INTRODUCTION:

This procedure explains how to hack apart a keyboard for use with MAME.

First, I have not completed this myself, but I have done all except the soldering so it should work.

Second, more than one person has fried their motherboard from a bad soldering job on a keyboard hack. If you are not comfortable, soldering on a printed circuit board, find someone who is. Use this procedure at your own risk and preferably test the controller on an older computer before plugging it into your brand new Mega-GHz PC.

Third, in deciding whether to use this method or a keyboard encoder, the most important considerations are in order: cost, available time, soldering skill, and number of gameplay buttons required. A typical keyboard hack will support two joysticks and between 8 and 14 total gameplay buttons. An almost unlimited (up to 104-keys) total inputs are supported, but these may have ghosting or blocking problems.

Finally, a keyboard hack is a very intensive solution for very limited results.   Yes, it can be virtually free.  Yes, it may provide enough inputs for most games in MAME for an arcade cab.  Yes, MAME (and many other emulators) can get past the limitation of not being able to reprogram your controller to match it's default inputs.   Having said that, I have come to the conclusion that a dedicated encoder such as the KeyWiz, I-PAC, or MK64 will be a better solution for the vast majority of arcade control users.  At the risk of sounding like RandyT's (KeyWiz) posterboy, I don't think I could summarize my thoughts on this better than he did in a recent BYOAC post.   

"I have a hard time recommending a keyboard hack to anyone who isn't chronically unemployed and to the point where every penny counts, or extremely bored and looking for a personal conquest.  It's a lot of work, it's easy for a novice to ruin parts (like keyboard fuses on motherboards), and the end result is usually pretty sub-standard.  The keyboard I hacked was noticeably poorer in performance than a dedicated encoder . . . I suggest you do a lot of reading before attempting this and don't base your decision on the successes or failures of a couple of individuals.  This is going to require a certain amount of introspection on your part, after educating yourself, to decide if you are "up to the job".  If you haven't soldered before, you'll need to practice.  If you don't own a multi-meter, you should get one.  It'll help tremendously.

For some, the task is trivial, but time consuming.  For others, it is impossible and cost them a motherboard to find out.  You should probably try to figure out which group you'll end up in before you get too far into it."

So now, if you're ready and still want to pursue this, let's get started.

GENERAL NOTES ON KEYBOARD CONSTRUCTION

The keyboard case is generally assembled in two halves.  The halves are held together by (usually Phillips-head) screws. In some cases, plastic retainer clips are also used to hold the halves together.  The keyboards that we are interested in will have a circuit card that the cable plugs into and one or more Mylar overlays that connect to this card and register keypresses.  In some cases, a metal or plastic plate is screwed to the upper half of the keyboard to hold the Mylar against the keys.

The Mylar overlay will generally have carbon or copper contacts in a "double C" configuration like this:

One lead of this circuit (the left trace above) goes to a row of the matrix and the other lead goes to a column.  The contacts may also be in the form of a split circle.   On some keyboards, there are three sheets of Mylar.  The upper sheet has conductive contacts on its lower surface and the lower sheet has contacts on its upper surface. A Mylar sheet without any contacts but with holes along the contact points is placed between the two sheets to ensure that only the selected keypresses register.

The Mylar overlay is generally clamped to a header terminal strip on the circuit card.   On some keyboards, the Mylar overlay has terminal strips that correspond to the terminal strips on the keyboard circuit card and screws hold the Mylar to the card and ensure contact. This connection method would probably be easier to solder to.

WHAT TO EXPECT:

Using this procedure, you should end up with a controller capable of handling two joysticks and six buttons per player with no ghosting during game play.  You may have to go through the entire procedure on a couple of different keyboards before you find a useable one, though.  Once you find one that works, it should also be able to support 3-player, 3-button games, or 4-player, 2-button games as well.

TOOLS/MATERIALS REQUIRED:

1. Analog or digital multi-meter with resistance (ohms, W) scale ($10.99 at Pep Boys).

2. Soldering Iron, preferably low wattage.

3. Office style correction tape, single line.

4. Rosin-core solder, preferably the small-gauge electronics type.

5. Some type of wiring. I recommend IDE-type ribbon cable, available at Computer Stores (or surplus from old computer hard drive cables), Radio Shack, or Digi-Key (Digi-key P/N R007-5-ND Multicolor $5.09 for 5 feet, P/N R027-5-ND Gray $3.51 for 5 feet).   I used 40-conductor cable because I will also be connecting my keyboard hack to removable control panels and to an optical controller board and 40-pin cable and 40-pin connectors are the easiest to find.  If you are just connecting the keyboard hack to a single arcade controller, you could probably go with 24-conductor cable, which is a few pennies cheaper and will make a cleaner installation. You also can peel off the extra conductors if they are not required.

BTW, I wanted solid rather than stranded conductor wire for my project, because I thought it would be easier to work with. Digi-Key does not carry this.  They gave me the 800 number to 3M.  3M only sells it in 200-foot rolls for a lot of money. However, they said the local distributor might be able to get me a sample of it to try.   I called the distributor and he sent me a five-foot sample, Priority Mail, for FREE. I don't know if this will work for you, but it's worth a shot.

