Project: SSK revival

[CPTBadAss]

MY HOUSE

no seriously, i'm going to print the complex plastic stuff at home and probably do final assembly of everything. i'm currently the only one with the tools to do the assembly of the internals anyway (and even then only with an assist from rknize)

What a baller

acetrak is responsible for this newest iteration of the case btw.

Acebro is the man

[fohat.digs]

That open-air frame in post # 170 is totally awesome.

[dante]

mkawa: is your intention to have this setup as bolt modded from the get go?

[mkawa]

yep

[dante]

MY HOUSE

no seriously, i'm going to print the complex plastic stuff at home and probably do final assembly of everything. i'm currently the only one with the tools to do the assembly of the internals anyway (and even then only with an assist from rknize)

rknize lives relatively close to me [both in Chicagoland] - though we've never met.  I can make myself available to help out with any manual labor if needed.

[mkawa]

do you know how to operate a manual mill without cutting your fingers off?

[rao2100]

How much are these and where do we order. :-P
Can't wait. 

[dante]

do you know how to operate a manual mill without cutting your fingers off?

Have no clue but I am willing to learn.  I put safety and quality of work over speed.

[CPTBadAss]

do you know how to operate a manual mill without cutting your fingers off?

I can operate a mill and not lose fingers. And I can make it to Chicago land if rknize and dante would like some help.

[dante]

do you know how to operate a manual mill without cutting your fingers off?

I can operate a mill and not lose fingers. And I can make it to Chicago land if rknize and dante would like some help.

Even if it turns out I'm not a good fit for mill work I'm open to doing any other tedious thankless tasks for the priviledge of being involved in this project.

[The_Beast]

I wish I had a mill

[sordna]

What kind of mill are you folks talking about? I have a manual mill but it's for grinding coffee and there's no way it could cut your fingers :-)

[Techno Trousers]

Just got pointed over to this thread, and I'll be watching with interest for updates!

[rootwyrm]

What kind of mill are you folks talking about? I have a manual mill but it's for grinding coffee and there's no way it could cut your fingers :-)

Well, for the lower full case mkawa posted... if you're a really good machinist, you can arguably do it with a vertical mill (provided you're also up for some welding.)
But this here is your typical starting point roughly for a CNC vertical mill. Tormach's got reasonable accuracy - if you want to go production grade accuracy, micro-machining or larger table the price goes up very quickly. How much? A Bridgeport Acu-Rite in used shape will cost you $18K+ - and that's on the small side. A beaten up Mori Seiki vertical SV-50 with 31.5" X-axis, basic controls (no Fanuc set), no vises, and 10K RPM 30HP will set you back at least $62,000. (Plus the concrete work to support a 9 ton machine.)

You don't want to know what a 58" traversal three-axis costs. Forget a five-axis Cincinnati.

But if you want to get started with a mill? You can toss 'high accuracy' at the start. But still figure on spending at least $5K - besides the mill, you'll also need a lot of tools and equipment. And that stuff is not cheap.

[mkawa]

the milling they're talking about is really simple, more of a drill press job. the plastic barrel carrying frames come in with plastic rivets sticking out of the back instead of holes for bolts. the challenge is drilling those rivets out quickly and efficiently so you can do batches of like 20 at a time. rknize has a manual mill and a jig that he uses to do it but is short on time (lead time in the half dozens of months ). he currently has about 10 or 20 frames that he's working his way through. people were volunteering to go over to his workshop and drill the rest out with his jig for the greater good

[Moosecraft]

Really interested in this , the regular IBM boards aswell as the SSK ones are so big I would go crazy. This however is a completely different story and makes me wanna go back to BS

[rknize]

the milling they're talking about is really simple, more of a drill press job. the plastic barrel carrying frames come in with plastic rivets sticking out of the back instead of holes for bolts. the challenge is drilling those rivets out quickly and efficiently so you can do batches of like 20 at a time. rknize has a manual mill and a jig that he uses to do it but is short on time (lead time in the half dozens of months ). he currently has about 10 or 20 frames that he's working his way through. people were volunteering to go over to his workshop and drill the rest out with his jig for the greater good

The current batch is done.  I'll send them off to you soon.  Might I remind you that you sort of sprung them on me!   

