Share this project


Share this project

Build mechanical computers powered by marbles to solve logic puzzles. Escape planet Eniac and discover how computers work.
Build mechanical computers powered by marbles to solve logic puzzles. Escape planet Eniac and discover how computers work.
4,198 backers pledged $404,071 to help bring this project to life.

Turing Tumble Progress Update #9: The Second Sample


Happy Holidays!

There's a lot to cover in this update. I'll start by walking you through the second iteration of the plastic parts that arrived on Monday. Then I'll explain the next steps and timing, there's an update on the Virtual Pack, and I'll save some juicy technical details for the end.

The Second Sample 

It took significantly longer than expected to receive the second iteration of the plastic parts, but they finally arrived on Monday. We had about 50 things for LongPack to fix from the last sample. We were more than a little anxious, particularly about the ball release system. Last time it released multiple balls at a time. With our changes, would it work reliably this time? 

The ultra-low friction of these injection molded parts is a blessing and a curse. I'm not sure how you can model their behavior before you make injection molds. CNC milled parts get you close, but not all the way there.

The package arrived on Monday afternoon:

The blurry picture I promised you
The blurry picture I promised you

It came in one of the old magnet boxes - this round was just to test the plastic parts, not the box. We'll get a final sample of the box in the next round.

The box everything came in
The box everything came in


I opened it up
I opened it up

And here's the board. Last time, remember how there was too much warp? The molders made a tool that puts pressure on the board immediately after it's molded to reduce the warp. You can see there still is warp there, but it does appear to be reduced. Also, remember the parts aren't made in their final color. They just used whatever plastic was already in the machine.

I took the board out and tried to put it together, but ran into the first problem right away:

A standoff
A standoff

It's missing a lip on the bottom that makes it snap into the other parts. Here's how it's supposed to look:

The missing lip
The missing lip

LongPack will fix that in the next round. In the meantime, I used my CNC milled standoffs to put the board the rest of the way together.

If you remember last time, the parts were really loose on the pins and wiggled around too much. This time they fit just right. Not too tight, not too loose. Also, it turns out the warp was reduced just enough to not cause any problems. It doesn't interfere with the board's operation anymore.

I put some parts on the board, held my breath, and ran it for the first time. Here's what happened:

 project video thumbnail
Replay with sound
Play with

The ball release mechanism worked! The crossover worked! The ramps worked!

Er...until a ramp fell off the board. In the video, I said that the ramps must still be too loose, but that actually isn't quite the problem. I spent a long time yesterday taking slow motion videos and trying to figure out what's really happening. It's a more complicated problem, but I figured it out and it should be fixed in the next round. I'll explain the problem and the fix for anyone who's interested in the technical details section below.

Then I tried a different setup involving bits. Last time, I couldn't even test the bits because the molders forgot to cut a section out of the arrow on top that's critical for them to function:

 project video thumbnail
Replay with sound
Play with

They worked perfectly! But what on Earth happened on the bottom? I've never had a ball get stuck there before. After messing around more with the board, I found that it happens only rarely, but frequently enough that I'll see if we can make a modification to stop that from happening.

So, as it stands:

The Board: The ball release mechanism works reliably, the parts fit well and turn smoothly. The warp is still there, but it doesn't interfere with its operation. We'll see if we can fix that rare problem where a ball gets trapped under one of the levers.

Ramps: Every once in a while they fall off the board, but they should be ok in the next round after a few changes (see Technical Details section below).

Crossovers: Every once in a while they fall off the board, but they should be ok in the next round after a small change (see Technical Details section below).

Bits: Good to go.

Interceptors: No problems, but this was a pretty easy one. :)

Gears: Perfect.

Gear bits: Last time there was a lot of sink near the teeth of the gear - so much that it left voids in the plastic. I made a modification to the part, adding little indentations in the places where the plastic was thickest, and it seems to have solved the problem. No more sink!

The new gear bits - no sink!
The new gear bits - no sink!

But, like I said before, the lower friction is both a blessing and a curse. It means we can connect 6+ gear bits together in a line and a single ball can flip them all. But it also means that if only two gear bits are connected together, they flip too fast. Sometimes that means they bounce back the other way a little, which can mess up the operation of the computer.

I'm not sure that's a problem we'll be able to fix without making some serious modifications to the part. And anything we do to make it work with two gear bits would make it stop working with larger numbers of gear bits. Fortunately, there's an easy way to make it work - if you only have two gear bits connected to each other, you can just add a third gear bit to the chain anywhere you want and it slows them down enough that they work reliably again.