ALTERNATIVES (AND THEIR PROS AND CONS):

Okay, if you read through this procedure and have decided that hacking a keyboard is not for you there are two other options:  a gamepad hack, or a pre-built encoder.   Detailed information on either option is available here.  Also, I will include some brief comments below.

KEYBOARD HACK ADVANTAGES:

1. Cost - Forget about wire and terminal blocks because you would have to buy these either way (Well okay, no TB's if you use a dedicated encoder and don't have removable control panels).  If you have basic electrical tools, your total cost is the keyboard, which you can often find for less than $5.00 with a rebate or for free (if the keys, case, etc. is broken) at computer surplus stores.  In a worst case, at $15.00 for a keyboard, $10.00 for a multimeter, and $10.00 for a soldering iron, the total cost of $35.00 is still less than any other option, and the cost of doing a second one is only $5 or free.  However, if you are only building one project, haven't soldered, and don't own a multimeter, you are probably much better off buying a dedicated encoder.

2. Number of total inputs - As previously stated, a keyboard hack will allow a total of 101 (or 104) separate inputs, however only probably 16 of these can be pressed simultaneously. This is more total inputs than any other option (and more than you would ever need).

3. Cost - This is the cheapest way possible to connect arcade controls to your PC.

4. Feeling of Accomplishment - It's pretty fun knowing that you created something yourself and it does what it was intended to and works better than anyone says it should.  Especially when most everyone you ask for advice says "Save yourself the headaches.  Buy a dedicated encoder."

5. Cost - Did I mention it is really cheap? If not, see items 1 and 3 above.

6. Size - In some cases, the circuit cards on newer keyboards measure approximately 1.25 by 3 inches.  This is about half the size of most dedicated encoders and is smaller than any of them.  This may be an important consideration for a desktop controller with many buttons.

KEYBOARD HACK DISADVANTAGES (REAL AND PERCEIVED):

1. Soldering - Be sure to read the warning at the beginning of this page about frying your motherboard.  Actually, all soldering is done to the input side of the integrated circuit, so I think you would be more likely to end up with stuck keys or non-working keys from a bad soldering job.  Of course, if the solder drips onto the IC circuit board and shorts out a couple of pins. . .  Well let's just say that could be bad. Basically, if you don't know how to solder well, and I don't, you are probably better off letting someone else solder this, as I plan to do.

Tip: if a keyboard hack goes wrong, i.e., the 5 volt supply is grounded, the motherboard keyboard port will appear to have died instantly.  This is because most motherboards supply the 5 volts through a tiny fuse which blows rather easily. While expensive to professionally replace, the fuse can pretty safely be wired across, saving an apparently useless board.  Keep in mind that this now leaves the other areas of the motherboard vulnerable, so if the original problem is not corrected or another faulty keyboard hack is connected, more extensive damage may occur.  Also, many A-Open brand motherboards use a resettable circuit breaker in lieu of a fuse on the keyboard and USB ports.  While A-Open typically offers a good selection of features per dollar and their boards have been well-rated in most reviews, I would not recommend buying a motherboard solely because of this feature. However, if you like their products, this is something to consider and maybe something to look for in other manufacturer’s boards.

2. Time - For me, I would rather spend time on a project than money. Those that disagree, should buy one of the alternative products shown above.   I would estimate the time required for this project as follows: disassembly - 0.2 hours, mapping out the matrix - 2.0 hours, reassembly - 0.1 hours, planning the input keys - 1.0 hours, testing the proposed input keys - 0.5 hours, second disassembly - 0.1 hours, soldering (rough guess, from here on I have not accomplished this) 1.5 hours, mounting - 0.5 hours, miscellaneous and sundry distractions - 1.1 hours. Total time: 7.0 hours (if my calculator is correct).  If you happen to find or purchase one of the keyboards mapped here, you can shave most of the key mapping and planning stages, so the total time should be about 3.0 hours.

3. Number of Independent Inputs – As explained elsewhere a keyboard hack will support from 12 to 18 independent inputs. This is sufficient for two joysticks and between eight and 14 total buttons.  The Player Start, Coin Input, and other maintenance keys are not included and may in fact have blocking problems.  Normally the likelihood of this happening is rare as most games do not allow the 1P or 2P Start keys to be depressed during gameplay. (It can be made even more rare by assigning the Coin 1, 1P Start, and 2P start keys to keys not in the same matrix column as any of the joystick direction keys or the B1 or B2 buttons for each player, but this added complexity is probably not required, unless you cannot EVER tolerate ghosting/blocking problems.) If you require additional buttons for gameplay (highly unlikely) or if you cannot tolerate even the possibility of ghosting with the coin or start keys, you should not use a keyboard hack.

