500k vs 1M paddle pots?

[Andromeda Stardust]

I'm in the process of adding paddles to my Atari arcade joystick. I bought a couple of 1M pots from Radio Shack (the ones with the long-arse rods on them), and I've already got them mounted in the sides of my joystick (one on each side). Among other enhancements, I've added a cutoff switch for the odd joystick game that behaves oddly with the paddles connected (Harmony menu is much easier to navigate with a joystick, for instance), and since the original paddle controllers only used half of the pot, I've installed two additional switches to "flip" or reverse the logical paddle direction. While testing, I observed the fact that a sizable chunk of the 1M pot is wasted, and for now I have to deal with the too-small 1" control knobs provided by Radio Shack (I'm in the process of making a pair 2" wooden knobs for the paddles, complete with set screws to lock them in place and a nice rounded edge for comfort, but I'm currently waiting for the black spray paint to dry).

 

Using KaBOOM! as a test cart, I measured the movement with my multimeter, and the usable range lies entirely between 77k and 297k. That's only a 220k range on a 1M pot (approximately 72 degrees total movement), less than half of the potentiometer is actually utilized, and I noticed I tend to over correct too much with my fast twitch reflexes: The onscreen object seems to move too fast for a given input. Since this is an entirely customized project, I'm not bound by the constraints of the original component specification. I believe a 500k pot would use a greater degree of freedom of movement on the dial, allowing greater precision in game. Obviously, going to the larger 2" wooden knob will help out some, but I think the default 1M pots just waste a lot of space on the dial.

 

Edited March 24, 2013 by stardust4ever

[BigO]

I agree that it seems most of the pot travel seems wasted. I would suggest that you test with many games before determining where the usable range lies. About 6 or 7 years ago, I started building an analog "teeter board" controller for the 2600. (I mostly envision playing Circus with it, but I suspect I'll have to hack Circus to play slower).

 

One challenge I had was trying to get enough resistance delta in a relatively small range of pot rotation. I solved that by finding a large value pot (I think 7 meg). I quickly found that the full range of motion for one game was inappropriate for others. So, I figured out a way to have a simple mechanical variable "gain" between the board and the potentiometer. Having solved that, centering was the next challenge. I got a working mockup built that addressed the issues and it's been sitting on a shelf in the garage ever since. One of these days, I'll build a version that will actually support my weight. At least you're not as mechanically constrained with centering and min/max mechanical position as my contraption is.

 

Long story just to say make sure that you experience a lot of games before you decide to change the pot value. Let's say that you go with a 500K pot and find that a game requires the range 400K to 700K to play properly. II don't remember if that actually happens in any of the games I tested, but there is a large variety of relative percentage and absolute portion of the pot that is used.

 

If I recall correctly, Blackjack uses a larger percentage of the total range than most. I can't seem to find my notes... I think Breakout uses a small range and it seems like it was in a substantially different part of the range than Kaboom.

 

 

Anyway, cool project. I saw an arcade joystick in my project pile just the other day and was thinking of doing something similar. Now I can just sit back and wait for you to identify and resolve all of the issues before I build mine.

Edited March 24, 2013 by BigO

[BigO]

Also, if you haven't tested with the 500k pot, you might consider doing that. One (possibly undesirable) side effect is that you have to turn the knob farther to get the same amount of resistance delta. That could affect game play.

 

When working on the variable sensitivity referenced above, I had a thought about using a gang potentiometer with 500K segments and switching between just one of the segments (total 500K delta) and the two in series for a 1 Meg delta (faster response).

 

If you do want the slower response of the 500K pot, and do find that there's a game requiring something above 500K, you could just put a separate 500K "trim" pot in series with the one you use to play the game. This would let you shift the center of the range and keep the less twitchy response.

Edited March 24, 2013 by BigO

[Andromeda Stardust]

Thanks for the feedback...

 

I did order some 470k "conductive plastic" pots with long shaft arms from an Ebay dealer in the UK. Cost me 7 Pounds (3.50 at 175 a piece for two pots, plus 3.50 shipping) which comes out to $10.50 USD. It's damn hard to find the long rod pots in certain values. Radio Shack only carries the long shaft pots in 10k, 100k, and 1M, and the long arms are necessary for mounting under plywood. I thought about getting a dual ganged 1M and wiring it for parallel, but I thought that might add a level of noise to the circuit, and I'd still need the "long rod" type as I'd rather not shoehorn a wooden dowel into an extension for a slotted rod. Also most "dual ganged" pots are tapered for audio use rather than linear. If it's indeed true that different games use a different resistor range, then I may well have to stick with the 1M pots. I've already drilled three holes in the back (in addition to the 9-pin male serial port) for switches to disconnect and invert the pots, and since my joystick is only an 8x8x4 box, I'd rather not have a clusterf*ck of confusing doodads sticking out of it. If I'd have to make setting adjustments on a per game basis just to use the less sensitive 500/470k pots, I might as well stick to 1M.

