The Dale C. Maley Family Web Site

Click here to edit subtitle

Brass Marble Lift Mechanism

I have built several marble games, and am always on the look-out for a design that catches my eye to build.

I am more or less a Mechanical Engineer, and I like linkages.  I have used the Linkage.exe program written by David Rector for modeling several of my past projects.  On his web page he has many different linkages he had modeled. You can download his free program, and also download some of the models he has already built using this link.

One of those is a 7 link mechanism for raising marbles up, then they take a 2-wire path to return to home position to be lifted again. You can watch a video of the linkage program running using this link.    David Rector also made a YouTube video of the finished model and you can view it using this link.


I took a screen shot of the mechanism in the Linkage.exe program.  I imported that into Sketchup, then arbitrarily scaled the long arm on the upper left to be 11 inches long, and I assumed a 1" diameter ball.  I then got all the link connecting points using the screen shot, then started designing in Sketchup.


I was messing around with the Linkage.Exe program trying to export the motion as a video...........when I saw you can export a DXF file of the mechanism !!!!!!!   If I had known this, it would have saved me a ton of time versus using the screen shot I took of the mechanism !!

The DXF file imported fine into Sketchup, and I scaled the long arm to 11 inches again..........then overlaid the DXF file on my design I made from the Screen Shot.   I did pretty good, but I have 2 significant errors !!!!    Either I dimensioned wrong from the screen shot, of the screen shot is wrong?   I guess I need to fix the errors now.

DXF versus my drawing - errors shown

My drawing fixed to match DXF output from Linkage.Exe Program

My Corrected Drawing

McMaster-Carr Parts to get started on this project

1st Fixture Design

I am going to use 1x4's to make fixtures to hold each of the 5 linkages in their correct position. 

1st Idea did not work

From past experience, I know that when you hacksaw a rod, the resulting face is not parallel to the other end.  I bought a small 3/8" end-mill and thought I could "kiss" the cut end and make it parallel to the other end.  But, in the drill press, at highest speed, the end-mill just kicked the 1/4" long 1/2" diameter brass piece right out of my vise!!  So much for that idea!  I need a vise with a v-groove in it to hold a round part, and I don't have one.  Instead, I held the piece with vise-grips and sanded it on the horizontal belt sander.

McMaster-Carr tube bender

I had this from a previous project, so I tried it on the 1/4" brass rod.  Although it is made for tubing, it works ok on the brass rod.

Soldering Iron

I tried my stained glass soldering iron on the 1/4" brass rods.  You have to keep the tip on both pieces that are to be joined, but you eventually get it hot enough to solder :)

1st Piece being soldered in fixture

I borrowed some techniques from my stained glass work. You can use blue masking tape to prevent soldering adjacent pieces.  You can also use it to hold pieces in position.

1st Piece Done :)

Brass bolts

I ordered a few brass bolts to use on the linkage connections.  I think they will work fine :) 

Trying to Get Smarter

My process to make the 1/4 and 3/4" wide brass pieces with 1/2" OD and 1/4" hole drilled thru them was......

1. Hacksaw blank

2. If 3/4" long, put in lathe and use 1/4" drill to mark center

3. Drill 7/64 hole in drill press with piece in vise, until it spins out of vise

4. Take piece to big bench vise, drill with battery drill [very hard on drill when it catches]

I switched gears and took a 9 inch long piece of 1/2" diameter bar, and marked the center in the lathe using a drill bit.

-put in wood 90 degree fixture I use for drilling holes in the center of dowels.  Clamp fixture to drill press plate. Drill 1/4" hole [plus 1/64" oversize bit] about 3 inches deep.  Hacksaw lengths as needed.  This is a much better and faster process.

Piece #2

Pieces #1 and #2 completed

An Idea

If I make this hand cranked, and kids can crank up to 100 RPM, and I probably need maybe 5 RPM on this model...............this means I need a 20:1 speed reduction ratio.  This is tough to do with spur gears.  Maybe I could use a worm gear to do it?

I made a previous hand-cranked worm gear with a 32:1 speed reduction ratio.

Desired RPM for Model

I went back to the Linkage.Exe model, and 5 RPM is about as fast as you want to go.  If you go faster than that, the mechanism might throw the marble or ball bearing out of the lifting forks.

