProfosist]

Not everyone knows what your layout was.  There were 3 plates produced:  ANSI, ISO and 7bit.  The ANSI and ISO were plain vanilla except for 7bit switch spots in the nav cluster.

i had a slightly modified left shift from standard 7bit it was iso instead of what 7bit wanted. My question is could i have had a ansi first row with a standard spacebar if the plate supported it. If so could someone modify the plate file for me so i could order one up?

[alaricljs]

Neat, there's 2 different 7bit layour images here:  http://deskthority.net/marketplace-f11/phantom-custom-keyboard-group-buy-closed-t1411.html   one is iso left shift the other isn't.

According to the plate images, 7bit's is wonky/reverse iso