Spark/SparkCE Optical Joystick Sensor for JLF


ETA on availability of SPARK CE?


I’ll check tonight. the newer link is on one of my custom sticks. I’ll check the dead zone to see if there is a noticeable diffrence


I just tried it with the link… You really have to play with the sensitivity but I was able to get it to cancel the dead zone phreak was talking about… Small issue is every so often the return to neural would cause an input… I’m sure if I played with the sensitivity a bit more I can correct that but wife made spam and rice! Pretty hungry will mess with it again later


the thread title should be changed to The official Spark/Spark CE OPTICAL JOYSTICK for the JLF… so people see it and not mistake it for the sparky LED thread…




Gate and Spring Testing

Recommended minimum sensitivity 55Ω at all settings for consistency.

I might refer to some SFxT character trials as execution examples, since that’s what I’ve been doing with my stick lately, no homo.


GT-8F (standard JLF restrictor)

4-way mode:

[INDENT=1]At the lowest sensitivity setting (110.8Ω on my sample) diagonals disappear entirely except for a very tiny spot at Down-Left. Best 4-way mode on a JLF. Tested on Pac-Man. :D[/INDENT]

8-way mode:

[INDENT=1]The standard for the games we usually play. Works well at all settings 55-110Ω+ depending on your preference. Recommend setting 55-60Ω and working your way up if you want less sensitivity, or want smaller diagonal zones.[/INDENT]

To some extent, the stuff below applies to the square gate, too, but I wanted to focus on the alternate gates since square is pretty straightforward.

GT-C (circular JLF restrictor by Toodles)

Diagonals are usually the issue with circle gates, especially in sticks with round actuators like the JLF. The tradeoff for the smooth circular motion is shallow diagonal zones. If zones for diagonal directions are too small, you will need higher precision to block and jump. Even if you are very precise with blocking, some charge moves may be easy to input incorrectly, i.e. Boxer dash low vs. dash straight (should end in DF vs F).

Luckily the SparkCE is adjustable so you can get good diagonals.

The circular gate can be a pretty solid execution assist with motion-based or 360 characters. Ogre trial 19 (multiple dash cancels) is easiest on GT-C. Balrog/Boxer types who care about ending execution in UB and DF still play decently, but are probably best suited for square if you don’t really love the circle.

55Ω setting
[]Diagonals very responsive
]Dashing is touchiest here, since cardinals are deep and neutral is small, it is easy to bang the gate twice without actually returning to neutral during a hurried dash motion
[]If you tend to cut your fireball motions short it might be an issue since the diags have higher priority
]Spring rebound can cause inadvertent input in opposite direction (LS-55 spring)
In general, the 55Ω setting is very good and playable; the above is nitpicking for picky nits.

65Ω setting
[]Diagonal zones well-represented and even
]‘Short’ qcf motions less of an issue
75Ω setting
[]Down-Right diagonal getting a little tighter
]No issues with abbreviated qcfs
[*]Spring rebound disappears in all but extreme cases
85Ω setting and higher (110Ω+)

Sensitivity lowers gradually, diagonals shrink some. It is still plenty playable at absolute minimum sensitivity without wonky inputs.

GT-Y (octagonal restrictor)

This one surprised me. I have never been a fan of the GT-Y; the octo gate is just too loose and sloppy in a stock JLF for my tastes. But it feels fine with the SparkCE and some good spring tension (see below). I normally hate the GT-Y and rarely pass up on an opportunity to ward people away from it if they don’t have the money to spend or the inclination to play with stick parts, but if you want to pair it with the CE, definitely give it a try.

I liked this at all settings 55-110Ω+. All directions still worked fine at minimum sensitivity. Cammy trial 20 (jump-cancelled qcb cannon spikes) felt pretty natural with the GT-Y.


The JLF becomes very light in feel when the microswitches are removed, as the switches provide a good portion of the total resistance. You’ll probably want to swap in a heavier spring to compensate for the lack of switch tension. I think the LS-55 is probably the best all-around spring paired with the SparkCE. It works well with all of the gates and has medium tension.

A single LS-32 spring gives it a lot of the throw feel of the stock JLF, but it doesn’t feel the same altogether. Switches make tension happen at the outside of the throw, which is completely lacking with an optical setup. I think going a little heavier on the spring makes the optical setup feel better.

Adding an LS-33 spring is an easy way to get an “in-between” level of tension. Take care when installing thinner springs like the LS-33 and LS-55 that they slide over the entire spring holder piece (JLF-P-6), as they fit snugly.

These are the springs I like the best for different levels of stiffness:

[INDENT=1]Light - LS-32[/INDENT]
[INDENT=1]Medium-Light - JLF-SP + LS-33[/INDENT]
[INDENT=1]Medium - LS-55[/INDENT]
[INDENT=1]Medium-Heavy - LS-32 + LS-33[/INDENT]
[INDENT=1]Heavy - JLF-SP + LS-32[/INDENT]

The Medium springs are closer in tension than the Light and Heavy, which feel much lighter and heavier, respectively. My favorite is probably the LS-55 or JLF-SP + LS-33, not sure yet. They feel pretty similar, and the LS-32 + LS-33 is not much different.

