Joystick interface for C-16, C-116 and Plus/4

Abstract

 In this document I'll explain some possibilities on interfacing standard 9-pin digital joysticks to the Plus/4 and also some practical tips.
 
 This is V1.1 of this document; some re-writing was done, and correcting a mistake in the theory that I made in the first version.

Table of contents

• Copyright
• Credits
• Review
• A simple interface
• A full featured interface
• Contact

Copyright

You can have this document in a zipfile together with all graphics here.

Credits

Review

C64

 The joysticks contain small micro switches, connected between GND and the appropriate direction line. When pushing the joystick towards a direction, a switch shorts the appropriate direction line to ground. The scanning of the state is realised in several different ways in different computers; in the C64, the direction lines are connected parallel with the keyboard scanning lines and can be queried with a simple procedure through the CIA1 ports. Since there are two joystick ports on the C64, scanning them independently needs them to be connected to different CIA input lines (as it is seen in the C64's schematics).

 The POTX and POTY lines are for querying the paddles (the popular game devices of the early 80's - does anyone still have such gadgets?!). The paddles are in fact potentiometers inside (variable resistors). One end of the potmeter is connected to +5v while the other to the POTX or POTY line (the paddles used to appear in pairs wired to one common joystick port plug). In the computer, these input lines go to a simple integrating analogue to digital converter (counter type). By measuring the time needed to charge a capacitor by the current going from +5v through the paddle potentiometer, one can deduct where the wheel is currently positioned. In most computers, the position is measured with 8-bit precision.
 

Plus/4

 Commodore decided to equip these computers with new joystick ports. (Just to mention, only the Serial,the Video and the RF connectors are identical to the ones found on other Commodore computers). For the tape and joystick ports, Commodore introduced Mini DIN plugs. The reason, well - I guess it's out of the scope of this document. They probably wanted to lower their prices, and thus decided to use cheaper connectors. Or they were in lack of free place on their new model (the Plus/4 is relatively small; there is simply no free place on the rear of the computer, even if the sockets are mini DIN's and the RF modulator appears on the left side). The rest is history; they also introduced their own joysticks with the computer, but seems like it didn't help much. Joysticks were always a problem ever since the Plus/4 existed.

 Speaking of the specialities of the joystick port, at first we notice that no POTX and POTY lines can be found. It's not possible to hook a paddle (as easily as to the C64) to the Plus/4. We find the usual GND, +5v and the direction lines. There is one more line which was not present on the C64s joystick port, this is the Select' line.

 Going deeper to the inners of the port, the Plus/4 hardware schematics tell us some differences. In the Plus/4, the joystick lines are also connected parallel to the keyboard scanning lines, the same way as it is found in the C64. However, both joysticks are connected parallel to the same keyport scanning lines (except fire buttons, whose keyport lines are unique for each ports). Thinking it over, it would mean, that we cannot differentiate between the two joysticks from the computers side, since they're both free to pull the same keyboard line in the same moment.

 Now, the problem is solved by the last line named Select' which is unique for each joystick ports. The Plus/4 joysticks direction switches must be connected between the appropriate direction line and Select' (instead of GND). In the scanning process, the software on the computer can make each Select' line to go low in a particular moment, independently from each other, thus making possible to scan the state of just one joystick instead of both. (One more advantage of this method is, that it's also possible to prevent the keyboard scanning process from getting false keyboard data by the joystick, just by inhibiting both Select' outputs, which was anyway not possible on C64).

 In a more deep point of view, the Select' line comes from a standard LS-TTL gate output from the computer, whose input is hardwired to one dataline of the processor data bus (D2 for joy1 and D1 for joy2). One would suspect a static, well programmable selector line in such a subject, but this design seems to rather result in a heuristic 'oscillation' on the Select' line (as the computers databus changes cycle to cycle). Still, it is suitable for this task, because of the way it is used for the selection. The answer is in the TED, the TEDs keyport input latch which is sampled in a certain minute.

 The TED's keyboard port is available at $FF08. When the processor writes to this register, the TED samples the actual states of the keyboard lines (this value can then be read out next, reading from $ff08), right after receiving the writing statement. If we want the state of one joystick, we need first to make the appropriate Select' line to go low, making the joystick able to pull the keyport lines to low. Because of the above mentioned wiring of the Select' lines in the Plus/4, this is only possible if a certain data is found on the data bus when (or rather, shortly before) this sampling happens. Luckily, the 6502 architecture of the processor helps that it can be done. For getting a joysticks state, one must write to $FF08 a value whose 1st or 2nd bit is 0. Since the writing statements (for example, an STA $FF08) 'last' transferred byte on the data bus is just the value to be written, in the right minute the lines of the databus will reflect the bits of the data written to $FF08 - thus driving low the respective D, thus Select' line. The TED samples its inputs before the next cycle, so when the program reads the sampled value from $FF08, the register will reflect the state of the input lines when the joystick selection happened.

