mirrored from: http://home.t-online.de/home/stephan.hans/tricks.htm

(last update: 24.02.1999)

Project # 1: DUAL-KEYBOARD Circuit

• First, the history: Why should somebody need a dual keyboard circuit:
  
  Some days ago, playing arcade games with mame32, I had the idea (like many before) to build my own superduperarcadejoystickconsole.
  There are many different ways to connect a joystick-like device to the pc.
  The easyest is to use the gameport, but the good old gameport is restricted to 4 digital and 4 analog inputs, so you can only connect 2 joysticks with 2 buttons (each) to the gameport, or if you like, a single joystick (even gamepad) with 4 buttons (or by using the remaining two analog inputs for buttons: 6 buttons).
  There are solutions with special interface-cards, solutions using the serial port and there is even a proposal for abusing the parallel port for joystick purposes. But all these solutions require special drivers (in mame/p the parallelport joystick driver is integrated) and sometimes won't work, depending on the os-environment.
  Another approach to the problem is to use the keyboard interface. With the power ramp mite gamepad for example, you can press/release a button (or move up, down, left or right) and the device simulates a pressed or released key to the pc. If you want to use your own buttons/controls you can take a power ramp mite, dismount it and reassemble your own controls to it; as far as I know it works with up to 20 buttons. Needless to say that the normal remaining keyboard is connected to the prm-gamepad and works as usual. But the power ramp mite is not everywhere available, and sometimes a little bit expensive.
  If you don't care about money, you can buy a special keyboard encoder which could do the same thing.
  For me, who prefers cheap selfmade solutions there is another way:
  With an old keyboard you can do nearly the same thing, but the question remains: how can I connect my normal keyboard and the old, as arcade console abused keyboard to the pc, without loosing the functionality of my normal keyboard ?
  The first thing I did was searching round the net, then asking at different message boards (davesclassics).
  Result: oops...
  So I had to use my own brain. I 've had different approaches to the problem, some of them using flip-flops, some microcontrollers.... everything too complicated and too expensive... But then, while driving to my office, I remembered the possibilities of the good old transistor.
  
  And now, here it is: the ciruit you need to use two keyboards simultanously at 1 pc:
  

PHOTO / prototype:

Sheet:

How it works:

The pc remains able to communicate with the keyboard 1 by pulling down its data line (pin 2 of the DIN connector) using diode D1. When one of the keyboards communicates with the pc, the data line is pulled to low (with the start bit) by the corresponding keyboard. This is detected by T1 and T3. Via T2 and T4 the cmos switch 4053 is triggered and pulls the clock line (pin 1 of the DIN connector) of the other keyboard to low. So the other keyboard buffers its data until the clock goes high again, this time depends on the combination of C1/R2 and C2/R4 and must be long enough for that the first keyboard is ready with sending its data.

By the way, with two transistors, two resistors, one capacity and 6 diodes more, the whole thing could be extended to work with 3 keyboards (the cmos circuit 4053 contains already 3 switches).

Updated:

Thanks to all of them who responded to my circuit here are some answers to the frequently asked questions:

Some explanations (for those guys who never heard of nanofarads):

Some hints for the transistors:

Instead of the BC548 you can use nearly every n-type bipolar transistor (e.g. 2N2222 or 2N3904) and
instead of the BC558 you can use nearly every p-type bipolar transistor (e.g. 2N2907 or 2N3906).

Something about the R/C part in my circuit:

As you can see, C1 and R2 (and C2 / R4) form a small timer. During the time determined by one of these pairs, the circuit is blocking the "other" keyboard while the "one" is transmitting its data. I've choosen the values for C1/C2 as 22 nF and for R2/R4 as 47 kOhm (47 000 Ohm). Because the timing factor depends on the product of both, you can also take 100 nF and 10 kOhm instead of it. But don't choose the value of the capacitor to high, because the transistor has to shorten its charged power.

And now to all of them who wanted to use PS/2 connectors instead of the 5 pin DIN connectors:

Sure it's possible, but take care of the different pinout for the PS/2 connectors:

And now for them who have begged me to show some more pictures of the prototype, especially the "layout" side, here they are, but don't complain if it's too confusing (so be warned):
(By the way, don't bother about the two LEDs that I used for "control" purposes)

"component" side

"layout" side

Project # 2: AUTOMATIC JOYSTICK SWITCH
(by popular demand)

• Due to the numberous responses to the dual keyboard circuit, I invented an automatic joystick switch for the use of two joysticks (or gamepads) at one gameport. It's not an ordinary Y-cable, it really switches between joystick 1 and joystick 2 by pressing button 1 at the desired joystick (take over control).
  So you can use an analog joystick for the games where this is the best input device and a digital type joystick (gamepad or something else) for the other type of games, where typical digital inputs are required (like frogger or pac-man and many many more).
  
  (sure you can do the same thing by plugging the one or the other joystick-device to the gameport, but this is really annoying)
  
  o.k. here it is:

prototype:

Project # 3: Automatic switch for two serial mices (or trackball, or spinner, or touchpad.....)