Next Steps and Timing

Last time we had 50+ changes for LongPack to make. This time we have only 6 changes. We think there's a good chance everything will be ready to go in the next sample. So we asked LongPack to add the final texture to the parts and they're going to make them with the correct colors.

This last iteration took a lot longer than we expected. We were expecting 10 days and it took a little under a month. We were told the next iteration would take 5 days + 2 days for shipping, so if all goes well, we're hoping to get this into production next week or the week after that. If we begin production mid-December and production takes 4-5 weeks, we'll have the games on boats by the end of January. Unfortunately, that means our timeline is pushed back another month. Instead of February, we'll probably have the game in your hands some time in March.

The real key is to have our production finished before Chinese New Year begins. China shuts down for about a month starting a week into February. If we don't have it on a boat by then, it will be another month of delay while we wait for Chinese New Year to end. Fortunately, LongPack thinks there's a good change they'll be able to get it on a boat before then.

Virtual Pack Update

There was a request to add STP files to the virtual pack, not just STL files, so I created a new version of the virtual pack that includes the STP files. You can find it at the same link:

Also, if you haven't had a chance yet, take a look at jesusaurus has been working to make a version of the game that can be printed on home 3D printers. He cut up the board into 16 sections and posted the STL files so others can print them. It looks promising!

A 3D printed board split into 16 sections small enough for a home 3D printer. This was made by jesusaurus.
A 3D printed board split into 16 sections small enough for a home 3D printer. This was made by jesusaurus.

Technical details

I thought some of you might be interested in more details on two key problems with the parts and how we're going to solve them.

The problem with the ramps:

I thought I'd start by explaining why the ramps are falling off the board and how we're going to fix them. It's a riveting tale of momentum and friction.

The ramps fall off the board now because they can rock back and forth too much. See the picture below showing the top side of a ramp placed on the board and the direction of the rocking that's causing the problem:

The rocking direction that causes the problem
The rocking direction that causes the problem

Here's how it falls off the board: After the ramp is pushed down by a ball, the counterweight returns it back to its original position. It stops returning when the pin running through the "smile" hits the end of the smile. When the pin hits the end of the smile, the whole part rocks - in the above picture, it rocks so the left side of the ramp moves toward the board and the right side of the ramp (the side with the counterweight) moves away from the board. When the counterweight gets pushed away from the board, it sometimes gains enough momentum to push the whole part off the pin (or at least it pushes it way down close to the end of the pin so that the next ball to hit it knocks it off the board).

The 3D printed parts and the CNC milled parts don't come off the board because they are not as smooth and the extra friction stops the parts from sliding off the pin. I don't think there's a good way to add friction to our injection molded part, so we're going to solve the problem a different way.

Our solution:  

Right now, most of the ramp sits about 1 mm off the surface of the board due to the small, circular protrusion right here:

The circular protrusion
The circular protrusion

I put that protrusion there to reduce friction as the part turns. It used to help, but now it's the source of the problem. It acts like a fulcrum, and the part rocks back and forth over it. So to get rid of the rocking, we're going to remove the protrusions so that the parts sits flush against the board surface. If it sits flush against the surface of the board, it can't rock anymore. I tested this theory by sanding off the protrusions from a couple ramps and it did indeed solve the problem. We're also going to make the ramp slightly wider to keep it about the same width as it was with the protrusions. Here's a picture of the top of the ramp with those changes:

The new ramp design - no protrusions and a little wider.
The new ramp design - no protrusions and a little wider.

The problem with the crossovers:

Friction (or lack thereof) is once again the problem. In the case of the crossovers, the lack of friction makes it possible for crossovers to fall off the board when a ball hits them at just the right angle. Before, friction with the pin would keep them in place. Not anymore.

Our solution:

We're adding more material to the "smile" on the back of the crossovers. We're pretty sure this will stop them from being able to come off the board unless they are pulled straight off.

The crossover has dimples now. :)
The crossover has dimples now. :)

That's all for today. Hopefully we'll have another update for you next week or the week after that. If all goes well, the title will be "Turing Tumble Progress Update #10: We're in production!"

Thanks again,

Paul and Alyssa

Dessart Marc, Michael Shick, and 45 more people like this update.


Only backers can post comments. Log In
    1. Missing avatar

      Mike Porter on

      Thank you for sharing the process details.
      I like your mindset.

    2. Zwergfell on

      Hi, I like the washer idea as a workaround. So you can do the programming and additionaly some mechanical experimenting. ;) ...and I‘m surprised too that changing the angle dosen‘t have this big effect... Thank you for your detailed answers.

    3. Paul Boswell Creator on

      @Everyone else: Thanks so much for the kind comments and the support! It means a lot to us. We read every single one of these comments and think them through carefully.

