XT/AT/PS2/Terminal to USB Converter with NKRO

[Retrete]

I will try to get a multimeter and verify that.

So as controller swaps go, it looks fairly straightforward - remove all the chips

I don't want to break everything xD,but If wiring a teensy would make it operative, I'll make you a statue hahahah.

Offtopic, do you know if theres any shop who sells your teensy in Europe? 86 dollars for shipping tracked 

[Soarer]

If the chips are removed carefully, nothing will break and it could all be put back how it was - if you remember where they all went

The untracked shipping from PJRC has been reliable to me (in the UK), and if you're buying more than one that works out to be the cheapest option overall. Otherwise, there's Floris.CC. Also, there are alternatives that use the same CPU as the Teensy 2.0 (warning: that page needs a clean up, some have posted non-compatible boards there).

I think a Teensy 2.0 would be enough for this - it probably has enough pins available. But it might be easier with Teensy++ 2.0, just because of being able to make a more logical set of connections, and less cramped. It's not like you'll be short of space to put it in that casing!

[Retrete]

Today I spotted a breakout titled "Leonardo Nano Pro Mini ATmega32U4 ..."

Still a good choice?On ebay for ~13€ with free shipping.

[Soarer]

Probably! However, two points...

The Teensy loader (as in, what gets code onto the chip) is by far the easiest to use.

You still need to check how many columns there are on that keyboard; it might be too many for the ATmega32U4 chip. There's 25 usable pins in all, but some may already be used by LEDs (one LED on Teensy). Plus, you might want to add the 3 lock LEDs. We've already spotted 8 rows, but then there's the shift keys which might need a pin each (since one side looks like it's joined to ground permanently). So 25 - 1 - 3 - 8 - 2 = 11 pins left for columns. It ought to be enough for 86 keys, but only just!

[Retrete]

So teensy loader will only work with a teensy board correct?
Analyze the front part of the PCB is nearly impossible, I've made a really awful looking scheme,


Sorry for my totally retarded questions 

[Soarer]

So teensy loader will only work with a teensy board correct?

Yes, essentially that's correct.

There's two other common ones... Atmel's FLIP works with the default Atmel bootloader on many dev boards, and dfu-programmer (or the Arduino environment) works with Arduino compatible dev boards.

Analyze the front part of the PCB is nearly impossible, I've made a really awful looking scheme,

Sorry for my totally retarded questions 

No problem!

Scheme looks about right... the blue wires can stay, since removing the chips means the wires won't connect to anything. It loks like the chips are in sockets, so they should come out with a bit of careful prizing with a screwdriver (a little twist at one end of the chip, a little at the other, ...). Analyzing all of the front of the PCB shouldn't be necessary - it's clearly 8 lines going to the socket at the end, and a group of 11 lines going to that other one - as predicted It looks like the shift keys (and maybe one other?) conect to the last three pins of that socket too.

Of course, I'm making some guesses, so this should be checked with a multimeter (not every single key's connection, but enough to know that I've guessed correctly).

[neocalm]

Sorry - misposted.

[neocalm]

Well Soarer - it only took nine months but after experimenting with conductive paint, and later on copper conductive tape - both didn't appear to be long-term solutions as they might wear down too much.  One bolt-mod is enough.

I took your advice and Unicomp does indeed sell the membrane.

Bolt-modded back together using 78 of:

• M2x0.4, 8mm length pan head machine screws
  
• M2x0.4 hex nuts

(Amazon amongst other carries these) and there was absolutely no issue.  Soarer's Converter is absolutely everything I could have dreamed, for 122-keys of mechanical switching goodness.

Thanks again!!

Rear defogger repair kit?

Yeah. Or from electronics suppliers (example).

I should say I've no idea how well or even if it would work, but in theory it might!

My reservation in this case is that I suspect a new membrane might not cost much more, so that's definitely worth enquiring about.

[Retrete]

I managed to get a multimeter, I don't know If is not working correctly or I am def stupid  ;
As you said a few posts back, So most likely to be simple contact switches. You could check with a multimeter easily,
I take the multimeter, both wires on it, test it with a 9v batterie and one of the resistances of the kb,the multimeter seems to work.
But, when I test the switches, voltage, resistance, amp all 0, I've tried it with all the precision positions available, which type of result I've to find?

