Unicomp Customizer 101 Bill of Materials

[msiegel]

I'm interested in designing an updated, reduced-cost version of the IBM Model F.

To do that, it would help to know the current manufacturing costs of a similar keyboard.

Here's the inventory of parts I have so far, for Unicomp Customizer 101:

case top
case bottom
2 feet
stamped steel backplate
plastic key-mounting plate
key matrix membranes (2 printed, 1 unprinted, 1 stamp-cut)
rubber sheet
101 springs
101 hammers
101 keys -- $20
controller board assembly (includes leds and connectors)
cable
fasteners

See something missing, or care to take a crack at an item's cost?

[InSanCen]

I'm interested in designing an updated, reduced-cost version of the IBM Model F.

See something missing, or care to take a crack at an item's cost?

...

What quantity are you wanting to produce in? The will have a very direct effect on unit cost for each part. I imagine something like the backplate will reduce a little in quantity, but the membranes would reduce a lot, because of material cost alone.

[lowpoly]

It will also depend on whether you want to have that done locally ($$$) or in the Far East.

[JBert]

key matrix membranes (2 printed, 1 unprinted, 1 stamp-cut)

Ehr, what do you want to do with a membrane?

Model F's use a bent PCB with capacitive pads on it. Using a membrane could mean that you'll have to use the model M's hammers and springs, meaning you would still have a keyboard with the same force graphs.

Or is there something else about the F that you want to reproduce?

[quadibloc]

Ehr, what do you want to do with a membrane?

The membrane was simply listed because it is one of the parts found in a Unicomp Customizer 101, which is like the model M. After all, if this is being used as a basis to determine costs for making a model F, one needs to know what to subtract from its price as well as what to add.

[talis]

Don't forget the $50,000 in tooling costs.

Depends on where you have the tooling done.  NA, the per-tool cost would be about $40,000, with a separate tool for the upper and lower housing, and a third for the key tray.

In Asia, you can get the tooling done for around $15,000-$18,000 a tool.  If you looked hard enough, you may be able to get that cost down to $8,000-$10,000 per, if you can meet the MOQ, and find a manufacturer willing to eat a portion of the cost.

[msiegel]

excellent-- reducing and eliminating tooling costs should be a high priority.

maybe an existing, current case design can be used, to cut that out of the equation.

i see why all the costar boards are packaged alike

[timw4mail]

Hey, I'm fine with the Customizer housing, if its affordable.

[msiegel]

are tooling costs proportional to the size or complexity of the parts?

[talis]

are tooling costs proportional to the size or complexity of the parts?

Both, a small complex part with multiple draw directions would cost nearly as much (or more) then a simple, large case size tool.  Usually for smaller, fairly simple parts they will do a multi-cavity tool (eg rather then injecting a single key cap, they will duplicate and inject 6-12 in a single shot) so even small parts tend to have reasonable sized tools.

[msiegel]

let's just see how close to zero we can get those tooling costs

also, why don't we call it the Model G


ps - where are all those model m2 teardown pics; hmm...

[wellington1869]

also, why don't we call it the Model G

I like it

[quadibloc]

In Asia, you can get the tooling done for around $15,000-$18,000 a tool.  If you looked hard enough, you may be able to get that cost down to $8,000-$10,000 per, if you can meet the MOQ, and find a manufacturer willing to eat a portion of the cost.

It would still be a $50,000 tooling cost if there were three tools. And since it would be hard to show a market for this kind of specialized item, I'd expect the latter to be difficult or impossible.

[ch_123]

If we're going to resurrect an over-complicated keyboard design, let's go all the way!

[lowpoly]

It would still be a $50,000 tooling cost if there were three tools. And since it would be hard to show a market for this kind of specialized item, I'd expect the latter to be difficult or impossible.

Plus the hammers unless Model M hammers will do.

Maybe just design a self-contained switch that uses the drill pattern of existing mechanical switches?

Similar to the Brother-made buckling spring modules.

[msiegel]

the Boscom TestOmatic does sound different (not feel different) as a result of the hammer hitting a hard surface.

that's cool


hmm, it sounds like the minimum new parts required for a Model G are:

springs
hammers
top plate / switch barrel modules
capacitive pads assembly (whether pcb or capacitive membranes)
controller card

unicomp's case, keys, and backplate sound usable as-is.

is there any existing board with a capacitive pcb, that could possibly serve as a foundation?

