Teensy RUMP

[Hellmark]

As a few of you may know, one of the projects I am working on is getting a good model M going. One of the things that would be cool to do, is do an updated version of RUMP, where you can drop in a Teensy instead of the plain jane Atmega32, and run it as an internal soarer's. I know a teensy runs an atmega32, but things were more kept in line with what most people have already been used, just giving a cleaner approach.

Most I've seen have been kinda hackish, with soldering the teensy to wires and jumpering it in.

One thing that may make things somewhat difficult, or redundant, is that soarer's expects output from a controller, and not directly from the keys. Without the source to tweak, it would be mean that an additional chip would be needed to act as a middleman. That would mean using something else as a basis if I wanted something with no additional chips, and miss out on any improvements from Soarer.

Am I wrong in any of my assumptions? Would this be something anyone else be of interest in? What would your recommendations be for this? Should I just say screw it, and have a normal RUMP, instead of wanting a teensy RUMP (I feel like this is a question my wife would be asking).

[snoopy]

I don't know much about that controller stuff, but a programmable replacement controller for the Ms, that doesn't need modification of the M, would be awesome.

[xavierblak]

Shouldn't be too difficult. It would basically be a breakout board for the teensy with headers to accept the model M's ribbon connectors.

One problem I see though is the sizing of the PCB. Looking at this photo. It looks like the original pcb has varied in size. If we picked a size for the new pcb would it work for all the model m's?

Edit:
I had to google the rump to see what it was. Here's the link: http://mg8.org/rump/

[wcass]

Soarer's is currently just a converter (you need a keyboard controller).

Hasu's TMK code is for a USB controller - suggest you start with that. It is used with many custom keyboards here. I have the FFC connectors already and some familiarity with the code and hardware, so i can help.

[mkawa]

what's wrong with the M controller? unicomp has huge numbers of those boards. in this case i would just say why bother reinventing the wheel?

for a design like wcass's, you need a custom eeprom (yes, there's actually a discrete thru-hole eeprom chip involved) and hence it makes sense to throw an mcu on with controller firmware instead of screwing around with the ibm/lexmark controller. otherwise, i don't see the need.

[Melvang]

I would have to agree with mkawa here.  Just stick with the convertor on this one.  Reason being the Model M switches are capacitive membrane based IIRC.  Not a standard momentary switch like MX or ALPS.

[mkawa]

no they're very simple spst membrane switches. however, getting a board made and designed to fit the controller spacers is more pain than it's worth

[Melvang]

no they're very simple spst membrane switches. however, getting a board made and designed to fit the controller spacers is more pain than it's worth

Alright I stand corrected.  So what is the deal with with the capacitive buckling spring switches I am hearing about?  Or am I just completely misunderstanding something?

[Hellmark]

Well, for me, part of why I was wanting to make a board, is I do like some aspect of customization, I have the parts, I've made my own PCBs from scratch, and I like keeping a removable cable. Unicomp's controllers, as far as I know, have the cable hardwired in. Plus, I like making things.

That said there are a few other reasons why I would want to go with a non Unicomp controller. Part of which have been oddly marked up on by Eric Raymond. Undervolting, no built in strain relief on the cable, etc are things that need to be considered. Yeah, I could modify the Unicomp controller, but that would kinda go against the basic concept of having a clean controller.

Not only that, but as was mentioned, the model M's PCBs changed size a few times. Would the one from unicomp fit my model M's? The one I am currently typing on is a 1390131 from 1986, which has the larger size PCBs. From what I've seen, everything from Unicomp is based on the later Lexmark keyboards. If someone is more familiar with the innards of the Unicomps in relation to how things would work with the older IBMs, please step forth.

So, I would be spending additional money for a controller that has some design issues I don't like, may experience USB problems, and there is a chance that it may not even fit my keyboards.