So if kids can crank at 100 RPM, and I want 5 RPM for the mechanism..........this means the speed reduction ratio needs to be 100/5 = 20:1 Speed Reduction Ratio.  A worm gear, versus a spur gear set-up, is probably the right choice for this model.  I updated the web site for the worm gear model I made back in 2014, because I may use the same techniques to make the worm gear for this model.

2 Identical curved pieces on 1 linkage

I made a form out of 3/4" thick pine with the right curve in it.   I knew from steam bending projects, there would be quite a bit of "spring back" after you cold bend it.  I ended up making the curve sharper in the wood form, until I achieved the bend I wanted.

All 8 brass linkages built :)

Starting on the Brass Base

When I was designing this, I wasn't sure if I should solder the axle supports to the main rails, or drill and tap, and attach with bolts.  I bought the right size drill bit and tap to make a 1/4-20 thread.  I was surprised. The brass drilled easily with this size bit, and hand tapped fine :)   I then step drilled the 1/4" axle hole at the top.   I guess I will go with bolted verses soldered connections of the axle housings to the bottom rails.

Scroll sawing 1/4" thick brass

Using a #9 blate, I was able to slowly scroll saw the mounting pieces for the rotating elements.  It turned out to be faster and easier to hack saw them, then grind to final form using the horizontal belt sander.  As the scroll saw blade starts to get dull, it makes the piece jump and down excessively, which is hard on your fingers holding the piece.

Drawing the worm gear in Sketchup

Back in 2014 when I made the hand-cranked wood model of a worm gear driving a spur gear, I did not know how to draw a worm gear in Sketchup. Yesterday, March 13, 2021, I figured out how to do it.  I found a Youtube video that showed step by step how to do it.  I documented the process in a Word Document for my future reference. You can download a copy using this link.  I made 2 worm gears, one for the 2014 hand cranked model with 1.5" diameter and 9/16" tooth spacing, and one for the marble raising model with 2.0 inch diameter and 0.94245 tooth spacing [to match spur gear with 6" DP and 20 teeth]. I am not sure which one I am using for sure on the marble lifting model yet.

In Sketchup, I made a new pattern to mark out a worm gear at every 90 degrees, for a 20 tooth mating spur gear.  It doesn' matter whether I put the marks on a 1.5 or 2.0 inch diameter worm gear blank, because it is done in degrees.

Worm Gear drawn for 2014 hand cranked worm gear model

1st Fit-Up of Linkages

I ran out of 1"x1/4" brass bar, but I went ahead and made oak temporary pieces so I could see how the linkages work, then work on the ball grasping mechanism.  I had to grind a little solder off some end faces, and ream a hole that a little solder had dripped into, but everything fit ok.  I do have some interference on the RH side where the 1 linkage operates between the 2 vertical bars on the RH side. I may have to rework that linkage.

Putting bottom linkage anchors in correct place

I printed out the distances between the 3 key anchors from Sketchup, then made a block of scrap wood with these 3 holes. I used the block to make sure all the linkage points were correct.

First Spur and Worm gear concept layout

I went with the smaller 1.5 versus 2.0 inch OD worm gear because it looked better to scale.

This concept looks pretty good. I may have to do some re-arranging when I add the ball track in, but I can modify the design when I get there.

Trying out the Mechanism with temporary set-up

Had to shorten linkage where it hit wood block

Had to shorten linkage where it hit the little c-clamps

Had to Remove Cross-Brace on linkage

Worst Interference Problem

Most of the issues above were minor, and easy to fix.  The last issue is a more major one because I have to rework or build a new link. The 2nd link from the LH side I made straight like the Linkage.exe model..............but I had interference and it appears the link needs to be curved instead of straight.  I went back to the Linkage.Exe model, and yes, it is very close to having interference.

Making new Curved linkage versus original straight linkage

The new pieces were 3.5" long straight before I bent them.  I came up with a simple set-up to bend the 1st piece, then be able to match the 2nd piece bend to the first one.

Temporary set-up to get linkages working


Before I installed the new curved link, I tried out the rest of the mechanism to make sure it works ok with no interference.  It worked fine.   You can use this link to see the partial model in action.

After I installed the new curved link, the whole contraption seems to work ok with no interference between linkages !!!!!!!!!

You can use this link to see the whole model in action.

Next step

Next is coming up with the lifting device for the LH linkage, then the next lifting device it transfers it to.