[lowpoly]

that's cool


hmm, it sounds like the minimum new parts required for a Model G are:

springs
hammers
top plate / switch barrel modules
capacitive pads assembly (whether pcb or capacitive membranes)
controller card

unicomp's case, keys, and backplate sound usable as-is.

is there any existing board with a capacitive pcb, that could possibly serve as a foundation?

Why not use the top plate with the barrels as well? Just add a pcb and and exchange the springs (I'm no expert on Model F/M differences).

[msiegel]

I think this is going to be the big problem.  Easy enough to pull apart a warehouse of old terminal Model Fs for all those other parts but the capacitance circuit has to be specifically tuned for the carbon content of the Model F hammer.

i'm treating the pcb as a "probe", and giving responsibility for the measurement of capacitance to the controller card...

so matching components (of an oscillator circuit, i'm assuming) can be put there

[msiegel]

Why not use the top plate with the barrels as well? Just add a pcb and and exchange the springs (I'm no expert on Model F/M differences).

i'm pretty sure the hammers don't fit -- does anyone have good macro shots of these parts?

[ch_123]

I was at a Computer Museum (euphemistically - it really was a warehouse of old computing junk) the other day and tried a bunch of terminal keyboards.  I wasn't impressed with all these leaf switch variations.  

Didn't see that one though.

Yup, that's because 99% of those old terminals use linear switches. Not really a valid comparison by any criteria. The beam spring switch doesn't really have anything in common with a conventional mechanical leaf switch.

Plus the hammers unless Model M hammers will do.

They wont - the design is quite different.

Maybe just design a self-contained switch that uses the drill pattern of existing mechanical switches?

Similar to the Brother-made buckling spring modules.

The A01 switches aren't really self contained in that the membrane is common to all of them, which, considering that the membrane is one of the most likely bits to fail, makes the whole point of the detachable BS rather unclear.

[msiegel]

Easy enough to pull apart a warehouse of old terminal Model Fs for all those other parts but the capacitance circuit has to be specifically tuned for the carbon content of the Model F hammer.

wait... do those terminal model fs all use individual modules for the barrels?

[ch_123]

In any Model F, the barrels are easily detachable once you open the plate up.

[msiegel]

In any Model F, the barrels are easily detachable once you open the plate up.

outstanding

this means all we need is:

• a boatload of terminal model fs
• non-functioning unicomps or model ms
• new top plates (maybe drilled plastic)
• the open source controller, with an oscillator module
• capacitive pads assembly (pcb or membranes)

and a upgrade of that awful foam! :D

am i forgetting anything? edit: yes, the stabilizer mounts... but we'll get to that ;)

[ch_123]

What are the old Model Ms for?

The problem with old Model Fs would be potential spring-wear. You'd probably end up replacing all the spring anyway, and I'm not sure what the advantage of finding old models to recycle is compared with making new barrels and hammers...

Again, Beam Spring switch much?

[ch_123]

Good thing we aren't trying to make an Alps board here...

Although vintage yellow is quite nice. Particularly on old DEC terminals.

[msiegel]

What are the old Model Ms for?

they have all the right keys, a ps/2 cable, and a case and backplate that support a modern layout.

The problem with old Model Fs would be potential spring-wear.

good point. what do new custom springs cost these days, in lots of 2500?

I'm not sure what the advantage of finding old models to recycle is compared with making new barrels and hammers...

there's no tooling cost. but yes, maybe barrels and hammers wouldn't be as expensive as a case would, etc

[JBert]

I think this is going to be the big problem.  Easy enough to pull apart a warehouse of old terminal Model Fs for all those other parts but the capacitance circuit has to be specifically tuned for the carbon content of the Model F hammer.

I thought one of IBM's expired patents described a calibration procedure which would be run every time the keyboard is turned on.

I would think that is the ideal way as it also solves any problems that may arise from corrosion of the copper padded PCB.

[msiegel]

I thought one of IBM's expired patents described a calibration procedure which would be run every time the keyboard is turned on.