I started getting cross-eyed after a while, so hopefully most the holes are in the barrel frames and not my hand.  I do seem to have a few extra holes, but they might have come from the weed whacker. 

[mkawa]

i totally did. it was more fu

the milling they're talking about is really simple, more of a drill press job. the plastic barrel carrying frames come in with plastic rivets sticking out of the back instead of holes for bolts. the challenge is drilling those rivets out quickly and efficiently so you can do batches of like 20 at a time. rknize has a manual mill and a jig that he uses to do it but is short on time (lead time in the half dozens of months ). he currently has about 10 or 20 frames that he's working his way through. people were volunteering to go over to his workshop and drill the rest out with his jig for the greater good

The current batch is done.  I'll send them off to you soon.  Might I remind you that you sort of sprung them on me!   

I started getting cross-eyed after a while, so hopefully most the holes are in the barrel frames and not my hand.  I do seem to have a few extra holes, but they might have come from the weed whacker. 

i totally did. it was more fun that way

extra holes are fine, and i can touch up with a dremel and rtv silicone (plus/minus). no worries.

[rootwyrm]

the milling they're talking about is really simple, more of a drill press job. the plastic barrel carrying frames come in with plastic rivets sticking out of the back instead of holes for bolts. the challenge is drilling those rivets out quickly and efficiently so you can do batches of like 20 at a time. rknize has a manual mill and a jig that he uses to do it but is short on time (lead time in the half dozens of months ). he currently has about 10 or 20 frames that he's working his way through. people were volunteering to go over to his workshop and drill the rest out with his jig for the greater good

Hmm... the barrel frames do present a problem and then some. They're poorly suited to my drill press to put it mildly. (Trust me: do NOT attempt bolt modding with a big drill press unless you have a curved shoe AND vises for it. I don't have the shoe.) This is kind of why I'd really love Unicomp to just change to a bolted through die. Alas. Optimally would want to use a CNC drill press to do the job, but given the variances in the plates from die wear, it'd probably ruin more than it actually produced.

BTW, for the bolts, I checked my notes. You want to use Loctite 290 "Green", not 242. "Blue" 242 has mediocre vibration tolerance and "Red" 271 is for absolute permanence. 290 isn't quite as easy to find in stores and yes, is the about same green you see on many bolts in IBM parts. (Also don't confuse with their superglue, which has a green label.)
Loctite 290 Green - $4.99 at JEGS
Great stuff. I recommend keeping a bottle around for projects regardless.

[mkawa]

i don't want loctite at all, as it ruins torque measurements on the bolts. i may or may not have mentioned that earlier. i will be transitioning to either jamb nuts or lock washers or both.

you're still very confused, unfortunately. the barrel plates are molded ABS and pretty precisely uniform, actually. a small 2-axis mill could make short work of the matter, but i don't have the money for that tooling at the moment, as i'm currently debugging our very expensive FFM printer.

[mkawa]

actually i should post this somewhere as i'm not sure i have it all out there anywhere but my head. here's the construction process for an ssk revival keyboard:

INTERNALS ASSEMBLY

1) drill out the plastic rivets from a new barrel plate, as in any other bolt mod (well, with a new barrel plate). currently, this is being done with rknize's manual mill, because he was naive enough to offer a while ago, muahahahaha.

2) modify a new IBM-matrix model M 104 membrane by slicing off the tenkey. bridge the cut traces as appropriate using a combination of small gauge magnet wire, and conductive copper tape. some of the details of this are TBA. i've used a variety of bridging materials and haven't been entirely happy with any of them. these seem to have the most promise out of those that remain. the beta boards actually just folded the excess membrane over, which works fine electrically and mechanically (for the most part... jury's still out on that actually, it may have contributed to some of the problems that were partly due to warpage of the backplates) but is inelegant, and may not work so well with the new case design.

3) sandwich 2 pcs new stamped steel backplate, ABS barrel plate, and M membrane, secure with 2mm-longer-than-normal-bolt mod screws (which iirc are M3, but i have hundreds at home and will correct this when i get there). secure with lock washers, 4mm nuts. torque to 2inlbs. (again, i may have to correct this when i get home).