I used the ball location at the LH side when I drew up the model in sketchup..........using the DXF file generated by the Linkage.EXE program.............but when I went to decide where to cut off the LH linkage to attach the lifting device.........the ball is too far to the right??

I tried to simulate the movement of the linkage in Sketchup, but that is way too time consuming.  After studying the Linkage.exe program while that model ran, I observed that when the linkage end starts moving up, that is where you want the ball. I added on to the horizontal bars by sliding the marble track to the I can handle this error on the final model ok.

I am going to use 1/8" diameter brass rod for the lifting devices and the track.  I designed the LH lifting device in sketchup......

I made a video of the finished device working ok, you can use this link to watch it in action.

Saving steps - used bowl of cold water by bench to cool hot parts

Fixed interference problem on LH link between 2 vertical uprights

I bent both pieces using the red c-clamp method shown above, and soldered them back together.  It solved the interference problem :)

I was able to use the same pieces, and did not have to cut new pieces :)

Making 2nd pick-up piece

Boy, I had to pull out every trick in my book with respect to how to clamp the parts while soldering them!  This is very similar problem you have when soldering 3D stained glass projects.

Eureka !!!!!!!!

Got 2nd pick up basket made and attached...............and both baskets work ok!!  They line up pretty well at the transfer point with respect to in and out of the model.  I may have to do some fine tuning with some lead-in chamfers or something to make them work 100% of the time.  You can use this link to watch the model in action.

Boy, it is tough to hold the parts in position for soldering on both pick-up baskets.  Maybe some modeling clay from mcmaster-Carr would help on this. I checked and they still carry it and it is part number 1334-T2.

Now on to the 1st piece of track on the upper LH side :)


I have 1/16 and 1/8" brass rod in stock. The 1/16" is just too flimsy for this model, in my I chose the beefier 1/8" rod.

I had to kind of wing-it with respect to the first piece of track where the marble exits the 2nd lift arm........because the YouTube video was shot at an angle, so I can't precisely copy it.

1st piece of chute

Boy, I had a heck of a time making this 1st chute!  You need to hold the chute in place so you can test the location, and I had no clamps that would hold 2 pieces of 1/8" rod together. Went to Ace, found some larger alligator clips that worked!


You can use this link to see the 1st chute in action.

Another Challenge

On the model in the video that I am patterning this project from, a steel ball bearing is used........looks about 1" diameter.  I don't have any 1" ball bearings in stock, but I do have red marbles, so I can start the project with these.

I made the chute that goes down and sends the marble upside down and around.  When I tested it, the marble would go down to the bottom but only came up about 1/2 way the curve!!

My basic physics is rusty, but the potential energy when the marble starts at the top is the weight times the height, so if a marble weighs less than a ball bearing, then the marble has less potential energy to convert into velocity as it falls.  I checked so see how many times heavier a steel ball bearing is compared to a glass marble.

So a steel ball bearing has roughly 3 times the weight of a glass marble. I think this means the diameter of a horizontal curve can be larger for a steel bearing versus a marble.  I ordered 5 ball bearings from mcmaster-carr.

I Googled it, but no data on what diameter of loop you can have for a given size of glass or steel marble.  I guess I will wait until I get my steel marbles tomorrow, to verify my diameter is not too big.


Boy, I thought the challenge of designing this model was the linkage mechanism.............boy was I wrong !!!!!           Having built wood marble tracks before, I though the track would be a slight challenge, it turned out to be a major challenge!!!!

I googled and found that basic physics says the drop height must be 2.5 times the radius of the loop.  The drop height must be more than the theory says to account for friction.  I started with a 2.5 inch radius and a 9 inch drop height. This failed and the marble got about 1/2 way up the curve. My ratio of the drop height to the radius was 3.6

Someone commented on the youtube with the basic physics equation, in the real word, the drop height needs to be 3.5 times the radius of the loop.  This didn't work either, because I had a ratio of 3.6 and it did not work.

So, I dropped the radius down to 1.5 inches, giving me a ratio of 9/1.5 = 6.  I built this and thought it worked, but slow motion video showed it is not working either !!

Here is link to the full speed video.

Here is link to the quarter-speed video.

Thoughts at this point in the project

Maybe I am looking at this project all wrong from a design point of view.  Maybe it should be, how high must the linkage drop the ball, to have enough height to go through a loop??