Also, Andy Warne ( I-PAC) pointed out and Randy T (KeyWiz) confirmed that most current keyboards employ two blocking methods.  There is the key blocking routine mentioned below, but there is usually a pressed key maximum hard-coded as well which varies but is usually eight.  This is done to reduce the amount of RAM required in the keyboard microcontroller.  I have no doubts that this exists, but both the keyboards that I have mapped have been able to register at least 15 simultaneous keypresses (I ran out of fingers on the PC Concepts verification).  Unfortunately, there is no easy way to tell and no easy solution if your keyboard has this limitation.   You basically have to map the matrix, test the keys, and if keys stop registering either limit yourself to that many keys, or start over with a different keyboard.

4. Ghosting, Masking, and Key Blocking - This is probably the single, strongest reason why people avoid the keyboard hack.  There is also a lot of outdated or plain wrong information about this on the web.  It is so important that I will spend most of the next section (Common Misconceptions) covering the technical aspects of it in detail. A short summary of the problem and solutions is included here:

1. What happens - In a nutshell, newer keyboards work by scanning a matrix. See Appendix A for example keyboard matrices. When three keys that form the corners of any rectangle anywhere on the matrix are depressed, one of two things happen, either the keyboard will assume that the key marking the fourth corner of the rectangle is also depressed and send this key to the computer (key ghosting), or the third key will not register (key blocking). In addition, if ghosting occurs, if the fourth key of the rectangle is depressed simultaneously and one of the other keys is released, the key release will not register (key masking).

2. How serious is this, really - That is a tough question.   Keep in mind that this condition only occurs when three or more keys are pressed simultaneously.   I played Asteroids for about six months before I read a review of the Stick-It on Saint's BYOAC site saying that Windows popped up when he pressed alt (thrust) and space (hyperspace) at the same time.   I tried it at home and it instantly did the same thing, although I had never noticed it before.  Also, consider that a two-joystick game with both joysticks on the diagonals generates four simultaneous keypresses without any other buttons being depressed.  The trick is to pick your keys carefully and avoid the problem altogether. BTW, the asteroids problem was a windows shortcut problem and not key masking, and was easily solved.

3. Why is there so much confusion: Basically, and I am guessing here, about 1989, keyboard chip manufacturers became aware of the ghosting problem, decided that having uncommanded keys show up was a bad idea for word processing and went to the blocking route.  At about the same time, MAME took off and people using older keyboards discovered that having uncommanded keys show up for emulators was a bad idea.  They recommended the use of diodes to solve the problem, which worked for some of the key ghosting keyboards.  Other older key ghosting keyboards used very low voltages for the switch scanning and thus were unaffected by diodes.  Later, people bought newer keyboards and found that diodes didn't make any difference and they were still having (now blocking) problems.  A huge debate ensued and is still raging about whether diodes make a difference, and when or if they should be used.  Hopefully, I will not respark that here.

4. What are the solutions: Ironically, the solutions end up being identical, at least as far as I am concerned.  If you have one of the older keyboards that actually demonstrates ghosting instead of key-mapping (and I don't think you will find a new keyboard which does), you could (theoretically) attach diodes to each input line and be able to use all 101-keys without ghosting (assuming the keyboard does not limit simultaneous inputs in firmware).  However, this doubles the amount of soldering required, and increases the possibilities of a solder connection coming loose and messing up your computer.  Additionally, if this works, I would expect Hagstrom to be aware of this solution and incorporate diodes in their encoders, and the fact that many people are posting suggested key assignments to use with Hagstrom encoders in matrix mode tells me that they still have ghosting/masking.  If your keyboard uses key blocking, diodes will not help and may actually hurt things. However, in either case, if you choose your keys strategically, you will not have any problems with ghosting or masking, regardless of keyboard type.

5. Lack of a keyboard when using the controller (no keyboard pass-thru support).  Some people mention this as a problem with keyboard hacks.   Actually there are several solutions.  Before I get to them, consider the following:

1. If you are using your controller in a MAME arcade cabinet, you shouldn’t need a keyboard at all except for set-up and configuration changes, or for adding new games or new MAME releases.  By designing the cabinet for this, it should be fairly easy to shut down the computer, unplug the keyboard hack, plug in the standard keyboard, make the updates, and plug the keyboard hack back in.  Or you could use a wireless keyboard for maintenance.

2. Now, in order of my personal preference, here are the solutions to the no-real keyboard problem:

   1. USB - if your computer supports USB, purchase a USB keyboard for general use and plug the keyboard hack into the regular keyboard port. You should be able to find a USB keyboard for around $15.00 or less. Under NO circumstances should you use a standard keyboard for general use and hack a USB keyboard for MAME, because the USB specification does not allow more than six simultaneous key presses (plus modifier keys). This does not apply to the I-PAC in USB mode. (Thanks to Andy Warne for this tip.)