    4. Paul Boswell Creator on

      @Lode: Thanks for the ideas! I like the idea, but I'm really hesitant to make any new molds for plastic parts at this point. I think it would add a pretty big delay.

      We tried a couple things this morning and arrived at a solution that could work. I'd like to know what you and everyone else thinks. So the gear bits actually sit on the end of the pin, not against the board. They are held off the surface of the board by about 0.5 mm to reduce friction. If I sand down the pin a little bit so that the gear bit sits against the board, the friction is much greater. Then you can only have like 2 or 3 gear bits connected together. Any more and the friction is too great.

      So here's the idea: We make all the gear bits so that they sit flat against the board. But then we also include little "friction-reducing washers" that you can put behind the gear bits if you want to connect more than 3 together. It's the best of both worlds, but it adds complication because you have to keep track of the little washers and you have to learn how to use them.

      What do you think?

    5. Paul Boswell Creator on

      @Matthew Smith: Thanks for looking at that! This morning I changed those platforms so that there's not even a depression for the ball to get stuck in. It's a pretty big change to the mold, so I guess we'll find out if it's something they can do tomorrow.

    6. Paul Boswell Creator on

      @ming: Absolutely! Someday, in TT 2.0, that will be the way it works. I didn't do that in this version because the injection mold needed to make a something like that couldn't be a simple straight pull. It would have to be more complicated and expensive.

    7. Paul Boswell Creator on

      @Zwergfell: I agree - those ramps are faster than I'd like, too. Hopefully the extra friction of having the whole part rub up against the board (instead of just that little protrusion) should help. I also made the cup 0.4 mm longer. That should help a little, too.

      Surprisingly, changing the board's angle doesn't make a noticeable difference unless you change the angle drastically, but then some parts stop working, like the ball release mechanism.

    8. Missing avatar

      Lode Vandevenne on

      This updates are awesome to read!

      About the two gear bits issue: rather than adding a third gear bit on the board, could the following work: Have some extra part that you can stick in front of the gear bit that adds weight to it. Then it doesn't take up a spot on the board, instead you can choose to make a gear bit heavier. Or alternatively, have a few heavy and a few light gear bits in the box. Just an idea, not sure if it can work :)

    9. Robert Hencke on

      Updates like this are honestly half the fun of backing a project. Thank you for taking the time to document everything out, and let us share in the journey. :)

    10. Missing avatar

      Matthew Smith on

      Regarding the second video with the ball getting stuck: is it just me (and the angle the video is shot at), or does the small ledge that halts the left lever warped downward slightly? It looked to me as though the stuck ball became stuck after bouncing off that ledge. If that wasn't just a fluke, you might try shifting everything in the bottom section (the ledge, the levers, and the entry holes) outward a bit (and adding a slight peak between the entry holes).

    11. Missing avatar

      Lucas Fowler on

      An excellent update. This is exactly why I back KS projects. I want to see how you solve problems and bring a product to market.

      Please, don't rush. Get this right. I would rather wait another week and get something which works than for you to have to send out 'bugfixes'.

    12. Missing avatar

      Don Ripperger on

      Thanks for spending the time now that the toy will work properly when we get them. Too much time in the past spent fixing newly opened toys with poor designs or manufacturing flaws. Your explanations and frustrations gives me trouble shooting incite into R&D of manufactured items. Fix it here now it doesn't work there;0)

    13. Missing avatar

      ming on

      For the falling off problem, i am wondering why dont simply make the pin ball headed, just slightly expanded on the tip so that it locks in the whole part inside. Currently it is straight from top to end. If there is a ball tip with the right diameter just equal to the centre whole of the parts, it would be good enough for the parts to be permanently locked in the board. It requires the pin to be slightly longer, but i suppose that's OK?

    14. Zwergfell on

      Thank you for this great detailed update. Never thought that friction makes such difference - and problems...
      In the videos the balls roll really faster than in your prototypes. For me a little too fast... What‘s about a lower stand angle of the board? But I think you tested that...
      Keep up the really great work!

    15. The 4th Jawa

      Never thought changing the production method would have such an effect on the performance. But you seem to have it well under control, good work. Looking forward to that "In production" update.

    16. Biomage on

      Excellent update. I appreciate all of the technical details behind the issues and your efforts to get them fixed.

    17. Scott Johnston on

      Great work Paul and Alyssa, my respect for anyone who tries to manufacture anything just rose another notch.

    18. Jan Wildeboer on

      TL;DR despite high hopes my kids will not get this for Christmas and I’m disappointed again. Hrmpf.

    19. Oglik Naes

      Thanks for the update. And thanks for making sure everything is working properly. Excited to see how it all looks in the next iteration!