 There is one more difference that I haven't realised for a long time; it is indeed a really stupid overlook anyway, I don't know how did I miss it since it is also in the above facts. The Select' lines come from a standard LS-TTL gate output (a totem pole output). This means, it is either GND or (something less than) +5v but nevertheless both levels are active. It would mean, that the joystick can indeed pull an input line high in some moments (to be precise, whenever it is moved and it is not selected). And this fact makes sure that the series diodes on the joystick input lines, seen in the schematics of the Plus/4 must indeed be present in _all Plus/4 (else, the joysticks, if some direction switches are shorted, would interfere with the keyboard scanning process). As a consequence, this makes sure that it doesn't hurt if one drives the inputs to active high, as I previously suspected - since the original design does it so.
 

How to...

• Seems like, it's not very neccessary, to put simply, to try hooking paddles to the Plus/4. You can do it however, but it's at least not supported by any software; and dealing with it needs a lot of extra electronics (not even such counter with capture option is available in the Plus/4s hardware). However, you can see an excellent design having Solders (Christian Schäffner) SID card that has a full featured C64-like joystick port on the rear.
• The different architectures may cause troubles. There is no problem with bare joysticks, because they simply act as a bunch of switches. The only difference is, the different common point. That's the way BTW that common C64 to Plus/4 joystick interfaces are wired. Troubles come in, when the joystick, or whatever external unit needs supply voltage for its operation.
• Because of this pretty dirty operation of the Plus/4 joystick port, one doesn't have to keep in mind not driving the input lines to high. Going to the joystick port directly from a a totem pole output is legal and violate no rules.
• Since the Select' lines change so fast, the interfacing circuitry, if any, should be fast too.
• There is another quirk that should also be kept in mind. The input latch, which is responsible for sampling the state of the scanlines, is integrated into the TED chip. It won't make problems anyway, but the TED is not well protected against peaks and high currents. A very common situation is when this latch 'goes to the moon'. Luckily, this failure doesn't mean that one must obtain another TED, nevertheless some problems still come in. The input keyboard scanlines will appear going to the keyport read register ($FF08) directly, instead of the 'in a certain minute' sampled behaviour. Once again, this doesn't result in unability to work. Since the other port - which is responsible for the keyboard scan (the keyport output latch) - is a simple 6529B, the selecting lines are static, it's no matter that reading $FF08 happens in one or another minute (one probably gets still the same result from the keyboard if reading the lines later by few microseconds; anyway, few programs like Laser Squad sense no keys on a defect TED machine- maybe they change $FD30, the keyport output latch after they wrote but before read $FF08). In opposition to this, the joystick read process is heavily dependent on the above sampling. With a defect TED, one most probably gets no answer from a properly interfaced (or even original Commodore) joystick. Why? The selection depends on the state of datalines in the sampling cycle. Now, when the real 'sampling' occurs, it's the very same cycle when reading from $FF08. In this moment, there is $FF on the databus, the last byte of the cells address to be read. Thus all data lines are high - no joysticks are selected.

A simple interface

 You'll need...

• A Sub-D 9 connector (male), with the plug housing.
• A Mini-DIN 8 connector (male). You may have troubles finding this part, but resellers like Konrad, RET or Radio Shack in the US can surely ship it.
• A short 7 wire cable. You can also substitute this by a piece of ribbon cable.
• A soldering iron, solder, some solvents (if needed...) and a good amount of patience... Whatever soldering iron will do (if you're good in this process...) but I'd suggest a Weller or similar termostate one, especially with sharp soldering pin.

 You can verify the routes on Figure 1. All lines go straightforward to their equivalents. The only exception is the Select' line: this is connected to the joysticks GND pin. +5v is connected because some autofire circuits will still work if it's present, even if the 'GND' level is present just occassionally. Note that POTX and POTY (Sub-D #5,9) are unconnected.

 After soldering everything, check if there are no loose wires, shorts on the connectors. Then assemble the plugs together.

 You'll most probably have to adjust the mini-DINs cover, because there's so small place around the Plus/4's connectors that these replacement plugs don't always fit to the computer (the original, cable manufactured ones do, but they can't really be obtained).

 Try your work out. Connect the interface, the joystick and the computer together. Switch the computer on. Try to move the joystick. If you see 'A', '4', '3', 'W' characters on the screen (joy2) or '5', 'R', '6', 'D', 'T' (joy1), your interface is perfect. If you see '2', 'Q' or the cursor jumps to 'HOME', you'd most probably connected the GND lines together instead of connecting Select'.