 

I was under the assumption that the min/max was set by the console. It's also possible that variances in the tolerance of the parts in each console could be different. The Atari Paddles are definitely a strange beast with the fact that both ends of the pot aren't tied to ground. If Atari had done like the Arkanoid NES controller did years later, with a simple voltage divider + ADC, then a wide range of pot values would likely have worked. Someone correct me if I'm wrong, but aren't the Paddle pots are part of an RC network, and the CPU/TIA/whatever determines the position by counting the number of oscillations? Weird thing is, the number of oscillations (assuming it uses some type of digital counter circuit) should in theory be inversely proportional to the resistance of the pot (for example, 100k would be twice the cycles of 200k which would be twice the cycles of 400k), and division is not feasible on the 6507, so I'm really surprised that the Atari is able to translate the angular movement of the pot into a proportional linear movement on the screen. Maybe if I really wanted, I could get some tape and a protractor and measure the corresponding angles of the paddle wheel and convert it to inches on my HDTV screen, but it feels balanced enough when I play.

 

I did some research online and I found out that some pots can have "taper", meaning the track is wider at one end than the other. This results in a 50% percent full rotation not being truly equal to 50% resistance. "Audio Taper" pots are very useful for audio volume applications because 50% voltage drop only corresponds to a 6db drop in sound pressure, cramming the more desirable lower volume levels very close to the bottom edge on a "linear" pot. Other applications such as voltage dividers or balance/fade/tone controls need a linear pot. I checked and the pots I have in my arcade box (that I purchased at Radio Shack) are 1M linear taper. The new homebrew paddle seems to move more smoothly than the stock vintage paddles I have. They were a little scratchy when I first bought them last year, but after I dismantled the paddles and sprayed contact cleaner inside the pots, they work much better now. Still, it's not surprising that a brand new pot from the Shack would turn more smoothly than 30 year old tech.

 

I haven't given thought to the fact that games may set the usable range of the pot. I had always assumed that there was some kind of ADC circuit inside the Atari that yielded an output somewhere between 0 and 255, which the game could then use. Current Paddle games I own are Video Olympics, Warlords, Kaboom, Medieval Mayhem, Breakout, Super Breakout, Circus Atari, Bumper Bash (which doesn't count because it only uses the paddles as digital "flipper" inputs), and Beat 'em and Eat 'em. There may be a few more paddle games in my collection that I overlooked. I'll have to give another whack at Circus Atari because the first time I attempted to play it, I didn't think to read the label and wondered why I couldn't get it to work with the joystick.

Edited March 25, 2013 by stardust4ever

[BigO]

There's no standard set by the console as to where min and max are. With the cheap A/D methodology employed there, having no inherent absolute min, max or center is really what makes it a workable setup. Variation in parts from console to console is a virtual certainty.

 

The paddle A/D doesn't work the way you are thinking. The variable resistor configuration of the pot is used to charge a capacitor. It is, technically, an RC network but it's not an oscillator like you're thinking.

The capacitor is connected to a digital I/O line. Under software control the hardware discharges the capacitor. The capacitor is then charged by a current supplied through the paddle control. The software counts how much time it takes for that input pin to reach a logic 1 state. The software then applies some meaning to that time and sets the player/token in a representative place on the screen. As the resistance of the pot is increased, the current flow is reduced and it takes longer to charge than when the resistance is low.

 

When the software wants to know the position of the paddle, it discharges the cap then lets the cap start to charge again and detects when that digital input goes HI.

(There's a portion of the non-linear cap charge curve that closely enough approximates linear to make this scheme workable. I'm assuming that charging for less than two time constants (tau) brings the cap up enough to be seen as a digital 1 else the whole arrangement would get very non-linear.)

 

 

Audio taper pots are logarithmic if I remember correctly. Yeah, you don't want those. I have the benefit of a local electronics salvage company that has bin upon bin upon bin of new old stock and "pulls". I can find darn near any kind of pot I want if I have the patience and time.

Edited March 25, 2013 by BigO

[Andromeda Stardust]

Good old RC time constant. I have fond memories of working those exercises in class.