4) attach to new SSK controller board. wire ground to somewhere sufficiently large and conductive.

CASE ASSEMBLY -- MOSTLY TBA

1) print 2 ABS assembly-side-pieces. the reason why we need these is because we need to want to form-fit the curved assembly at the edges. for support reasons. now that our printing costs are significantly lower, the sketches you saw will probably be widened, actually -- the original design attempted to minimize volume due to printing cost.

2) print a pcb retainer. because we can.

3) fasten both to cut and press bent sheet metal frame that will serve as keyboard outer. we're probably looking at thin stainless, because it's cool, and because the assembly mostly supports itself if you give it a nice but minimal platform.

[wcass]

i'm going to order something from inkjetflex.com - would you be interested in adding to my order? you would save $30 for processing and delivery.

if you are interested, i could convert a flatbed scan or picture of your matrix to proper format. could send you back a PDF that you can print at 100% on 11x17 to double-check.

[mkawa]

i am very interested. what's your timeframe for ordering? i can send you a matrix but you'll have to scan both sides and then we'll have to convert and hack on the photos together.

[wcass]

i can order any time.

i have IBM membranes on hand that might be the same as yours. i'll convert these to CAD file then PDF and JPG. you can print out the PDF at 100% and lay your membrane on top to check that everything lines up.  if anything does not match, mark up the JPG with Snagit, Greenshot, MS Paint, or other graphics mark-up tool and send it back to me.

[bazemk1979]

Guys whats the general idea behind this mod? DIY and then have a GB for some sort of casing or give the SSK mod to Unicomp and hope they take one the bait copy/paste the design and roll out new SSK model production based on this mod?

[mkawa]

i can order any time.

i have IBM membranes on hand that might be the same as yours. i'll convert these to CAD file then PDF and JPG. you can print out the PDF at 100% and lay your membrane on top to check that everything lines up.  if anything does not match, mark up the JPG with Snagit, Greenshot, MS Paint, or other graphics mark-up tool and send it back to me.

sounds good

[mkawa]

Guys whats the general idea behind this mod? DIY and they have a GB for some sort of casing or give the SSK mod to Unicomp and hope they take one the bait copy/paste the design and roll out new SSK model production based on this mod?

i'm going to build small batches and then sell them, with funds going into a geekhack trust fund.

[rootwyrm]

i don't want loctite at all, as it ruins torque measurements on the bolts. i may or may not have mentioned that earlier. i will be transitioning to either jamb nuts or lock washers or both.

I've done scale assembly engineering for sensitive parts with torque-limited drivers - specifically motherboards where overtorque cracked them and backout fried them due to short. Hence why I specified 290 wicking - Loctite does not have an impact on torque unless you're using the wrong compound. Wicking compounds are less susceptible to application error generally. The whole point of Loctite is to ensure torque is maintained and does not change over time. Really should be using Loctite 222MS "Purple" but Grainger's website is being broken AGAIN - prevail on 222MS is 30in-lbs, which is pretty much free-hand tight. 290 will not affect torque, but it's very much "GFL getting it apart ever again." And 222MS requires you apply it correctly. Blue is entirely inappropriate here as it expands quite a bit and isolated to the application area. (Also 222MS used to cost a lot more than 290.)
Loctite is specifically indicated for torque sensitive, tight tolerance parts all the time. Loose fitting OTS M3 fasteners (anything over 0.005 is loose) should be using Loctite. I recommend looking around Practical Machinist sometime - it's a frequent topic of discussion. The TL;DR is that most problems with Loctite and other thread lockers are summarized as 'using the wrong one for the job.'

you're still very confused, unfortunately. the barrel plates are molded ABS and pretty precisely uniform, actually. a small 2-axis mill could make short work of the matter, but i don't have the money for that tooling at the moment, as i'm currently debugging our very expensive FFM printer.