 When you notice no effect, first check the cabling, the wires, the joystick etc. If nothing works, your Plus/4's TED is most probably broken. This interface doesn't work for you.
 

• Take a new TED. ...This may be impossible. No TEDs were manufactured since 1990 or so.
• Design a small board (or just circuit, piggy backing the chips in the Plus/4) that captures the TEDs input lines right (or rather after a short time) when $FF08 is written. This may be difficult, since it implies a lot of decoding logic. But it's possible.
• Disassemble your interface (the mini-DIN plug), remove the wire from pin #8 (Select') and connect it to pin #7 (GND). This will mean that your Plus/4 won't be able to differentiate between joy1 and joy2. But the joystick will somehow work. If you experience strange behaviour when using certain programs, it should be because both joystick ports are selected statically from the point of the Plus/4.

A full featured interface

 My interface is built around a 74LS244, a dual quad tristate bus-driver chip. But you can use several other chips too. In fact, five OR gates would do the trick (but unfortunately, simple OR chips contain only 4 gates). Here's the schematic diagram.

It looks like spaghetti. However, the connections are simple, and this 'spaghetti' look is the consequence of drawing the schematic from the ready interface. All direction lines go to the bus driver chips inputs. The chips gate inputs are connected to the Plus/4 joystick ports Select' line. The gate inputs are low-active, so when Select' goes low, the chips outputs will reflect the respective inputs. Else, the outputs are in high impedance state - say, they don't even exist from the point of the Plus/4 joystick port. In theory, the joystick side would need 5 pull-up resistors of about 2-3K ohm each on the direction lines; I left them, since the TTL inputs are on a weak high state as default thus it works with or without those pull-ups. A bypass capacitor of 100 nF (between pin 10 and pin 20 of the chip) won't hurt either, but there was no place in the housing for one more ceramic capacitor. The series resistors limit current, as the original Commodore joystick does with its sensors.

(If you built my previous interface design with the diodes, don't worry - applying the diodes or using an open collector bus driver doesn't appear to be hurting, just more than neccessary.)

 What you'll need is...

• A Sub-D 9 plug (male), with cable-end housing
• A mini-DIN 8 plug (male)
• A short 8-wire cable
• 74LS244 chip. This is one of the most common chips ever used in digital electronics. If you have some old wrecks, I bet you have a lot of them. But if not, any part resellers can ship it. You can also use any other family (74F244, 74S244). I'd avoid, however, using standard, C, HC (maybe even HCT) parts because of logical level incompatibilities. You can substitute the '244 by a 74LS757, which is almost the same component but has open collector outputs.
• 5 pieces of about 100 ohm resistors. Limits currents (fakes Commodore joysticks).
• A lot of short wires, to connect the chips pins to the Sub-D 9 connector.
• A small amount of epoxy resin
• A screwdriver, pliers, solvents (or resin) and your favourite soldering iron. You may use instant soldering iron, because this TTL chip is not much sensitive to peaks; but you'll be at least hard on soldering in small places.

 Here is the chip with its pin-out. The red arrows indicate the connections and data directions inside the chip

• Cut some short wires which will connect the Sub-D 9 pins to the chip. Now you know what is seated where; then you can approximate the lengths. Blank the wire ends.
• Take the 8-wire cable and blank it's ends.
• Take the resistors, and cut their ends 'short' (as needed). Bend them like they forms 'U's with the resistor body on one arm and just wire on the other - it will be the easiest to solder them this way to the chips pins, so you can still solder the wires to the other ends.
• Take a short, naked wire. Connect pins 1 and 19 under the chip together, solder the wire and cut the ends. Make sure that the wire doesn't come off the pins when you'll solder to them.
• Take the 8 wire cable. Solder one wire to pin 19 of the chip (Select'). Take a short wire together a with a wire of the cable, and solder to pin 7 of the Sub-D connector. Do the same with pin 8 of the Sub-D. Solder the other end of the short wires to pin 20 and pin 10 of the chip. (+5v and GND)
• Connect all appropriate Sub-D direction pins to the chip's inputs (see figure, which to which).
• Pick up one resistor, and solder it to one output pin of the chip. This may be difficult. You may want to use some solvents or resin (it makes soldering much easier). Repeat it with the other resistors.

• Put the plastic cover of the mini-DIN plug to the cable. Solder all wires to the appropriate pins (see figure).

• Loose connection somewhere.
• Defect 74LS244
• Defect joystick
• Defect TED

• Finally, I'd suggest to make it secure. Mount the cable fastener part of the Sub-D connector to the cable. Now the Sub-D should be quite compact inside, with a lot of wires and simply no space. Take the epoxy glue, mix a small amount and fill the Sub-D 9 housing around the chip and the wires. When done, put the two sides together and fix the screws in the connector housing. The glue solidifies in few hours.

Contact