 

So, if I'm getting this correct, the console is counting the number of clocks necessary to charge the capacitor up to logic 1, not the frequency of the RC circuit. That also explains why the pot needs to be connected to VCC+. Yes, the time constant should be fairly linear compared to the R value.

 

I've formulated a min/max table out of the paddle games I own.

 

Procedure: I opened the bottom panel of my joystick and loaded the game. I rotate each paddle just until the character stops movement. Then I flick the disconnect switch before measurement. My multimeter is kind of shaky on the 2M setting and it takes some time for high impedance readings to stabilize. A few of the values I took twice and typically got higher or lower by around 10k or so. As a result, I'm placing an error tolerance of 10k for the low end values and 20k for the high end ones. All of the games' input ranges stayed under 400k, which is good. The highest R value I recorded was for the Harmony boot loader menu at 396k ohm. Perhaps the Harmony boot-loader needed increased definition due to the menu size?

 

Video Olympics:

left = 51 - 348

right = 48 - 340

 

Kaboom = 77 - 297

Breakout = 90 - 348

S. Breakout = 93 - 273

Circus Atari = 80 - 288

Beat 'em & Eat 'em = 63 - 246

Harmony Menu = 57 - 396

 

Medieval Mayhem:

menu = 57 - 315

left paddle = 28 - 292

right paddle = 29 - 285

 

Warlords:

left = 37 - 302

right = 32 - 294

p2 left = 36 - 287

p2 right = 31 - 293

Given the RC time constant method of determining position, I can see why it would be favorable to keep most of the movement on the low end of the pot because this would waste fewer clock cycles to get readings. I'm going to go ahead with the 470k experiment when the new pots arrive from the UK. If I do stumble across any games I cannot play with my custom built controller, I can always go ahead and use the original pair of VCS paddles I own, which are still in good working condition.

[BigO]

I'll be interested in hearing how the game play is when you have to use twice the mechanical rotary motion to make the same movement on screen. I would expect Kaboom to be more difficult that way, but that's just a gut feeling.

[Andromeda Stardust]

I'll be interested in hearing how the game play is when you have to use twice the mechanical rotary motion to make the same movement on screen. I would expect Kaboom to be more difficult that way, but that's just a gut feeling.

I searched on my old friend "Google" and I discovered this thread from a ways back...

http://www.atariage.com/forums/topic/163689-is-there-a-way-to-make-paddles-less-sensitive/

 

Ironically, it is possible to increase the sensitivity from inside the paddle without swapping the pots, but not to decrease it. According to AJ Franzman, it is possible to connect a capacitor between ground and the paddle pin to increase sensitivity.

Too bad the idea of adding capacitance wouldn't work. Parallel capacitance can be added inside the paddle controller, but this would have the opposite of the desired effect. To increase the travel of the knob (reduce the charging time constant), one would have to add series capacitance. However, if it's added inside the controller, it will be in the wrong part of the circuit, such that the console would be unable to dump the charge to take a new reading. The only way to make it work would be to add a series .068 uF capacitor inside the console (or simply replace the existing .068 uF one with a .033 uF). Adding a series capacitance could easily be switchable, though -- simply have a switch short across the added capacitor to restore original function.

There's a lot of valuable info in the thread. Apparently, the Atari contains a .068uF cap tied between ground and the signal pin on the TIA that reads paddle controllers. Increasing the capacitance will increase the RC time constant. So by adding extra capacitance between the paddle input and ground, it would be possible to "reverse" the effect of using a smaller pot. Adding an additional .068uF cap in parallel with the first will double the RC time constant. Using half the resistance will half the RC time constant. By using a smaller pot in my controller to decrease sensitivity, I can gain that sensitivity back if I want by adding parallel capacitance. So I could simply install an additional toggle switch to enable/disable the parallel capacitance as a simple form of "gain" control. However, I believe there is a limit to how much you can decrease resistance while increasing capacitance:

My wife voluntarily played Circus Atari last night. She definitely needs less sensitive paddle controllers.

 

She seems to be expecting a full turn of the paddle to correspond to full travel of the player across the screen. After a while, I was able to get her to pay attention to the screen and correlate rotation directly to on-screen movement. But, the paddles are just a bit too twitchy for her liking.

 

I guess I will experiment with this sometime in the near future. My plan is to use a smaller pot and tune the sensitivity with external capacitors and see what happens.