Melvang, the capacitive buckling spring switches are the earlier model F keyboards. The Model M kept the buckling spring aspect (which is part of why they last so long, the contact of the membrane isn't direct like with rubber domes, so it is a constant force), but really all about making things a lot cheaper for IBM.

[wcass]

Alright I stand corrected.  So what is the deal with with the capacitive buckling spring switches I am hearing about?  Or am I just completely misunderstanding something?

IBM buckling springs come in two varieties*; Model M (m for membrane maybe) and Model F (for faraday maybe). The feel on these switches is subtly different (think Cherry Blue vs Ergo Clear). The M uses a conductive matrix controller and the F uses capacitive matrix controller. Most BS afficionados consider the F to have the superior feel.

[mkawa]

Well, for me, part of why I was wanting to make a board, is I do like some aspect of customization, I have the parts, I've made my own PCBs from scratch, and I like keeping a removable cable. Unicomp's controllers, as far as I know, have the cable hardwired in. Plus, I like making things.

we're certainly not against making things for the sake of making them here. heck, it's about half of what we do. that said, the unicomp controllers have SDL jacks, as they are literally new old stock model M controllers.

That said there are a few other reasons why I would want to go with a non Unicomp controller. Part of which have been oddly marked up on by Eric Raymond. Undervolting, no built in strain relief on the cable, etc are things that need to be considered. Yeah, I could modify the Unicomp controller, but that would kinda go against the basic concept of having a clean controller.

Not only that, but as was mentioned, the model M's PCBs changed size a few times. Would the one from unicomp fit my model M's? The one I am currently typing on is a 1390131 from 1986, which has the larger size PCBs. From what I've seen, everything from Unicomp is based on the later Lexmark keyboards. If someone is more familiar with the innards of the Unicomps in relation to how things would work with the older IBMs, please step forth.

So, I would be spending additional money for a controller that has some design issues I don't like, may experience USB problems, and there is a chance that it may not even fit my keyboards.

no, they have new old stock controllers. give them the model number on your board and they will poke around for the correct controller (it's almost certainly there in some box somewhere). there are no usb/ps2 draw problems with the ID innovations adapters i sell. they are the preferred adapter for the model M. they handle all kinds of funky stuff without hiccuping. also, ESR has an account here , so in case he's reading he may want to update his faq

[wcass]

Well, for me, part of why I was wanting to make a board, is I do like some aspect of customization, I have the parts, I've made my own PCBs from scratch, and I like keeping a removable cable. Unicomp's controllers, as far as I know, have the cable hardwired in. Plus, I like making things.

That said there are a few other reasons why I would want to go with a non Unicomp controller. Part of which have been oddly marked up on by Eric Raymond. Undervolting, no built in strain relief on the cable, etc are things that need to be considered. Yeah, I could modify the Unicomp controller, but that would kinda go against the basic concept of having a clean controller.

Not only that, but as was mentioned, the model M's PCBs changed size a few times. Would the one from unicomp fit my model M's? The one I am currently typing on is a 1390131 from 1986, which has the larger size PCBs. From what I've seen, everything from Unicomp is based on the later Lexmark keyboards. If someone is more familiar with the innards of the Unicomps in relation to how things would work with the older IBMs, please step forth.

So, I would be spending additional money for a controller that has some design issues I don't like, may experience USB problems, and there is a chance that it may not even fit my keyboards.

Melvang, the capacitive buckling spring switches are the earlier model F keyboards. The Model M kept the buckling spring aspect (which is part of why they last so long, the contact of the membrane isn't direct like with rubber domes, so it is a constant force), but really all about making things a lot cheaper for IBM.

When you ask Unicomp for a replacement IBM controller, you get an NOS IBM controller. If you have a part number, you usually get that same part number.

All of the (bottom fit) M controllers have the same width, but might have different length. IIRC either "long" or "short" controller will fit in any. The big difference is how the LEDs are connected. some use a 4 pin SIP, some use 4 pin FFC, and some combine the LED pins with the matrix rows FFC. So a "good" replacement would need to adapt to all 3 configurations.
 