Thanks again!

[Soarer]

Right... on a resistance setting, you should find the meter reads "OL" or goes blank or something, then if you touch the probes together it should read 0 (or close to 0). You might find it has a continuity test setting, which will beep when you touch the probes together...



So to check the switch, hold a probe each on the contacts marked red and black - we're expecting it to read 'OL' without the switch pressed, and close to 0 with it pressed. (Yes, you might run out of hands!)



Then you could try to confirm the connections from that switch to the sockets... my guess is that the red one connects to one of the pins on the purple socket, and the black connects to a pin on the blue socket.

[Retrete]

Multimeter doesn't work well, in "AUDIBLE CONTINUITY TEST" never beeps, In the manual says that if the resistance is lower then 1k ohm buzzer will sound, so putting both probes together must return into a beep as the screen value goes near to 0 (default 1, not OL)
But it doesn't, and when I make the contact with those marked,return this;

Only a few switches return that value, the rest stay at the 1 deafult.

[Soarer]

Shame about the buzzer, that does make it easier

Some relatively high resistance instead of OL (I'll say OL instead of 1, so it's clear what I mean) is quite possible, and 1811 is high enough. (The resistance is elsewhere in the circuit, not in the switch itself).

Does it go close to 0 if you press the switch?

[metalliqaz]

Get a proper meter dude.  It is a worthwhile investment.

[Retrete]

Get a proper meter dude.  It is a worthwhile investment.

Yep, I don't learn the lesson, months ago I bought a (welder?) for 3€ in a chinese shop instead of paying 10€ for the local one. The welder exploded, comical scene(and also scary...)
Multimeters can explode? 
Tomorrow I will try to return it.

Shame about the buzzer, that does make it easier

Some relatively high resistance instead of OL (I'll say OL instead of 1, so it's clear what I mean) is quite possible, and 1811 is high enough. (The resistance is elsewhere in the circuit, not in the switch itself).

Does it go close to 0 if you press the switch?

Yes! It goes closer to 0 when pressed

[Soarer]

Yes! It goes closer to 0 when pressed

Cool, the switches are simple contact switches then

How about finding a connection between red and a pin on the purple socket? The switch doesn't need to be pressed for that test, just hold one probe on red and try the other probe on each pin of the purple socket, to see if it goes close to 0 on any of them.

Then the same between black and the pins on the blue socket

And then return the meter!!  )

[Retrete]

Found! I will record a video , it's a little bit weird
The video;

[Soarer]

Not weird at all, it's just how I'd expect it! Pressing the switch joins one connection on the purple socket to one connection of the blue socket. All the other switches will join a different combination, forming the matrix.

Here's a switch matrix from a pinball machine



Anyway, I think that testing confirms that rows are on the purple socket, and columns are on the blue socket. The controller would scan the matrix by setting the columns active, one at a time, and each time reading the rows.

Apart from three keys - Left Shift, Right Shift and Reset - which look like they have separate connections...



The trace from the Reset key (black) is mostly visible. But can you check that you get a low reading between the green circle and the end pin on the socket, and also from the cyan circle to the next pin?

Then we have almost all of the information we need. The last question is which of the pins on the purple socket are actually connections to rows. We already know pin 2 is. (well, the second pin on that side - I think it might actually be pin 10 going by the marking on the top of the resistor pack). The rest I can see in this picture, but not quite clearly enough to tell for sure...

[Retrete]

can you check that you get a low reading between the green circle and the end pin on the socket, and also from the cyan circle to the next pin?

Both goes to ~0

The last question is which of the pins on the purple socket are actually connections to rows

Im now getting which pins makes nearly zero, I will post now the scheme, I also took a new photo of the board;


The scheme;
http://www.flickr.com/photos/95230480@N02/8694766662

[Soarer]

That pic's nice and clear

So these are all of the 22 places that a new controller would have to connect to on the keyboard...



Then maybe the controller also needs a few pins for LEDs (onboard and locks).

edit:

Scheme is cool 

I think the line to the left shift should go to the key next to it though

The line you've drawn to the right shift is just a ground... you would need to make a connection from it to the new controller, but it won't use up an IO pin.