   2. Keyboard splitter circuit: Saint's BYOAC site has links to two keyboard splitter circuits: Stephan Hans, and one from everyday practical electronics.   Either one could be built for under $10.00. The following information regarding these circuits was provided by Bill Lash in response to questions I posted on the BYOAC message board and should be helpful for someone considering these options:

1. What are the differences - Stephan's circuit is set-up so that one keyboard is primary and the other is secondary.  The LED's on the primary circuit will always work and on the secondary one they never will. If you are building a MAME Cabinet with LED's on the panel, plug your keyboard hack into the primary port and the computer keyboard into the secondary.  If you are building the controller for a general use computer that also plays MAME, plug you computer keyboard into the primary port and your keyboard hack into the secondary one.  The other circuit is easier to build and costs a few dollars less to build.  It will activate the LED's on the keyboard or hack which last sent an input to the computer. Thus, your LED's in MAME will not work until you press a button and your LED's on the computer keyboard will then be off until you press a key on it.

2. In both circuits, the keyboard port pins are numbered looking at the cable side in on the receptacle (looking at it the way the cable would be inserted). Also notice that the simpler circuit shows a plug rather than a receptacle for the computer connection.

3. Integrated circuits (IC's) are numbered as follows:   Pin 1 is indicated by a dimple or U-shaped indentation on the side of the chip that contains Pin 1 (or by markings on the IC). Looking down on the chip, Pin 1 would be on the upper left corner and pin numbering continues CCW down the edges and around the chip, so that the highest pin number ends up at the upper right corner opposite Pin 1.

4. The simpler circuit shows a 4093 IC in two different places on the circuit schematic. Actually, only one 4093 IC is required.

5. The simpler circuit implies that +5V and ground are coming in on a external wire into the circuit. In fact, these inputs are supplied through the keyboard port.

3. Keyboard switch: Lew's wheels includes a circuit using a SPDT (single pole dual throw) switch and three sockets to switch between two keyboards.

4. The simplest solution is to hot-swap between your computer keyboard and your keyboard hack.  I asked Andy Warne about this and he said that theoretically there is a danger in this, but he had done it at least a hundred times setting up the I-PAC and never had any problems.  I also had tried it several times with no problems before I asked him.  Digi-Key (among others) sells an M-F keyboard extension cable which will allow you to do this without crawling around behind the computer.

5. Y-Key-Key : Also on Saint's BYOAC site, PI Engineering sells this adapter for about $50.00 which will allow you to have two keyboards connected to your computer.  However, as far as I can tell, the device operates identically to Stephan Hans circuit and the price will eat up any of your savings by going the keyboard hack route.

6. Micro-center (among others) sells a cable splitter (for $7.95) for laptop computers which allows a keyboard and mouse to be hooked up to the same port.  The cable has three connectors, a female PS/2 and two male PS/2 's which are labeled keyboard and mouse.  There were no instructions as to whether it was intended to plug into the PC keyboard port or the PC mouse port.  I tried it in various configurations with two keyboards, couldn't get it to work, and returned it.   There might be a way to get this to work, though.  If someone finds one, let me know and I will include the information here.  This is a similar device which is designed for connecting two keyboards, but I have not tried it and don't know if it works or not.

On the I-PAC or MK64, if you press the shift key nothing happens until you release it.   Then it either sends its own code (1) or (if you have pressed another key) the shift code.  If you map the same combinations in MAME, you will always get the first key send followed by the combination key send.  The effect of this for each shift key function is as follows:

1. Coin Input: The I-PAC uses 1 and P1B1 keys to mimic coin input.  In MAME, from the attract screen, if you map this combination, the first coin input will register.  If you try to input a second coin, the game will start a 1-player game, because MAME reads the 1 key first.  Therefore, you will never be able to start a two player game.  For this reason, I recommend having a dedicated coin input key on your panel if using a keyboard hack.  (You CAN get around this from the attract screen for most games by pressing the P1B1 key FIRST, followed by the 1 key; but a dedicated key is the easier solution).

2. ESC: The I-PAC uses the 1 and 2 player start keys to mimic ESC.  In MAME, the game will either quit immediately or will start a one or two player game and then immediately quit.  This is perfectly acceptable and I recommend mapping these keys if you don’t want a dedicated escape key.   However, a drawback to both of these methods is that you might exit your game if both players press the start buttons at the same time.

3. Pause: The I-PAC uses the 1 and J1 Down key to mimic Pause.  In MAME, if you are on the attract screen with credits already input, MAME will start a new game before pausing the game (assuming you press the 1 key first).   If you are in one of the game playing screens MAME will pause properly unless you press the J1 Down key first, in which case the player position will move down slightly before the game pauses. This is acceptable, although I personally prefer a dedicated Pause key so that I don’t have to remember key combinations when I need to stop gameplay at a critical point.

4. Tilde and Tab: The I-PAC uses the 1 and J1 Up or J1 Right keys, respectively, to map these functions.  These keys will function the same way as the Pause key above.  See the section on assigning keys for my personal key mapping recommendations on these keys.

5. Enter: The I-PAC uses 1 and J1 LEFT to mimic this key.  This will work in MAME.   This key is only used in MAME for items on the Configuration or On Screen Display Menus.  The only difference between using this key in MAME is that if you enter these menus without pausing the game, the player may move left while you are hitting enter!  You are not likely to do this, though.  Again, see the key mapping section for my personal recommendations on this key.