I scaled my linkage to give a reasonable size hand cranked model.  I did not scale it to put the marble at a certain  height such that it could do 1 loop.

I'm going to make a dummy base from pine which lifts the brass linkage mechanism up 25/32" from the base. I need to raise it up to give enough clearance for the 6 inch PD spur gear driven by the worm gear.  This will let me mount the brass mechanism in its final configuration, before I attached the loop track.

The ball does have some velocity when it leaves the 2nd pick-up basket and rolls down the slope. Maybe this extra velocity will be enough to make the 3 inch diameter work.  When the ball leaves the loop, it needs enough velocity to go through the other 2 horizontal loops to.

Since the 2.5 ratio of height to loop radius doesn't work, and neither does the 3.5 ratio, there is no way to know the drop height this model needs, except real world trial and error.

Switching to Final Base design

Usually on a 1st time project like this, I make a temporary base until I get the design developed and verified.  On my temporary set-up, my hand crank is on the front of the model, instead of the back, and it is in the way of the loop chute.  Yesterday, 03-19-2021, I moved the model to the final design of the base.  I made this base from pine instead of oak, because I may need to modify the base as I finish the chutes.

I learned it is important to put the 8 links back in the same position when you dis-assemble the model.


One of the challenges of installing the loop, is that you need to hold it and do trial  & error until it works right!  I used a bunch of stuff to get it set just right before I soldered the ring around both segments.........

After soldering, I got excited because the ball seemed to make it all the way around the loop, and I added some track to curve it over the 2 main brass main rails.  There does not seem to be enough momentum left after going thru the loop to go over the main rails, then do two more loops like the original model on youtube :(   I made a video showing the model in slow motion, about 5% of normal speed..............and the ball is still falling off the loop !!!!!!!!!  The video also shows the ball leaving the top track and bouncing against the curved track down to the loop, which is what I expected.  You can use this link to watch the slow motion video.

Now what?

One possibility is that I did not scale the model large enough from the Linkage.EXE get the ball up high enough to have enough fall to go thru the loop and still have some momentum left to go thru 2 more curves back to home??

I went back to the Linkage.exe program and exported the .DXF file again, and imported into Sketchup.  The ball came into Sketchup as 6" diameter, so I scaled it back to 1" diameter ball.   The DXF file says the longest link on the RH side should be 10-39/64 and my design is 10 - 45/ I don't think it is a case of my model not giving enough drop.

Maybe the ball hitting the track when it comes off the top section is wasting energy that could be used to give more momentum to the ball going down to the curve?

Reducing Loop diameter from 3" to 2.5"

I bit the bullet and bent the loop down to 2.5" diameter, or 1.25 inch Radius...and Success !!!!!!!!!!! 

Below is photo of the very small loop and a video in slow motion showing the ball going all the way around the track ID :) 

The ball has some velocity left when it exits the chute!

Here is a link to the video.

Hopefully, last adjustment on the loop

I have 2 issues to fix before I continue the track after the loop.

1. Moving the marble very slowly by hand, I noticed it almost sticks below the ring connecting the 2 chutes. I need to increase the clearance just a hair.

2. Sometimes the marble goes through the ring, then falls to the base instead of going down the track. I think my track is too vertical and needs the angle moved a little


I got the diameter increased were the marbles were sticking in the ring, and I decreased the slope from the ring down to the loop.

I still don't have much velocity left when the marble exits the chute, so rather than trying to add 1 or 2 more horizontal loops, I think I will just curve back to the entrance, and make sure that works ok. If I can not make it back to the start position, that will be a bad thing!

Getting all the brass to work

I just put in a curve to get the ball back to the beginning spot.  Bad news, it would not make it all the way!!

So I raised the loop up so the exit was just above the top of the main base brass bars, and it started to return ok !!

Thought I had it whipped when I dropped balls from the 2nd pickup onto the track, they went down, around the loop, and promptly hit the arm where they go underneath the linkage  :(

Boy, everything is so touchy on this model, you change 1 thing, and many other things have to change !!