[Retrete]

Trying to understand the left&right shift issues I've found this;


Tomorrow I'll make the Teensy order, the mini-usb cable used its a special one? Or I could use any other?
Teensy 2.0 or ++ 2.0? 6€ more, worth it?

[Soarer]

There will be lots of ground connections around... and one side of each of Left Shift, Right Shift and Reset is connected to ground also, by this trace on the right:



The other sides of those switches we already know use three pins on the blue socket. In a sense, they are wired separately from the matrix. It would be possible to modify the PCB to make them part of the matrix, but there's no need.

I think Teensy 2.0 doesn't have quite enough pins - only 24 are available, and one of those is awkwardly in the middle of the board. But it depends if you want to connect LEDs for caps lock etc, and how many. I'd go for the Teensy++ 2.0, if it were me. The mini-usb cable is just a standard one.

[Retrete]

I didn't metioned yet,I also have a (space saver?) Teletype M43 keyboard, the clickiest I've ever heard.
I think that It's also teensy capable, but I am frightened with this;


Little gallery (It looks like all this new space saver mini mechanical keyboards  )
http://www.flickr.com/photos/95230480@N02/sets/72157633371311121/

Im doing the scheme of the pcb, I will post it now.

So, 2 teensy's? 

[Soarer]

Sure, why not! Maybe get one of each - since that has less keys a Teensy 2.0 might be enough. As long as the switches measure ~0 resistance when pressed, it should be possible.

You should be fine working on it if you touch something which is grounded (like an unpainted radiator pipe) before touching it, to get rid of any static charge you might have. It's usually only an issue when removing chips, and then you can try not to touch the pins.

Removing the chips looks like it would be difficult, so you would have to hope that the new controler can work OK without removing them (but fairly good chance of that).

[rknize]

If anyone is old enough to remember the 4000-series CMOS logic gates, they were/are very sensitive to ESD.  People accustomed to TTL circuitry got a nasty surprise when they handled early CMOS stuff without a static strap.  Wear appropriate protection if you dig into this device.

Modern gate logic is pretty much all CMOS, but they have many layers of protection diodes on each leg of the part to prevent most ESD problems.

[Retrete]

Sure, why not! Maybe get one of each - since that has less keys a Teensy 2.0 might be enough. As long as the switches measure ~0 resistance when pressed, it should be possible.

You should be fine working on it if you touch something which is grounded (like an unpainted radiator pipe) before touching it, to get rid of any static charge you might have. It's usually only an issue when removing chips, and then you can try not to touch the pins.

Removing the chips looks like it would be difficult, so you would have to hope that the new controler can work OK without removing them (but fairly good chance of that).

W/o pressed ~0 , and when pressed default 1 value, the scheme;


Edit: Teensy-with his cable and Teensy++ incoming  , 56$ 
Hope it works 

[Soarer]

I'm confident that the first one can be made to work.

The M43 though, after seeing that diagram, may take some effort... it doesn't look like a matrix, and I'm guessing that all the other pins of the switches (the left-hand pin of each switch, in that pic) connect together. The best approach might be to work out how to use it via the cable connector instead. The first step would be to try and find a Technical Reference or a Repair Manual for it.

[Retrete]

The package arrived today! Only 1 week 
It's extremely necessary to remove all the chips which do nothing? Or I could let them stay?

[Soarer]

1 week is good!

Technically you might get away with not removing the chips, but there is a risk that they could get damaged, or that they could obstruct the operation of the Teensy controller. They're really not all that difficult to remove (just do it carefully, and record what went where). What you might not have is something good to store them in - either anti-static foam or an IC tube is best...

[Retrete]

Chips removed, but I don't know If I also need to remove the resistances or&and the " special chip";


Now, how I should wire the teensy?


Thanks a lot for your help,w/o your knowledge this would still taking dust in the lab 

[Soarer]

The other resistors etc can stay. Most will now be disconnected from the switch matrix, after removing the chips.

As for wiring, there is some flexibilty... let's start by labelling the pins on the keyboard:



We still need to find a convenient point to connect GND to ground, as well.