 

I spent quite a bit of time playing around around with this methodology on the 5200 and ran into a point where the scheme just wouldn't work with a significantly smaller pot and a large external cap. I think what I was seeing was that the circuitry that reset the cap couldn't sink enough current to discharge it in the short time the system was holding that I/O line low. That was with a pot of 10K or maybe 25K max.

 

It was somewhat game dependent which, to me, is evidence in favor of my theory. However, I never did put a scope on it and compare between games to see if some hold the discharge state longer than others. I'll probably do that if I ever get back into my 5200 custom controller project.

 

I hope not to encounter this issue on the 2600 if I keep the pot size > 100K, probably 250K.

500k + .068uF should definitely not be an issue here.

[+SpiceWare]

Test Backgammon - when I maintained the OS/2 port of Stella I recall that a change to the emulated paddle range caused it to become unplayable as you could no longer select all board positions.

 

Paddle readings tend to be how many scan lines it takes to charge the capacitor. It's not necessarily every scan line:

I fired it[Warlords] up in Stella and did a trap on INPT0. It reads it from scan line 59 thru 206, but not every scan line.

 

59-91 is every 4th scan line

91-131 is every other scan line

skips 5 scan lines

136-178 is every other scan line

178-206 is every 4th scan line.

[Andromeda Stardust]

I have decided that once I get the 500k pots, I'll add in a parallel .068uF cap for each paddle, toggled by a switch. If Radio Shack doesn't stock the .068 caps, then I'll use two pairs of .033uF caps. This will enable me to restore the orignal 1M paddle range on the fly while still permitting the increased travel of a 500k pot when i want it. Once stting for longer travel and the other setting which should match the shorter travel of the original 1M pot. I'll fire up backgammon on the Harmony later and see, but if my plan works I'll have a dual setting. I determined one time constant to be 3.4ms with a 500k resistance and a .068uF cap. Assuming it takes slightly more than one time constant to trigger a high logic state, you've already used up a sizable portion of the video frame to do the testing.

 

Quick question; I have a 10 foot serial extension cable that connects my joystick to the Atari. I'm wondering if the distance of separation between the parallel caps will matter in my setup. I think not given that the Time constant is in the milisecond range. The propagation delay of a ten foot cable probably won't matter unless you're dealing with sub-microsecond timing. I do know that if you daisy chain enough SNES extension cables, it will eff up the controller read, but that is comparing apples to oranges.

Edited March 27, 2013 by stardust4ever

[Andromeda Stardust]

Bump. I just wanted to give a little update to the 470k pots project: I have replaced the original 1M ohm Radio Shack pots with the 470k ohm pots I recieved from a UK dealer. One of the pots measured ~430k ohms and the other pot measured ~495k ohms. Both are over ~400k, so they should be compatible with the vast majority of games out there. Rotation is extremely smooth as these pots are conductive plastic rather than carbon as I assume the radio shack pots are, so they should last a very long time. I tried out a few of my games, Video Olympics, Super Breakout, Night Driver (I recently picked this one up), and KaBOOM!. Overall, the larger rotational area of the 470k pots allows some much needed increased precision, and even with KaBOOM!, I didn't seem to have much difficulty sweeping the board quickly enough. I can even play a two player game such as Video Olympics by myself by using both sides of my Atari joystick. The pots still seem plenty responsive despite the extended travel, so I believe adding a toggle switch with optional extra capacitance to emulate the original range will be unnecessary.

 

If anyone wants to build their own paddle controllers, or refurbish a pair of originals, 500k/470k pots are a great alternative to extend the usable range of the paddles. I haven't found a game yet that uses more than ~400k of paddle range, so the 500k pots are great!

[+stephena]

Maybe try Casino, on game variation 4. This is the one I use when tweaking the paddle emulation in Stella, since it seems to use a wider range of movement than most other ROMs.

[Andromeda Stardust]

Maybe try Casino, on game variation 4. This is the one I use when tweaking the paddle emulation in Stella, since it seems to use a wider range of movement than most other ROMs.

 

I'll keep that in mind. I think my local Game-X-Change has a copy. Is the gameplay any good?

[+stephena]

I'll keep that in mind. I think my local Game-X-Change has a copy. Is the gameplay any good?

 

Personally, I think the game sucks But then again, I'm not into that genre of games, so maybe it isn't bad if you're interested in it.

[Ethaniel]

...to my Atari arcade joystick

My friend, would it be too much to ask for a schematic? I tried making one myself but i suck at electronics.

[Andromeda Stardust]

My friend, would it be too much to ask for a schematic? I tried making one myself but i suck at electronics.