LITTLE KNOWN FACTS AND COMMON MISCONCEPTIONS:

1. Matrix Mode:

Most keyboards use some kind of matrix to interface with the keyboard IC.  This keeps the manufacturer from having to run 105 inputs (104 keys + common) to the controller.  Instead, only 25-30 inputs are used, by assigning the inputs to rows and columns and assigning keys to each combination on the matrix, all of the keys are supported.  For some reason, 24 input matrices are almost the industry standard.   For example, the DR-104 keyboard has 26 terminals connected to the IC.   However, terminals 17 and 18 are used to support the Windows 95 menu keys only.   Since these keys are not recognized by early versions of MAME and most all other programs, this results in an effective matrix of 16 x 8 = (you guessed it) 24 keys.

Ideally, one would like to find a matrix with 22 x 2 (rows by columns).  However, this is impossible since such a matrix would only support 44 total keys.  The best keyboard we could hope to find would be an 18 x 6 matrix (108 keys theoretically supported).  This keyboard would support two joysticks and 7 buttons per player (or two joysticks, six buttons per player and the player 1 and 2 start keys), with no chance of ghosting.  If someone finds a keyboard with this configuration, let me know and I will add the information to this document.  The worst option would be a 12 x 12 matrix.  This will only support two joysticks and 4 buttons per player (or one joystick, six buttons, the 1P, Coin Input, Pause, and ESC keys) with no chance of ghosting.  As you can see, even this is not a bad option, unless you really like the Mortal Kombat / Street Fighter style of games.

2. Ghosting, Masking, and Blocking:

Dave Dribin does a much better job of explaining the technical side of key ghosting and masking than I could ever hope to.   Stephan Hans also has some excellent info on his site.  Follow the links to SuperDuperArcadeConsole (using Cherry Core).  Stephan's page doesn't provide much detail on how you pick the keys to avoid ghosting (that's my job), but he uses the same concept that I am and his site was the first place I ever saw this mentioned.  I think his information about diodes is incorrect, though :-((.  Thanks to Saint for the links.  I will try to cover the practical side of the problem here:

1. Key Ghosting - Most keyboards use a matrix to determine key presses. The DR-104 uses a 16 by 8 matrix. To show a random real example configuration:

Ghosting may occur when you press any three keys that form a rectangle (any number of rows and columns apart) in the matrix. In the example above, pressing E, Q, and C, simultaneously will generate X. This is rarely a problem as the X key, for example is not usually used by another game input and all keys need to be pressed simultaneously.

3. Diodes

There has been much heated debate on the use of diodes in a keyboard hack.  It has been very ugly at times.  I don't want to rehash that here.  Diodes may or may not work on older keyboards which show key ghosting.  They won't work on keyboards that use key blocking.  They are not required on either if you pick your input keys carefully.  See the FAQ section for more info on diodes.

HOW GOOD ARE THE RESULTS:

Since the DR-104 keyboard uses an 18x8 matrix, I should theoretically have eighteen keys available with no blocking.  In fact, this and many other keyboards put the right and left Windows Menu keys all alone on two rows and since earlier versions of MAME do not recognize these keys, only 16 keys are available.  However, since a joystick cannot be both up and down or both right and left at the same time, these keys can be allowed to block and only two inputs per joystick are required.  Thus, my method allows two joysticks and six buttons per player to operate with no chance of key blocking.   I still will get blocking if I press the 1P, 2P, coin, or pause keys with certain other keys depressed, but this should rarely create any problems.  I was even able to use the MAME default keys for the Player 1 inputs, although the joystick and Player 2 inputs are nothing like the MAME defaults.  Later, I was able to expand this concept, and the keyboard now also supports 3-player 3-button games and 4-player 2-button games.

HOW TO DO IT:

SECTION A - USING ANY TYPE OF KEYBOARD.

If you have one of the example keyboards for which the matrix has already been mapped, you can skip most of these steps. See SECTION B for the procedures to follow: During these procedures, I will use the DR-104 keyboard in developing the examples, however, the general procedures should apply to any keyboard.

1. Tearing it down:

1. Place the unplugged keyboard to be hacked face down on the desktop and remove the screws holding the keyboard halves together.  (Generally, eight to ten philips-head screws.  Some of the screws may be under labels or some keyboards use plastic retaining clips to hold the halves together.  If the halves don't separate easily, look for this). Do not lose the screws as you will need them again later.

2. Carefully, remove the back cover from the keyboard.   Depending on the design, this may leave all of the little key buttons and rubber pads which go under them loose.  Do not lose these either.  You will see one of the following three things:

   1. There may be a bunch of metal plates inside the keyboard.   On older keyboards these were used for magnetic shielding.  I have never run across one of these, but one of the example websites said that they cannot be hacked.   However, the Mitsumi keyboard uses a metal plate to cover the Mylar membrane, so you might try removing the metal plates and seeing if you see a Mylar sheet. Otherwise, trash the keyboard, buy another one, and start over.