On the Teensy++, it could go something like this:



I wonder, would the Teensy mount on the pins 0 to 10? (tilted at an angle to avoid touching the pins behind them). That would save a lot of wiring! (It would be harder to remove though).

[Retrete]

Like this ?

Now I'll try to see how the rest pins could be connected w/o soldering.

Any advice with the type of wire?

[Soarer]

Unfortunately (?), you will most likely need to do some soldering to get this done!

I was thinking of putting it this way round (and tilted so it doesn't touch the next row of pins along):



Another option would be to flip it over - still connecting to the same pins. But then the reset button would be hard to get to, since it would be a) underneath, and b) under the metal plate.

The main problem with the way you've shown it is that there are the GND and AREF pins in the middle of that side, which we can't use for the matrix connections.

You might be able to use some breadboard wires to avoid some soldering, at least on the keyboard side. Something like these female to male or these female to female (which are probably available more locally to you). You could solder them at the Teensy end, or solder pins to the Teensy and then plug them on.

[Retrete]

With patience and bending slowly a little bit the front pins I finally managed to connect it with that position;

I bought a good amount of female-male arduino connectors, kb-female---------male-arduino, and I would probably solder these (theres always the possibility just with pression)
With that position the GND and AREF problems dissapear?

Thanks again!

[Soarer]

That way round, but moved along 3 positions - so the X,Y,Z pins are not connected into the Teensy. The GND and AREF are not a problem along this side, but PD6 is, since it has the LED on the Teensy connected to it already. Pins PD7 along to PC7 gives the most pins together  in a row that can be used. Something to insulate the Teensy from the next row of pins would be wise - a few layers of electrical tape on the bottom of the Teensy should be enough.

A warning here: if you solder the Teensy directly to those pins on the keyboard, it will be quite hard to remove! I can't make the choice for you - you can either fit it there with less individual wires to do, or elsewhere with more wires but far easier to remove. Friction fit on those pins might be OK for testing, but I wouldn't trust it long-term.

My controller code is progressing, the main part of it has been tested and works fine. What I'm busy with now is allowing the connections and key mappings to be defined using a config file, and most parts of that are written but not yet tested. Maybe ready for a try-out in a few more days!

[Retrete]

I bought enough pin-adapter, so I'll probably connect all the pins to the teensy using them

not a problem along this side, but PD6 is, since it has the LED on the Teensy connected to it already.

I didn't get that , I won't put any leds, at least for now, don't want to complicate it even more 

[Soarer]

Just that there is already an LED on the Teensy, which uses PD6

[Retrete]

Solderless cables arrived, I will edit now with some photos
If the position are the definitive I'll solder them now
--------
Soldered and plugged;

[mkawa]

this post is to for force alaric to post the bom and howto for this awesome soarer he made for me

[Soarer]

Solderless cables arrived, I will edit now with some photos
If the position are the definitive I'll solder them now
--------
Soldered and plugged;

Show Image

Dang... stuff happened!

My code basically seems to work now, at least on Teensy 2.0. It should work on Teensy++ with a recompile, I'll build it if you wanna try it out!

[Retrete]

My code basically seems to work now, at least on Teensy 2.0. It should work on Teensy++ with a recompile, I'll build it if you wanna try it out!

Of course! It will be a pleasure testing it 
I'm also making a video-resume of the whole steps, im building your statue 

[Soarer]

OK, I'll try and put it together tomorrow evening. Are you running Windows?

[Retrete]

yes, Windows 7.

[Soarer]

Cool, I won't need to build the Linux or Mac tools then

[Soarer]

this post is to for force alaric to post the bom and howto for this awesome soarer he made for me

Show Image

Show Image

That is a really nice little case!

[Soarer]

My code basically seems to work now, at least on Teensy 2.0. It should work on Teensy++ with a recompile, I'll build it if you wanna try it out!

Of course! It will be a pleasure testing it 
I'm also making a video-resume of the whole steps, im building your statue 

Right, here's a zip of controller stuff, which should be all you need for the moment. You'll also want to get the latest converter zip, for the docs, since the method of loading a config is the same. Then you'll need to load Soarer_Controller_1.2_at90usb1286.hex onto the Teensy++, and then compile (scas) and load (scwr) the kevex.sc config file.