You could get away with skipping the LS-33-tweaked settings and getting only LS-32 and LS-55 springs with the SparkCE, then set as follows:

[INDENT=1]Light - LS-32[/INDENT]
[INDENT=1]Medium - LS-55[/INDENT]
[INDENT=1]Heavy - JLF-SP + LS-32[/INDENT]

I did try Joe’s method of cutting the LS-32 in half and pairing it with the stock JLF spring. It feels good, very similar to the LS-55 level, maybe a little stiffer. I think if you want heavier than the LS-55, but the JLF-SP + LS-32 is too stiff (and it is quite stiff), cutting the LS-32 down to 3/4 or 1/2 would get you in between.


Any tests on using the “home depot” spring as to how it would feel without the microswitches?


I haven’t found a good alternative just yet on the home depot

Oh and spam>“aso” haha…it’s already canned and doesn’t make so much noise when you cut it up…


Thanks for laying the knowledge RoboKrikit.
The consensus is that that the usable lower range hits about 40-50 ohms. I guess my big question is about the far end; how is it on a square gate on the max resistance? Does it get to a point where its wonky before maxing out? Does it get wonky at all?
Since there’s so many turns available on the pot to dial in the preferred sensitivity, I’d really like to have it cover the entire range from wonky tight to wonky loose. I’m just trying to get an idea of if the max resistance is far enough or if I should make it go further. I’m sure it may make the circle or octagonal gates unusable at max, so this is really a square gate question.


For me the stick is still plenty usable cranked all the way to the far end on all gates, even the GT-C. When I max it out it measures 110-111Ω. I would give up some of the lower range to get a higher ceiling.

On all gates the wonkiness is 100% eliminated for me at 55Ω. Can someone else test to find the Upper Wonky Range and see if we match? Or is it the Ceiling of Wonkiness? CoW?

First find your wonky zone(s) by setting sensitivity somewhere around 40Ω. It is very very easy for me to find the spots at 40Ω; I have only one and it is barely above Right. In Game Controllers on a PC I see a Cthulhu or Kitty with current firmware register neutral when I hit this spot. It’s easiest to see with the stick set to LS, since the app has to draw the + for neutral.


The 360 TE board shows DownLeft instead.


As you dial up the ohms, it should get harder to duplicate.

Unfortunately I don’t think you can find the real CoW in the Game Controllers app. As the sensitivity gets lower (ohms up), the input that was easy to find is a mere 1 frame blip. You need something with an input history, and not on a console using the Cthulhu/Kitty, because if it is sending Neutral you won’t see it. I used SFxT and SF4 on Xbox360 to find it.

Find the funny input at 40Ω. Mine makes my character, when on the right, walk left while holding right or barely above right. If you have a different spot you might have to find something else to use.


As the ohms go up you have to keep an eye on the input history, as the wonky inputs will get more rare and happen too quickly to cause character animation.


Yeah, that wonkines is because the cardinals become so sensitive that it can activate opposites from the little light reflected off of the actuator. That 360 downleft is how the 360 board handles the SOCD.


on: springs

I may try out a few rc shock springs to see how they work on a spark. I had tried a few on a JLW going for a happ feel, but never bothered anything with a JLF (considering I rarely used a JLF, I guess that made sense). I did get a multi-pack of springs from Home Depot but none really felt right, even with cutting some down.


:smokin: Yep. I think it feels identical to the FLASH1 spring.
I also like that it’s a single spring solution and you don’t have to deal with springs bunching up and popping.


Can’t wait for SPARK CE to be released!
Is there a log of any current issues the spark ce has in beta testing? Just curious to how it is coming along. Keep up the good work Toodles!


As far as the spark CE goes there isnt really any issues we have seen just yet… as long as the user stays within 80 - 50 ohms the response is really crisp and there isnt any funny business with lost directionals or strange occurrences when installing… Toodles made the “CE” to help with the voltage issues from a few pcb’s… cool little trade off is that now adjusting the ohms helped with the sensetivity of the spark making the response more to the players liking… I’m waiting for him to mass produce these beauties as well… I will want these in all my sticks…

Only issue would be the spring… we are still trying to find one that will match close to what the tension is with the micros in a JLF… the LS-55 is good and if you are a little picky like me the two spring method where you cut the smaller spring to help with the tension works really well… just finding that single alternative is what we are looking for atm… the actual Spark CE is playing out pretty well up to expectations…



I havent received any information on them other than what’s been posted here, and I’d prefer to have all of the info, good or bad, out in the open anyways.
So far, the only adjustments that I can see that need to be made is to tweak the resistances so the lower end of the adjustment is a little higher than what is currently possible. So instead of a range of 17-113, Im thinking of pushing it more to 32-128. There’d still be the some wonkiness at the lower end, but less of it, which means more usable area to select from.
PCB support, other console voltages, all appear to be going to keikaku.


The spark is the real deal… response is scary good… still looking for that one special spring toodles…


You should have a cool dust washer come with the spark ce similar to the flash 1 dust washer.


I would be most pleased with your input on that. Maybe I should talk to blklightning about scouring McMaster for some springs to try.

That would be nice, but I can make no promises.