No, I've worked on M's - a lot of M's. I see two problems. One, if you run the RPMs too high or move too quickly you produce heat distortion and damage. Two, by my calipers I'm finding variances of 0.3mm to 0.7mm between two same-year (91) back plates and molds. And no, not Greenock. I'd presume these are somewhere around the same, which obviously is moot if you go oversize by +1mm.
And I've been advised the fastest way would definitely be drill press with tilting table using a horizontal clamp rather than a mill. Don't look at me - this is from someone who's been turning parts for over four decades. Pretty sure someone who can get better than 0.001 on full manual three-axis and do 144 holes in an hour knows what they're talking about. Plus, looking around, looks like it's also a very cheap solution in terms of equipment - can be done under a hundred bucks including the drill press!

[wcass]

for your approval.

print the PDF on 11x17 at 100%, lay your membrane on top and see if i need to adjust anything. if anything does need to be moved, mark up the JPG in paint using red or blue and repost. the extra bit on the top membrane is so that the tail will fit either 8 pin or 12 pin FFC connectors - depending on where you trim it.

i don't like what had to be done on the right side of the bottom membrane. just a thought - you might consider a custom controller; we could change the matrix so the paths are more efficient and add a Winkey option. the controller would also be a drop in programmable replacement for most older M's.

[mkawa]

actually wcass, the routing here is not ideal. i just realized the not-dumbass thing to do is to just send you an ssk membrane to scan (i have an embarrassingly large number of them), most of them corroded to ****, but quite scannable. i have your address. 'twill hit the mail tomorrow-ish

if i can figure out how to massage my macro setup into a copy stand in the next hour two i'll just send you tiffs. otherwise, mail it is!

rootwyrm: if you would like to help, you are free to contact unicomp, order 20 plastic frames, have them drilled out however you'd like, and then send them to me. if you'd like to go one further and construct the entire internal assembly yourself, that would be even better. thanks.

[rao2100]

Really admire you guys for doing this. I am a ready customer, even though it might cost a lot to ship to where I am. (Typed on my SSK )

[mkawa]

completely doable, but yes, it will cost. (hint: i have had packages of 5-6 SSKs sent to me from china before )

[wcass]

added the odd cut-outs at the top of the membranes and part numbers to match those on the stock sheets.

i made the rivet holes round so that they can be easily punched out with a cheap die. should i make them match the shapes of the originals?

[mkawa]

no, round is perfect. hell, i'll probably just shove the screws directly through them after poking them by hand with an awl. the weird shapes were due to tooling that unicomp had long ago but no longer use.

[rknize]

Well...if you look really carefully at a barrel plate, there are oval shoulders at the base of every peg.  I believe this provides a fixed offset between the back plate and the barrel plate.  Doesn't the membrane need to match the shape of these?

[mkawa]

i believe those are horizontal positioners and tehy don't, as far as i know, provide an offset because they go through the ellipses in the membrane. the large sills that run horizontally across the barrel plates provide the fixed offset. the holes just need to be large enough to fit those shoulders (ie, the holes should not be 2mm diameter, they actually need to be 5mm diameter, which i believe wcass put in his pdf, as he just went for the closest circle to the weird ellipses and partial arcs that are on the membrane). that said, i think all of these weird features on the membranes are positioning functionality for the tooling to exploit to lower the x-y tolerance on barrel to membrane, as ideally each hammer should be exactly centered above each pad. that said, the actual membrane tolerances are huge, and i don't have a large dedicated die that can press out these shapes. i can do 5mm circles though. that's pretty easy (and yes, it's me smashing a 5mm awl through. THERE I SAID IT

[fohat.digs]

and yes, it's me smashing a 5mm awl through

Everybody knows that you can't build anything without using a hammer.

[wcass]

i use this to make my holes:
http://www.harborfreight.com/9-piece-hollow-punch-set-3838.html

the holes are just for alignment. if you look at the membranes, you will see that the smallest hole on the membrane (about 2.8mm and round) is just top-left of the quote key - this is the membrane position anchor.  other holes on this row allow some horizontal position play, but little vertical play. holes on other rows need more vertical play due to the curved back plate; the greater the distance to the anchor row - the greater the allowed vertical play.

[mkawa]

i should be able to print out a punch set even for the weird shapes. can you just dot the center of each hole? i can go off a production FCB to get the exact weird hole to punch for each bit

[AKIMbO]

What's going on with this project kawa? 