More Debug

Had quite a few balls leave top position, enter curved track to loop, but fly off and hit the base.  Could not figure out why until I put a 1 inch diameter dowel in top track and moved it to the ring. Found out my top track was not perfectly aligned with the down track, so ball was hitting track nearest you, then bouncing inwards off the track.  Repositioned down track using the 1 inch dowel, and almost 100% eliminated balls coming off the track :) 

But, I just don't have enough velocity on the ball when it leaves the chute to finish the horizontal loop to the home position :(  I tried raising the top drop off point by 2 inches using a piece of old wood track, it did not help the velocity exiting the loop at all.

You can use this link to see a video I made showing everything working ok, except not enough gas for marble to make it around the last horizontal loop to the starting position.

How to Fix it?

Raising the marble height by extending the 2 arms will not help, because I simulated this by adding some old wood track.

I could raise the loop, but then I lose drop height, which means I lose velocity coming out of the loop.

Tonight, I'm thinking that reducing the loop diameter is my only hope?  I got the marble to stay on the loop when I reduced the radius from 5 inches down to 3 inches.

Why does Tom Rector's Youtube model have so much more velocity?

Assuming our models are the same size, with same drop height, the only thing I can think of he has a bigger curve from the top drop off point to where it enters the ring and the down chute. I don't know what size his loop is.

By my slow motion video, by ball leaves the top curve and bounces off the vertical down track. Maybe this wastes energy??  Would increasing my curve radius give more speed?

I scaled his model from the youtube video, and his base length is 2.8 inches, and the top track goes down in to the ring at 1.1 inches to right of vertical upright.  My base is 13.5 inches long, so his top curve is 5.3 inches to the right of the vertical support.  My top curve is 4.5 inches to the right of the vertical support.

So my curve is just 0.8 inches shorter than his??  I can see my ball leaving my top track and bouncing off the vertical track.  If I increased my curve by about 1 inch, would this dramatically change the ball velocity coming out of the loop?

So Close, Yet So Far Away :)

After sleeping on it, I wondered if a slow motion video of Tom Rector's model would show the marble not leaving the upper track like mine does, and this causes his higher marble velocity.  You can use this link to see a slow motion video of his model, and it does not leave the upper track.

I can see my marble leave my upper track with the naked eye, but this link shows it leaving in slow motion video.

This web site creates a free 5% speed slow motion video file.

It is hard to believe that 0.8 inches differences in curvature explains the huge difference in marble velocity leaving the loops??

Below is photo comparison of Tom's model and mine.  He took his video at an angle, so the exact curve and placement of the ring is distorted from the angle.

More optimum marble path?

Concept verification

I don't know how to calculate the curve such that the marble does not leave the top track until it's velocity is pointing down, so it smoothly transitions to bottom curve with no energy loss?  Maybe I can make a test curve in wood or cardboard to verify the geometry before I make another metal track?

It also bothers me how poor my efficiency is with respect the physics guideline of drop height being 2.5 times the loop diameter and the real world guy that said 3.5 times.  I am at 13 inches with a 3" loop diameter, so my ratio is 13/(3/2) = 8.7.

I previously saw a Youtube video where people made curved marble tracks using cardboard. They wetted the cardboard so it would bend to the curve form easily, then they hot melt glued the track to a carboard base.  I decided to try out my new design in cardboard since it should be easy to do.

Cardboard Design of New Curved track

I dried the wet cardboard after the curve was hot glued on, in the oven at 120F.............but you can't heat too long or hot melt glue melts!!!!!!!!  [I learned on a previous project].  I designed the cardboard model to have a straight line at the bottom, plus 1.5 inches below the line, so I could clamp it in the bench vise. I tipped the cardboard inwards slightly when I dropped the marble, so it was assured of staying in the track.

I changed my Canon Powershot camera from 30 to 60 frames per second on this video. I used the free web site noted above to slow it down to 5% of full speed, and it appears the marble is tracking the first curve and exiting into the 2nd curve about as smooth as you can get it. You can use this link to see the slow motion video.

Much as I hate to do it, I'm going to remake the top drop off track, and the vertical track to the loop. Hopefully this will solve my problem of not having enough velocity after the loop to make it to the start point.  Based upon how far the marble dropped after it hit the bench vise, it should have more velocity going into the loop!!   The marble went 15 feet across the shop concrete floor !!


I printed off the new track pattern and then laid my old upper track on top of it.  My new track only goes about 3" to the right of the vertical support, while the old one, and Tom Rector's goes about 5 inches to the right !!!!!!!!!