   2. The keyboard may have a huge circuit board which covers all of the keys and little mechanical switches below each key.  These were typically used on the old IBM-type "clickety-click" keyboards.  Many people have reported that these keyboards did not use a matrix and all keys could be pressed simultaneously. Andy Warne reported that these keyboards do use a matrix and therefore, I would expect some type of ghosting/masking behavior.  All I can recommend is that before you start, you check for the presence of ghosting/blocking using one of the keyscanning programs as shown in Step 5.  If ghosting/blocking is not present, you connect the keyboard by just soldering two wires to the back of the circuit board opposite the mechanical switches for the keys you want your buttons to use and running these wires to the appropriate switches.  In some cases, people have reassembled the keyboard after they finished and still been able to use it as a regular keyboard and as an arcade control!

   3. The keyboard may have a circuit card in one of the upper corners (follow the input cable) which connects to a Mylar sheet(s) which records the key presses.  The Mylar sheet may be retained under a plastic or metal cover plate. Remove the plate so that the Mylar sheet and circuit board are exposed.

2. Mapping the Matrix:

An alternative method which many people recommend involves disassembling the keyboard and plugging the circuit card into the keyboard port and shorting between the terminals and recording the generated keystrokes to map the matrix. I recommend my method over this one for the following reasons: 1)  In my method, the keyboard is unplugged so there is no possibility of damage to the computer system during testing.  2)  It's not always easy to tell where the connection points are and which ones are rows and columns.  I'm not sure what happens if you connect two row inputs together, but it might not be good.  3)  Many inputs (L Alt, Windows Key, etc.) won't display on a screen like NotePad.  (I realize GhostKey or some of the other utilities will work; however, even some of the utilities can't pick up the NumPad keys, or the difference between R Alt and L Alt.  And if it doesn't detect these keys, they will be left out of the final matrix.  With my method, you are using the physical locations on the keyboard, so every key gets picked up, and no effort is wasted).   4)  A 16x8 matrix is 128 inputs, but MAME can only use 104, so you have 24 unused and unusable inputs that you still need to map out.  (Add another 36 to that for an 18x8 matrix). 5)  If the keyboard ends up being a poor choice to hack, with my method you can reassemble it and either use it as a regular keyboard or yard sale it.  With the other method, you have spent a couple of hours soldering wires to a keyboard encoder that you will never use.

I recommend using either a spreadsheet program or a word processing program to map out the matrix, but it can be done on graph paper as well.

1. Pick one terminal of the circuit card to be terminal one.   Label the first line of your spreadsheet or word processing file as "Terminal one is right edge of card (connector side of board) looking at component side of board." or similar wording.  You may also use correction tape to label terminal 1 on the Mylar sheet so that you don't get confused as you start mapping out the matrix.   This ensures that you will still know how to wire everything up when your joysticks and buttons arrive in three weeks later.

2. Select the top row and the first column of your table or spreadsheet to bold. This will make it easy to distinguish between row 5 and an input for key 5 on the keyboard.

3. On many keyboards, the Mylar sheet(s) will have more contacts than the actual number of keyswitches on the keyboard.  This usually occurs around the shift or bottom row keys, or around the enter key.  I think it occurs so that the keyboard manufacturer may use the same Mylar with L-shaped or straight Enter keys, etc.  These contacts are not required for the keyboard being tested and might be confusing. Cover the contacts with correction tape on the conductive side of the Mylar to ensure they are not mapped.

4. Lay the Mylar sheet (and possibly the circuit card also) flat on the desktop as follows:

   1. If the Mylar sheet(s) and circuit card are connected (clamped) together, remove them as an assembly and lay them flat on the desktop with the conductive side up. (The same orientation they would normally be in when you actually use the keyboard.)

   2. If the Mylar sheet is held down by the screws holding the circuit card to the keyboard, remove and retain the screws and place only the Mylar sheet on the desktop with the conductive side up. (The same orientation they would normally be in when you actually use the keyboard.)  In this configuration, (looking at the disassembled upside-down keyboard), the circuit card is usually mounted with the component side and the contacts pointing up and the Mylar sheet is placed on top of the terminals with its contacts face down.  Thus if you label the circuit card terminals from 1-26 starting at the left, you must label the Mylar sheet from 1-26 starting at the RIGHT when the sheet is flipped over.

5. You will now have a Mylar sheet with contact points corresponding to the key positions on the keyboard under test. These should be similar to your standard keyboard so you may use it to determine the key associated with each contact point. The contacts will look similar to the following

"double-C" configuration, or possibly a split circle configuration. You will want to check for continuity between each leg of the C and the traces leading to the circuit card as follows:

1. Turn the Ohmmeter (multimeter) to the highest resistance scale.

2. Touch the positive and negative ohmmeter leads together. You should see full scale deflection of the needle reading. (A digital Ohmmeter should read 0.0).

3. Touch one lead to each side of one leg of the double-C trace as shown by the red dots below. You should see full scale deflection.  This will ensure that you are testing on the conductive side of the Mylar.

Keep one lead on one leg of the Mylar contact and touch the other lead to the traces leading to the circuit card, in order, until you see full-scale deflection.  Repeat for the other leg of the contact. For example (on the DR-104 keyboard), for L CTRL, you would see full scale deflection on pins 4 and 25.