I agree that a removable cable would be nice.

[Findecanor]

One problem I see though is the sizing of the PCB. Looking at this photo. It looks like the original pcb has varied in size. If we picked a size for the new pcb would it work for all the model m's?

The issue is not the size of PCBs, but the attachment points. Some Model M PCBs are laid on the bottom, others are bolted to the backplate; the Unicomp's is under the LEDs.
One needs to survey which different cases and backplates there are (did they really change at all?) and make sure that the new PCB design has holes for mounting them in all these different ways.
If Unicomp's membrane/matrix is compatible but just in another location, then maybe the new controller could support Unicomp keyboards too.

Hasu's "TMK" firmware already has supports for a hacked Model M keyboard (hbkb). It should not be too difficult to extend it to the full one, I think.

I have previously been toying with the idea of making a controller board myself,  using a Teensy (or Teensy++) in a socket on a stripboard bolted to the back of the backplate with an internal cable going to a USB socket at the bottom of the case.

[Hellmark]

we're certainly not against making things for the sake of making them here. heck, it's about half of what we do. that said, the unicomp controllers have SDL jacks, as they are literally new old stock model M controllers.

no, they have new old stock controllers. give them the model number on your board and they will poke around for the correct controller (it's almost certainly there in some box somewhere). there are no usb/ps2 draw problems with the ID innovations adapters i sell. they are the preferred adapter for the model M. they handle all kinds of funky stuff without hiccuping. also, ESR has an account here , so in case he's reading he may want to update his faq

But getting a NOS controller kinda defeats the purpose of what I want here. I already have an M with a SDL jack on the controller. Why buy an NOS controller when I have an original controller that works.

Tonight, I've been working on bolt modding it, and I want to test it out, and I don't have any running machines with PS/2 at the moment, so this is a time when USB would be handy. My options are spend $40 on a cable to switch to USB and just hope nothing happens to the cable, to replace the controller with something custom, or have an inline adapter.

Third option kinda sucks if I want to tote the M around with me, as it is more crap to carry and keep track of. First option isn't all that attractive because of the price and has some of the same vulnerabilities as any other SDL cable (plastic clips, high price if I need a spare or something, etc).

When you ask Unicomp for a replacement IBM controller, you get an NOS IBM controller. If you have a part number, you usually get that same part number.

All of the (bottom fit) M controllers have the same width, but might have different length. IIRC either "long" or "short" controller will fit in any. The big difference is how the LEDs are connected. some use a 4 pin SIP, some use 4 pin FFC, and some combine the LED pins with the matrix rows FFC. So a "good" replacement would need to adapt to all 3 configurations.
 
I agree that a removable cable would be nice.

For me, that more or less is going to be a sticking point going forth. I've had cables wear out, and fail. I want to be able to easily replace it if I need to.

The issue is not the size of PCBs, but the attachment points. Some Model M PCBs are laid on the bottom, others are bolted to the backplate; the Unicomp's is under the LEDs.
One needs to survey which different cases and backplates there are (did they really change at all?) and make sure that the new PCB design has holes for mounting them in all these different ways.
If Unicomp's membrane/matrix is compatible but just in another location, then maybe the new controller could support Unicomp keyboards too.

Hasu's "TMK" firmware already has supports for a hacked Model M keyboard (hbkb). It should not be too difficult to extend it to the full one, I think.

I have previously been toying with the idea of making a controller board myself,  using a Teensy (or Teensy++) in a socket on a stripboard bolted to the back of the backplate with an internal cable going to a USB socket at the bottom of the case.

On the M I was working on tonight, it definitely was different than the later M's I am used to. Had pegs for the controller to latch onto, where as the others I've had apart, would bolt the PCBs down. From what I saw though, the plate and everything looks the same. Only real difference being the PCB, and the case backing.