I then took the paper pattern and set it in front of my actual model, and the down track will interfere with the RH upper linkage arm, that is why it needs to swung out further to the right.  DARN

It is literally "back to the drawing board" !!!!!!!!!

2nd Attempt

Pushing the RH side of the top curve out, makes it darn difficult to design the 2 curves. Here is my 2nd attempt for the chute designs, and I think I have the curve to the loop far enough to the right to avoid the top linkage arm now.  

I don't think this is that much different than my original design, I will  lay it on top and see.


I laid my original hardware on top of the  Sketchup print-out of my latest design, and it is not significantly different than my original hardware!!

So why not take my original design and make a cardboard model, and move the down chute away so when the marble leaves the top track, the trajectory hits my downwards chute nicely, instead of how it bangs now?

So if I accept that my marble must shoot off my upper track, why not use my existing hardware and keep moving the down chute to the right so it matches the trajectory of the falling ball, then attach it there??  Worth a quick try versus building 2 new chutes?  I may need to bend the straight section at top of vertical chute a little.

Ready to throw this project in the Vermilion River !!!!!

Decided to tighten up the Loop  from 3 inches to 2.5 inch diameter.  I also raised the height of the loop about 1" so their is fall left to get to the starting position.  But this reduces drop height about 1 inch.

After all that work, marble won't complete the smaller loop !!!!!!!!!!!!!                  RRRRRRRRRRRRRR

I may just rip out the loop and have a simple return back to start tomorrow, so I can build the worm gear, spur gear, and finish the model.

More Thoughts

As I thought about it, the only way Tom Rector got the shape of the 2 curves was to do the following:

1. On top curve, have it curved also in and out from the model, such that it delivered the ball in front of the RH vertical support.

2. Now there is room for the downward curve since it is front of the front RH vertical support.

Unfortunately, there is no angle on the Youtube video that shows this, but I studied it some more and I think this is how he did it.

I was surprised how much velocity I had from the cardboard test above, it seemed to be a lot more than I am getting.

Now, do I give it one more try with the loop, or throw it away?

Throwing in the Towel on the Brass Track

I decided to quit on using a brass track and loop, and switch to wood track because:

1. I spent many frustrating hours trying to get the top 2 tracks and loop to work right, to no avail.  My best guess is I don't have the transition right between the top piece and the the downward piece to the loop.  

2. My first attempt at brass track turned out very ugly.  Maybe using 1/8" brass rod is too thick and I should have used 1/16" rod.  My
  hat is off to Tom Rector and others on Youtube videos that make such elegant looking metal track.

3. My focus is building a model that showcases the 8 piece mechanical linkage, and the hand-cranked worm gear that runs it.  My focus is not on exotic metal track designs

4. I know how to do wood track and do it well, so it should look very nice.

5. The solder joints on the brass track are surprisingly weak. It will be very easy to damage the model taking it to shows and having kids operate it.  My wood track will be robust.

6. My wood track concept design is shown below.  If I need to slow down the ball some more, I can put in the "wiggle waggle" triangles in the long run to slow it down some more.

On wood track, you need a minimum of 3 degrees slope.  I am going to use 6 degrees on this one to allow for some margin of error.

Photos of set-up before I removed the brass track

Brass track removed and ready to start adding wood track

Finally ------- A successful day in the shop :)

Got a great start on the wood unloader and track today. Got it all done except the 2 straight sections of track, which should not take long to finish :)

You can use this link to watch the model in action. Right now, the marble exits the vertical tower at a bottom hole, which will be linked to wooden track tomorrow.

Wood track all done

I made a video of the model in can use this link to watch it.

Now on to the 2 wood gears and hand crank to finish the project :)

Worm Gear Blank

My design is for a 1.5 inch OD, but that is a weird size because 2 standard 3/4" thick boards gives you 1.5", but you loose some turning in the lathe.  I chose to cut off a piece of 2x3.5" oak board, then rip it to a 1.5 inch square.

I marked x's on both ends for the center, then hand turned it onto the small faceplate with a screw.  I put the tailstock center into the other end.  I ended up with a 1-3/8" OD blank, which I think is good enough to work.  My paper patterns at 0, 90, 180, and 270 degrees work regardless of the diameter.

I used a small piece of flexible cardboard to connect the marks and make the spiral curves.