6. Test all keys and complete the matrix as follows: Place the lower number terminal on the row and the higher on the column. Insert rows and columns in order as necessary. (I usually just work left to right and up the keyboard, but it doesn't really matter. These examples are from the DR-104 Key keyboard).

1. For the Left Ctrl key (contacts 4 and 25):

2. For the L Win95 key (contacts 17 and 20):

3. For the L Alt Key (contacts 10 and 22):

4. For the Space Bar (contacts 12 and 22):

5. For the R Alt Key (contacts 10 and 24):

6. And so forth. . . Continue this process until all of the keys are mapped. When complete, add rows or columns for any unused circuit traces. Usually there won't be any.

7. You should now have a table similar to the matrix in Appendix A.  The only other possibility is that if you chose the wrong end of the card for terminal 1, your table may have more columns than rows. In that case, you can either use the transpose function of your spreadsheet or word processor (save and print the current configuration first), or you can leave it as it is and just substitute column for row in the rest of this document.

3. Sizing things up:

• The Power, Wake, Sleep, and Windows Multi-Media keys all fall in the category above, but MAME can't use them, so there is not much point even mapping them.

• While I recommend avoiding these keys even for dedicated encoders, the problem is orders of magnitude worse with a keyboard hack.  Dedicated encoders have been optimized for speed and throughput, because they expect to be used with arcade games.  Keyboards are designed for typing in a word processor.

• There is a catch-22 with the arrow keys.  While MAME itself can be easily programmed to get around using them, many Windows programs and probably even some front-ends for MAME cannot.  However, the HotRod is non-programmable, and it does not use them.  Furthermore, the odds of a keyboard having the Up and Down arrows on one row and also having the Left and Right arrows on the same row but different from the Up and Down row is pretty slim, so you most likely can't use them even if you wanted to.

• In some cases, you may have no option other than to use these keys.  (For example, a matrix row does not contain any keys that aren’t on the “Avoid” list.)  This is acceptable, the keys will still work, they just slow down the processing, so their use should be minimized.

• The windows GUI keys deserve special consideration.  These were not useable in older versions of MAME and are on the "Avoid" list above.  I did not use them in my example, because when I wrote this up, MAME could not use them.  I did not add them in when I revised the page because the 16x8 matrix without them was representative of most keyboards.  However, MAME from version 0.60 up will recognize them and I would gain two additional inputs by using them with the DR-104 keyboard.  OTOH, you can't just use the Windows Key in MAME because if you just set the FIRE button to L WIN (for example), MAME will fire a shot and freeze as the Start menu pops up, fire a shot and continue as the Start menu disappears, fire a shot and freeze . . . etc.  You can avoid this by installing either WinKey Killer 1.7 (homepage/download) or LogoLess.  Here is a comparison of the two programs (either one of which I can recommend):
  

WinKey Killer 1.7 - Very simple (double-click to run, double-click again to exit).  Transparent.  Can be loaded at startup by placing in a shortcut and adding to the Windows Start folder.  Kills Windows key instantly in all apps.  Has several advanced set-up options like Disable Ctrl+Esc, Disable WinKey+E to launch Internet Explorer, Disable Ctrl-Alt-Del, Hide from Ctrl-Alt-Del, Display message on shutdown, etc.

LogoLess - Runs with an Icon in the Systray.  This is the better program if you ever use the Windows key because it only eliminates it in windows beginning with "MAME" (user-specified), or in full-screen applications.

2. Map the non-gameplay keys. These keys will be allowed to block or mask each other so any matrix position is acceptable (unless you have additional rows available and can make them non-ghosting.)   Find the keys you want to use on the matrix and change the shading to yellow (for example) to highlight them.  (This prevents you from selecting these keys for gameplay and then finding out you need them for MAME functions).  (It's not a bad idea to highlight ALL the keys listed below, even if you don't plan to use them on your panel.  This will remind you to change their function in MAME if you decide to use them for action keys when you start assigning the action keys).  The following is a list of maintenance keys that I have seen on people's cabinets and my personal suggestions:

• 1 - For 1P Start (recommended)

• 5 - For Coin 1 (recommended)

• 2 - For 2P Start (recommended, unless you never play 2-player games)

• 6 - For Coin 2 (or map another button such as P1B6 to Coin 2 in MAME. Coin 1 will work for either player except for a few games (Gauntlet, etc)).

• 3, 4 - For 3P and 4P Start (not recommended unless doing a four player panel)

• 7, 8 - For Coin 3 and Coin 4 (same as above)

• P - For Pause (recommended)

• Esc - For Quit (acceptable or recommend mapping UI CANCEL = 1 AND 2 in MAME.)

• Tilde - MAME On Screen Display Menu (recommend using a keyboard to adjust this during setup)

• Tab - MAME Configuration Menu (same as above, however this key is required to bring up the joystick calibration menu. Many MAME games will not play correctly with an out-of-calibration gameport joystick, even though the game may not require it. I recommend having access to this key on any MAME computer which will have a device connected to the gameport. Personally, I recommend mapping CONFIG MENU = 1 AND 5 to provide access to this menu. Then pressing the 1P Start and Coin Input buttons will access this menu.)