It should be fairly simple to draw one up. Most Atari controllers are direct wired and do not contain any sophisticated circuits. The paddle pots and switches wire directly to the control port with no active components.

 

That means no circuit board is necessary to house components and you don't need any advanced electronics knowledge to build a circuit. I used a 9-pin serial interface for my controller, which uses a standard strait-through serial cable. A continuity tester is highly recommended to ensure the contacts work properly and there are no shorts. I used the following diagram for my joystick and paddle pinouts.

http://atariage.com/2600/archives/schematics/Schematic_2600_Accessories_High.html

 

The Paddles and joystick share different input lines so you can stuff both schematics into a single controller. I recommend using a switch to optionally disconnect the paddle lines if you decide to build a 2-in-1 joystick paddle controller. In very rare cases, a game can detect the presence of a paddle and may not operate properly with paddles connected. The Harmony cart, and some Homebrews designed for Genesis controllers may have compatability issues with a 2-in-1 paddle/joystick controller, hence my recommendation to include a switch to optionally cut off the two analog paddle inputs.

 

Polarity of the paddles can be confusing. The Paddle should read 0 ohms when rotated all the way clockwise, and have maximum resistance when rotated counter-clockwise.

[Ethaniel]

Yes the atari schematics are pretty straight forward with no extra components but my problem is with the paddle fire button... It's using a direction pin and that's where I lose my head.

[Andromeda Stardust]

Yes the atari schematics are pretty straight forward with no extra components but my problem is with the paddle fire button... It's using a direction pin and that's where I lose my head.

Left Paddle Fire button uses Joystick Right; Right Paddle Fire Button uses joystick Left. If you are building a joystick with built in Paddles, and assuming you are using Microswitch arcade parts, you wire the switch for Joystick Left in parallel to R. Paddle Fire, and visa versa. Basically, the common input on all your micro switches (joystick, pushbuttons) are connected to ground, which is pin 8 according to the diagram I linked above. Pins 3 and 4 are the L/R inputs. Basically, you connect the input line to the NO (Normally Open) contact on the microswitch. The normally closed inputs are not utilized. For the paddle fire buttons, the NO inputs on the microswitches are wired directly to the NO switches corresponding to the JR and JL on the joystick assembly. Whenever the switch is pressed, it closes the circuit, connecting the digital input to ground. So, tilting the joystick left or pressing the right paddle button will ground pin 3. Pressing both will also ground pin 3. Pressing neither button will leave pin 3 floating. Inside the Atari, this input is held high by a pull-up resistor, so a logic low will represent a button press.

 

A couple of caution points: The diagram on Atari Age assumes you are facing the controller plug, not the port. If you are soldering onto a 9-pin header inside your controller, you will need to mirror-image the diagram. Also, if you are building your controller with an arcade joystick, the microswitches on the bottom of the joystick will usually be opposite the direction you tilt the joystick. For instance, tilting the joystick up will actuate the bottom microswitch. Keep that in mind when making connections to the bottom of the joystick. Since all of the buttons use direct connections to the controller port, you need to make sure it is wired properly. The TTL/CMOS chips used in more advanced console controllers such as NES, etc are largely fault tolerant, meaning if you short an output pin or fail to properly connect VCC or Ground, it is not likely to cause catastrophic damage to either the console or the controller, but it will not operate properly either. However, Atari controllers are directly connected, with the analog paddle pots connected to pin 7 (VCC +5V) and the digital buttons L,R,U,D, and Fire are wired to pin 8 (Ground, 0V). These pins are adjacent on the controller port. YOU NEED TO MAKE TRIPLE SURE YOU DO NOT INADVERTENTLY CONNECT VCC AND GND, OR YOU WILL DAMAGE THE CONSOLE!!! A simple ohmmeter can be used as a continuity checker. Zero or very low ohms represents a closed circuit (or short) condition. You can use an ohmmeter to test proper operation of the paddles and digital inputs prior to connecting it to real hardware. You may also want to buy an Atari or Genesis USB adapter for testing the controller before connecting it to vintage hardware. The USB ports on modern PCs are going to be a lot more fault tolerant than vintage hardware, and a USB adapter is usually cheaper to replace than a console. Be advised that the paddles will not operate with most USB controllers since the interface is digital only.

 

Good luck on your build; I hope this helps...

Edited April 21, 2013 by stardust4ever

[Ethaniel]

Thanks a lot buddy I will try when I get back from work and reply ASAP

[kenfused]

Plug it into an Atari 8-bit computer and see if it returns the full 0-228 range.