[mkawa]

it's mostly waiting for me to get back to it. acetrak is finishing up the sheet metal drawings. the break points are there but the various cutouts need to be drawn in still. the pcb mount also needs to be put into the assembly file, and all the plastic pieces need to be revised to be bigger and thicker for easier printing

on the internal assembly side, i have all the parts i need and a new construction rig to try making a folded membrane 1.5-2mm thick backplated assembly that should hopefully not warp or really flex at all. the one issue with this is that the plates need to be bonded together so that they don't just flex out of alignment with each other (since the bolt holes have fairly large tolerances). the boring solution to this is is a little bit of jb weld or other metal adhesive. the fun solution to this is capacitive discharge welding.

the final bit is the membranes. wcass and i want to try this short run FCB vendor, but their order requirements etc. etc. are too much for me to deal with right now. i have a living room full of stuff for the geekhackers store, an active 3d printer that breaks a part every 75 build hours (and it's up to 150-200 or so ) and i have some pretty serious real life stuff that is incredibly urgent.

also i think i have to go to keycon or something.

THANKFULLY I WON'T HAVE TO GIVE A TALK THERE MASHBY

[AKIMbO]

Sweet...thanks for the update.  Hope your irl stuff works out for the best.  I still haven't broken this kawa ssk. 

[mkawa]

you know boost literally broke his quickfire rapid by dropping it on his beta ssk right? :-)

[wcass]

will this work, or do you want a smaller dot?

[mkawa]

those dots are perfecto. only one thing i can see offhand: that dotted line on the FCB connector for the 386 layer, just get rid of everything to the left of the dotted line and make the dotted line a solid line. i'm not entirely sure why that bit is there, honestly.

also, i'm guessing this place is not going to cut the outlines for us, so could you make the FCB outline maybe 3-5 pix wide?

[wcass]

3-5 pixels at what resolution? the border line in the original is 5mil (3 pixels at 600 dpi). the new border is 10 mil (3 pixels at 300 dpi). if you would like that changed to any value, that is a 2 minute fix. just give me a value.

flap is removed in this version, but my understanding was that this project would be using vintage 101key controllers. some of those controllers will have 3 FFC connectors (16-8-4) but some might only have 2 connectors (16-12). the extra bit is "padding" so that the membrane would fit proper in a 12 position FFC or could be cut back to the dotted line to fit an 8 position FFC.

also, 3mil top sheet fits too loose in those FFC connectors (connector spec calls for 5-15mil). older membranes fold a flap to bring the contact area up to 6mil. newer membranes use stickers to bring it to about 9mil.

[mkawa]

3-5 pixels at what resolution? the border line in the original is 5mil (3 pixels at 600 dpi). the new border is 10 mil (3 pixels at 300 dpi). if you would like that changed to any value, that is a 2 minute fix. just give me a value.

flap is removed in this version, but my understanding was that this project would be using vintage 101key controllers. some of those controllers will have 3 FFC connectors (16-8-4) but some might only have 2 connectors (16-12). the extra bit is "padding" so that the membrane would fit proper in a 12 position FFC or could be cut back to the dotted line to fit an 8 position FFC.

also, 3mil top sheet fits too loose in those FFC connectors (connector spec calls for 5-15mil). older membranes fold a flap to bring the contact area up to 6mil. newer membranes use stickers to bring it to about 9mil.

10 mil at 600 dpi would be perfect.

the project is using new old stock 84 key controllers (and old old stock, i literally can't telll the difference and they're all in the same box). all of them have the shorter connectors, so i've been doing a lot of cutting.

hmm.. if we were doing high speed mass production, i would say go with the flap, but my experience has been that even with fcbs at the specced thickness, there's a thin layer of oxidization that requires a bunch of fidgeting with the connectors. i'd rather purpose-build an army of small pieces of spudging out of another material that is stiffer to help induce electrical contact than screw with folding over flaps.

[wcass]

thicker packaging tape or label stock would probably work nicely. place it before cutting the membrane outline.

[mkawa]

exactly

[terran5992]

Any updates on this ? :3