• In Win 95 you can use a ps/2 mouse and a serial mouse-like device simultanously. So you can use your standard ps/2 mouse for standard input purposes and a trackball or a spinner for gaming purposes.
  
  But what can you do if you want a trackball AND a spinner connected to the serial port ???
  Here's the answer.
  
• My circuit consists of two parts. One part is the "power-splitter" and the other one is the automatic switching circuit.
  The power-splitter supplies the two mouse-like devices with +/- 6.1V from the serial port (most mices work above +/- 4.5V).
  The automatic switching part detects which mouse is in movement and switches over to it after messing around with the mouse approximately 1..2 seconds. This avoids accidential interferences when you've touched the -undesired- control unintentional.
  
  If you don't want that automatic switch, you can simple use a standard change-over switch for the RXD-Lines (pin 2 of the 9 pin serial port).
  
• Now here's the circuit:
  

• Due to the limited power supply (+/- 6.1 V) it might be possible that some mices won't work with the circuit properly.
• The circuit has been tested with an original microsoft serial mouse, a genius mouse (gm 303) and an glide/touchpad. It worked with any combination of them without any problems.
• Ooops, nearly forgot to say: J1 is a female connector and should be connected to the PC, J2 and J3 are male for the connection to the mices.

If you would say the power splitter is rather complicated, you're right. It's only because I've had no more zener diodes but plenty of transistors (btw: for me a transistor is even cheaper than zener diodes). Now here's a proposal for a much easier power splitter, which should work instead of the "complicated" one (haven't tested it 'til now):

Project # 4: SUPERDUPERARCADEJOYSTICKCONSOLE

Now, lets go on with the arcade console:

• 1st: The required parts (pushbuttons, joystick-units (with microswitches)) are ordered and delivered. Btw, I replaced the cheap microswitches at the joystick units by better ones.
• 2nd: The dual keyboard circuit works without complaint.
• 3rd: The keyboard I decided to dismantle is a Cherry G81-3000 (with german layout)
  After gutting the keyboard, it was easy to find out that it's keyboardmatrix is made by 8 columns and 19 rows.
  Fortunately there are dedicated rows for Alt, Ctrl(left), Shift(left), AltGr, Ctrl(right), Shift(right). This defuses the "ghost-key" problem.
• Let's count together what we have: 2 buttons for 1 or 2 player selection, 1 button to insert coins, 3 buttons - oh no, better 4 for each of the two players (so that we can play asteroids), and two joystick units with 4 switches each.
  Makes 19 switches in total !
  Here's my proposal for connecting the switches as new arcade-console-matrix to the cherry keyboard controler:

• A few words about the "ghost-key" problem:
  If you press 3 buttons which are at the edges of a rectangle in the keyboardmatrix, for example the keys:
  <F8>, <F10> and <5> you can't detect with a matrix if the key <4> is pressed or not, so the keyboard controler generates a "ghost 4" and send it to the pc. And furthermore, if you have all keys at the edges pressed and you release one of the keys, nothing happens (no keyrelease-code would be send to the pc) cause the keyboard-controler in fact did'nt notice that one of the keys is released.
  So you have to be careful which points in the keyboardmatrix you will use for your controls. Such keys like <1>,<2> and <3> wont cause any problems because they are only used for "game selection" purposes. But when it comes to action, there should be no interference between the controls used.
  

  
  

  Updated:

• How to eliminate the ghost-key problem (thanks to George for the hint):
  If you connect a standard diode in serial to each button, the ghost key problem vanishes from the electrical point of view. Check out, in which "direction" the diode could be inserted at your gutted keyboard (it depends on whether the keyboard manufacturer uses "running zeroes" or "running ones" to scan the matrix):
  

  "running zeroes"	"running ones"

  
  
  I haven't checked it out 'til now, but it should work in theory.
  

  
  

OK. that's it for now. I'll make some pictures when the "mechanical" part is ready. (It's still not ready ;-( )

Stephan

Project # 5: SPINNER for superduperarcadeconsole

Hi folks, my spinner device is ready:

• It's very reliable! The heart of the spinner is the 52mm steel-axis with 13mm and 11mm ball bearings. Then I've two wooden disks (35mm diameter, 10mm thick). At one disk I attached a stroboskop pattern which is then be scanned with two opto-reflex sensors (CNY70). The little board with the opto-sensors is assembled (with hot-glue) to the bottom of the second disk. At the other side of the second disk, there is a 5mm deep 13mm hole with a 11.5mm core-hole, so the steel-axis fits exactly into the wooden disk. The electronic uses two opamps and 2 CMOS gates and builds the interface to the LPT-port from the PC, where I use SNESKey from Benji York as driver. That's all.
  Btw, the electronic is supplyed by the printer port, where I use one data pin for the core-electronics and two data pins (via diodes) to supply the two infrared-LEDs from the reflex-sensors (which I hooked in serial with a 100 Ohm resistor).
  