I would think that is the ideal way as it also solves any problems that may arise from corrosion of the copper padded PCB.

with a custom controller, this should be possible

[msiegel]

i wish there were a way to use the already-printed model m membranes as capacitive sensors.

wouldn't a model f hammer change the capacitance of a couple of conductive pads, if it were nearby?

[ch_123]

Has anyone pointed out the fact that the feel of the Model F has nothing to do with the capacitive sensing? Although there's a theory that suggests it might be in a roundabout way, really what is needed is better springs.

[msiegel]

Has anyone pointed out the fact that the feel has nothing to do with the capacitive sensing? Although there's a theory that suggests it might be in a roundabout way, really what is needed is better springs.

it sounded like the springs and hammers are the main source of the model f's better feel...

but those springs and hammers aren't designed to activate a membrane, so if they're going to be used, they have to be used with non-membrane switches :-/

[ch_123]

Contrary to what some people would have you believe, the hammer bit doesn't have any serious part in the what it sounds like - it's really a combination of the spring buckling and unbuckling, and reverberation inside the plate. And has NOTHING to do with what it feels like.

There is some speculation over whether the spring was readjusted on the Model M to ensure that the smaller hammer could actuate the membrane, but nothing really conclusive has come out of this. I'm sure that a more substantial retooling and retweaking could come up with a Model M with a more satisfactory feel, but again, that's speculation.

[msiegel]

has anyone swapped model f springs into a model m?

[ch_123]

Ripster did, it apparently failed spectacularly.

[JBert]

Has anyone pointed out the fact that the feel of the Model F has nothing to do with the capacitive sensing? Although there's a theory that suggests it might be in a roundabout way, really what is needed is better springs.

Here I come with my hammers again: because the hammer needs to be changed to be able to press a membrane, the spring will need to be changed as well to provide the correct amount of force.

If you use capacitive sensing, you don't need to use that much force; all you need to do is move the hammer close to the sensing pads or lift it. So to use a lighter spring, just make your hammer from a lighter material. As long as the spring can tilt it, it should be fine.

Contrary to what some people would have you believe, the hammer bit doesn't have any serious part in the what it sounds like - it's really a combination of the spring buckling and unbuckling, and reverberation inside the plate. And has NOTHING to do with what it feels like.

The hammer just plays a minor part in the sound because it bounces, but you are right that sound is mainly caused by the spring. I think it is different between M and F though.

[msiegel]

ok, at least we know one thing that *won't* work (thanks ripster

back to capacitive senors...

Q: can model m membranes be turned into capacitive sensors? this is a big question.

[msiegel]

:O gaaah!

dem bones dem bones
gonna walllk around

howdy ripster, it's a model f revival XD

[talis]

ok, at least we know one thing that *won't* work (thanks ripster

back to capacitive senors...

Q: can model m membranes be turned into capacitive sensors? this is a big question.

Unlikely, small local movements in the material when neighboring hammers were actuated may in fact produce enough of a change to trigger another capacitive sensor.

Designing and manufacturing a new "custom" PCB for such a project really isn't a major undertaking (taking time cost as $0, prototype quantity boards can probably be made for $80-$100 with little to no tooling charge).  There's plenty of free tools out there for both schematic capture (TinyCad/Eaglecad) and PCB layout (FreePCB).   In production quantities, the board cost would probably be in the order of $4-$5.

good point. what do new custom springs cost these days, in lots of 2500?

If you search alibaba.com, you could probably find an Asian manufacturer that would sell you some for $0.04-$0.08 each.

[msiegel]

small local movements in the material when neighboring hammers were actuated may in fact produce enough of a change to trigger another capacitive sensor

hmm, this may call for testing. if we can get it to work, it shaves $40-$80 off the end user's cost. unfortunately i'm model m-less at the moment...

If you search alibaba.com, you could probably find an Asian manufacturer that would sell you some for $0.04-$0.08 each.

bueller?

bueller?

where's that handy lab assistant when you need him?

[msiegel]

LOL - this Model F mod is going well.  Wake me when the prototype is ready.

:lol: you're awesome ripster, and imma let you finish... but model f is one of the best keyboards of all time! of A L L - T I M E!!
XD


edit: http://noiseforairports.com/post/178842028/piano-piece-13-by-george-maciunas