That will set it up to scan the matrix, but it won't be outputting any keys (it would get messy if it happened to spew out random keys!). It will be printing stuff out that can be viewed using hid_listen - if it's basically working you'll see nothing much until you press a key, and a hex code prefixed by either \ (key down) or / (key up). The hex code corresponds to the matrix position of the key - the strobe number first, and the sense number second.

If that all looks good, you can start editing a new config file, replacing all the UNASSIGNED entries with useful HID codes (such as A or SPACE, see the Codes page of the docs). So you press Q, see what the code is, plug Q into the config at the right place. Rinse and repeat!

Oh, the three lines near the end beginning with 'unstrobed' are left shift, right shift and that reset key... but I'm not sure what order. They will have matrix codes of B0, B1 and B2, in that order.

[alaricljs]

Show Image

Show Image

Case:  Mouser - 789-S3A-301210
Connector: Mouser - 568-NYS325  -PS/2 alt: 161-381/6-E
3 LED resistors: 1kohm Mouser - 603-CFR-25JR-521K
3 LEDs: 3mm flat top orange, 2.9v nominal from ebay
USB connector:  "Mini USB Female 5Pin Socket Connector 180Degree" http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=170588741155
spare mini USB cable to destroy
Teensy
loose wire, used old PC case mobo leads (power switch and the like) ~22ga?
JB weld epoxy to hold the USB connector in
2 spare PC screws for AT connector.

The 1k resistors make the orange LEDs a nice indicator brightness.  Yes, I made a sacrificial USB lead to connect to the teensy.

[Retrete]

Right, here's a zip of controller stuff, which should be all you need for the moment. You'll also want to get the latest converter zip, for the docs, since the method of loading a config is the same. Then you'll need to load Soarer_Controller_1.2_at90usb1286.hex onto the Teensy++, and then compile (scas) and load (scwr) the kevex.sc config file.

That will set it up to scan the matrix, but it won't be outputting any keys (it would get messy if it happened to spew out random keys!). It will be printing stuff out that can be viewed using hid_listen - if it's basically working you'll see nothing much until you press a key, and a hex code prefixed by either \ (key down) or / (key up). The hex code corresponds to the matrix position of the key - the strobe number first, and the sense number second.

If that all looks good, you can start editing a new config file, replacing all the UNASSIGNED entries with useful HID codes (such as A or SPACE, see the Codes page of the docs). So you press Q, see what the code is, plug Q into the config at the right place. Rinse and repeat!

Oh, the three lines near the end beginning with 'unstrobed' are left shift, right shift and that reset key... but I'm not sure what order. They will have matrix codes of B0, B1 and B2, in that order.

After loading the Hex onto the teensy, opening scar or scwr result in a pop up of less than 1/10 sec. I think that's normal without dropping the sc file, but,dropping it still the same pop up.
If I drop the sc file onto the scaswr I get this message;

The program also automatically creates a new file, "kevex.sc.scb",is this normal?

[Soarer]

Apart from the not found message, that's all as expected. The tools are command line programs. For now, dropping kevex.sc onto scaswr.bat is all you need. The file it creates is the output from scas, and that's good.

For my tools to find the Teensy it should be running my hex. Once the hex is loaded, click the Reboot button (if you're not using the Auto mode) but don't press the reset button again. Then scwr should be able to find it.

[Retrete]

No more error! It ends w/o problems;

BUT, hid_listen doesn't work, I have touched all the keys and no one respond;

[Soarer]

Hmm, that'a puzzle. It seems that hid_listen has connected OK, just nothing is being printed... which probably means the matrix scanning isn't working. Just to check, have hid_listen open when you drop onto scaswr.bat - stuff should be printed about the config being loaded.

Did you make a ground connection? It probably won't help with most of the matrix scanning, but it's essential for the 3 unstrobed keys.

The only thought on the matrix scanning is that maybe the resistor pack on the sense lines is pulling down instead of up (I expected it to pull up). Could try again after removing it... I mean the whitish thing that looks rather like an IC.