  It should be possible to use the mechanics and the opto-reflex sensors together with a gutted mouse; then the -circuit- should not be needed.
  

  From left to right: knob, steel axis, wooden disk with sensor board assembled, wooden disk with stroboskob pattern, LPT-port interface (for use with SNESKey)	Sensor board assembled on the wooden disk. the black squares are the opto-reflex sensors (CNY70)
  	here the 13mm hole with the 11.5mm core to pick up the steel-axis (with its 13mm*5mm upper ball bearing) is visible

Project # 6: universal spinner interface for keyboard, joystick or gamepad

Hello arcade freaks, my universal spinner interface is ready now:

• Some of you mentioned that it'll be a good thing if there were a interface that allows the use of the spinner in an arcade console which uses a gutted keyboard as core, or with a standard joystick or even a gamepad.
  
  I came to the same idea, because I've been running into some problems with the actual SNESKey version.
  
  Now it was nescessary to develop a circuit that does the following things:
  
  1. detect in which direction the spinner spins
  2. generate pulses when the spinner spins
  3. trigger two "switches" which can act as keys (in case of a keyboard-core) or buttons when used with joystick or gamepad

  First an introduction into the gray code which is usually used for rotary devices:
  
  The signals A and B represent the signals which were delivered by the photo-transitors (after amplification)
  
  

  A B Rotate Left: 0 0 0 1 1 1 1 0 0 0 0 1 1 1 1 0 0 0 Rotate Right: 0 0 1 0 1 1 0 1 0 0 1 0 1 1 0 1 0 0

  Here you can see the "gray code" life, while the spinner is spun by a drill.

  
  It is possible to change the motion direction whenever you want. And you can see that you can always determine the direction of movement if you know the last state! So it's nescessary to store this state in some way.
  
  Now for something completely different, the larch.... oh no, the verbal circuit-description:

First stage of the circuit is a 4 times OpAmp (TL084) which amplifies the signals from the photo-reflex sensors (signals A and B). Two of the OpAmps just act as inverters to get the inverted signals A* and B*.

The second stage of the circuit consists of 4 D-flip-flops (2* 4013). There the "direction" is going to be stored. The flip-flops are triggered (positive edge) by the signals A, B, A* and B* while the D-inputs are connected to B*, A, B and A* so you always have for the right direction the Q-output set (QA, QB, QA*, QB*) and for the left direction the Q*-output (Q*A, Q*B, Q*A*, Q*B*). Btw, all the "set" inputs are tied together and all the "reset" inputs are tied together, they will be connected to the outputs of the fourth stage.

The third stage:
With four monoflops (2* 4098 or 4528 or 4538) a pulse signal for each possible transition in the gray-code is generated. So we have a Ap, A*p, Bp and B*p pulse. I've designed the pulse duration time by 17ms. In fact I used a 47nF capacity and a 1.8 MOhm resistor.

The fourth stage:
The pulse outputs are combined via 8 AND gates (2* 4081) to the direction outputs of the flip-flops. (Ap AND QA, A*p AND QA*, Bp AND QB, B*p AND QB* for the "right" direction and Ap AND Q*A, A*p AND Q*A*, Bp AND Q*B, B*p AND Q*B* for the "left" direction).
Each output for one direction is tied via a diode (Anode at output) to a single 10 kOhm resistor which is then tied to ground (2* (4 diodes, 1 resistor)). In fact these diodes simple simulate two OR gates with 4 inputs.
Let me name the outputs of this stage as "right-pulse" and "left-pulse".

The fifth stage:
Is just a CMOS switch (4066 or 4016, only two of the 4 available switches are used) where the steering inputs are connected to right-pulse and left-pulse.
(In the prototype shown above, I use the remaining switches to drive two LEDs, so I can survey the proper function of the circuit)

Oh, not to forget: the "set" inputs from the D-flip-flops are connected to "right-pulse" and the "reset" inputs to "left-pulse" too.

Now the outputs of the two CMOS-switches could be used to shorten a keyboard-key or a joystick-button or a gamepad-pad....
Note that the ground and the supply voltage level (+5 V) of the interface should be the same as for the driven device (otherwise you can damage the switches within the 4066). The easiest way to achieve this, is to supply the interface by the keyboard or joystick port.

I think the circuit is nothing for beginners ! When I write the sheet down to paper, I'll scan it and put it at my site. Maybe someone will find the time to "write it down" with a CAD-program.

Update 02.04.98: click here for the scanned sheet.

Very experienced electronic freaks should be able to build the circuit following to the verbal description (availability of CMOS-databook assumed).

I just checked out the interface and connected the circuit to my (second) cherry keyboard controller.
IT REALLY ROCKS !
I tried out ANNOID (the arkanoid emulator) and it worked "like a charm".

I'm ready now for arkanoid within mame v0.31,
hope the mame team will be ready soon !
(really can't wait any longer.....)