• Enter - Used for accepting menu changes or for calibrating analog joysticks/pedals. (Recommend including this under the same circumstance as above. Many people have recommended setting one of the less used buttons (say P2B6) to Enter. This is acceptable or you can simply configure UI SELECT to use the same key as P2B6 is assigned to. The only drawback to either method is that if you are playing a game that requires P2B6 and you do not pause the game before entering the menu, the game may continue to carry out the P2B6 action while you calibrate, etc.!!).

• F12 - Snapshot (not recommended)

• F11 - Toggle speed display (not recommended)

• F10 - Toggle Speed throttling (not recommended)

• F9 - Adjust frameskip (not recommended)

• F8 - Adjust frameskip (not recommended)

• F7 - Load Save State (not recommended)

• F3 - Reset Game (not recommended)

• O and K - To get past "Press OK to continue." See this page for better ways around this. (Not Recommended)

3. If desired (not recommended), try to assign the joystick keys to the arrow keys.  This will only be possible if LEFT and RIGHT are on the same matrix row, and UP and DOWN are on the same row, but a different row from LEFT and RIGHT.  If this is the case, change the shading for these key to a contrasting color, say teal. Most likely, this will not be possible.
   
   If this is not possible try to assign the Joystick 1 keys to the numeric keypad arrow keys (2, 4, 6, and 8, especially 2 and 8 which are the down and up keys).  These are the HotRod and X-Arcade defaults.  If any of these keys can work, shade them in teal.  If you can select KP 2 and KP 8, then you can tell EmuLoader (or any frontend with HotRod or X-Arcade support) that you are using these controllers and use the P1 Joystick to scroll the games list.  Interestingly, 4 of the 5 keyboards tested allow this and the 5th one doesn't have either key even listed on the matrix.

4. Try to assign the default Player 1 Keys to buttons as follows:

   1. Locate the L CTRL key, if no key on this row is shaded teal, change the shading to aqua.  If L Ctrl is on the same row as a teal key, decide whether you would rather have the joystick keys remain or be able to use L Ctrl, and change the shading accordingly.  Do NOT leave both L Ctrl and the joystick keys shaded in the same row.

   2. Repeat for L Alt, Space, L Shift, Z, and X.   However, if a row already has a key selected, skip it and move on.

   3. You should have most of the Player 1 keys available and highlighted.  If any keys were skipped, select replacement keys from any remaining open rows.  At this point, you should have six rows with one aqua key in each one highlighted.

5. Assign the joystick inputs as follows:  Look through all of the rows with no teal or aqua shading.   Either look for logical assignments like Up = U, Down = D, Left = L, Right = R or look for keys which are in the same row and physically oriented like up and down or side by side, or don't worry about it and just pick two keys on each row.  Shade the P1 Joystick keys in teal and the P2 Joystick keys in green.  When you finish, you should have six rows with one aqua key in each one highlighted, two rows with two teal keys in each highlighted, and two rows with two green keys in each highlighted (ten rows used).  The  teal and green keys will represent:

• J1 UP and J1 DOWN

• J1 LEFT and J1 RIGHT

• J2 UP and J2 DOWN

• J2 LEFT and J2 RIGHT

6. Assign the remaining rows to P2 button inputs as follows:    In any row that does not have any teal or aqua or green shaded keys, pick one key and shade it lime for six of the rows.  It is better to pick letter keys than commas or periods (which will be hard to see on the MAME configuration menu), but it really doesn't matter.

7. Double-check your work. At this point, (for a dual joystick panel) you should have six rows with one aqua key in each one highlighted, two rows with two teal keys in each highlighted, two rows with two green keys in each highlighted, and six rows with one lime key in each one highlighted (16 total rows used).  If this is not the case, make corrections.  At this point, you now know which of the keys you will be using in MAME.

8. You still need to decide which keys will be assigned to which joystick motion or button, but this is basically personal preference as MAME will accept any choice.

What is the effect of this? The biggest thing is that if there are two keys that you absolutely must have for gameplay that are positioned on the same row, you probably can rearrange your matrix to use these keys.  A second advantage is that you may be able to pick up additional available keys using this method.  However, in the idealized method above, we only gained two keys.  In point of fact, most keyboards will have too many "open spaces" in the matrix to allow keys to be added.  In the "real world", you are probably better planning things out using the "one key per row except joysticks" method and only using this method to tweak the matrix if you absolutely want to have some keys on the same row available for gameplay.  For example, using this method with the PC Concepts keyboard matrix (see the Appendix), it would be possible to leave the joystick mapped to left and right and also add the L Alt key which is the MAME default for P1B2.  However, to do this, you would no longer be able to use the R, Z, X, C, Up, or Page Down keys for gameplay. Each of these keys would have to be moved to a different key on the same row, resulting in a lot of less desirable keys, such as the numeric keypad or function keys.

Here is an even more extreme example of this provided by RandyT (KeyWiz).

(Click image to enlarge)