Then to the Dremel. I used the green deburr tool a lot, because the sanding drums plug up pretty quickly and they quit cutting, and burn the red oak.  When I was happy with the form, I used the 1/2" Dremel drum sander to sand the teeth.

Then I used my 3/8" wood dowel with a slit sawn it, stuck in a small piece of sandpaper, used the drill press, and sanding down in the grooves better than I could reach with the dremel drum sander.

To drill the center, I used my jig for drilling in the center of dowels. On the c-clamp, I put in a piece of scrap pine across several teeth to clamp.

My design was for a 1/2" hole in the worm gear.  This looked to large, so I did a 3/8" hole with the Forstner bit, then finished in the vise with a long 3/8" wood drill, very carefully.

Next is mounting the spur gear on the model, then adjust the center distance from the spur gear the worm gear until it runs smoothly, then drill the 3/8" holes to support the worm gear.

How to Adjust center distance between spur gear and worm gear

On the hand-cranked worm gear I made a few years ago, I struggled to get the tooth mesh just right, and I actually got 1 support off a little from the other. Since then I have learned that hot melt glue is an efficient way to make a part that you want to remove and reposition.  I marked my starting worm gear shaft center hole on the 2 supports, then made a 1.5x1.5 inch square block with a 3/8" hole in it.  I hot melt glued the blocks onto the supports, by looking down the hole and visually aligned the hole in the block with the mark.  This worked well.

Too good of day in the shop

Everything was going perfect, made the spur gear, made the worm gear, got the 2 supports in place for the spur gear.  To test the worm gear height and  axle location, I hot melt glued on 2 blocks temporarily so I could adjust the fit.

But, when I tried to run the spur gear, I had way too much run-out of the axle hole to the OD of the teeth !!!!!!!!!!  As best as I could measure, I had 1/4" TIR (true indicator reading) using a tape measure. I guess my dowel drilling fixture did not work so great, the bit must have wandered down the hole.  I glued in a piece of 3/8" diameter red oak dowel, and I will try drilling from both ends next time to reduce the TIR.  Wonder how I drilled the other worm gear so good many years ago??

Here is a link to my video showing the excessive run-out on the lathe.

I sanded off the 2 ends of the 3/8" dowel I glued in so I could put the worm gear back into the lathe and pencil mark a circle on each end.  I then used my dowel drilling fixture and drilled from each end about 1.5 inches.  I then put the worm gear into the vise and used a longer bit to drill from both ends and connect the hole.  I got much lower run-out, not perfect, from drilling from both ends. You can see a video of this using the link.

Another "Learning Experience"

For this model, you will turn the crank clockwise as viewed from the RH side of the model. I need to convert this to CCW rotation to run the linkage in the right direction. The linkage will not work properly if rotated in the other direction.

I was all excited to set the shaft location for the worm gear and essentially have the model finished except for the crank handle...............but I found out I had the wrong "Hand" on the worm gear angle.  I did not think I needed to worry about the hand, because I could just flip the worm gear end for end to get my correct rotation.  WRONG, it does not change when you flip the worm gear end for end!!!!

When I made my first worm gear several years ago, on my Sketchup file that I created the patterns at every 90 degrees, I labelled the patterns "starting at LH side of gear blank", so I did the same thing here.  I googled it, and sure enough, you can have either a Right Hand or Left Handed Worm Gear!

Worm Gear OD

From my photos below, it looks like I could get more tooth engagement between the worm and spur gears if I increase my 1-3/8" OD.

Sketchup of 1.5 versus 2.0 inch Worm Gear OD

2nd Worm Gear

I think more tooth engagement is better, so I think I will shoot for a 2" OD LEFT HAND worm gear. If I glue up 3 boards, each 3/4" thick, I will get 2-1/4" square blank, which will come down to about a 2" OD after lathe turning.

If I glue up 4 boards, I will get a 3 inch square blank. Maybe I will do this process:

-make 3x3 inch square blank

-turn to round on lathe using small screw faceplate

-use dowel drilling fixture, drill from each end, connect 2 holes with longer bit

-put 3/8" diameter steel bolt in lathe chuck, tighten with washer and nut on tailstock end

-lathe turn to final size. Put tailstock live center into small drilled hole in end of 3/8" bolt to support better.

-mark and cut worm teeth

This should give as perfect run-out of the worm gear as I know how to make.

I used the process above, and ended up with a blank about 2-1/8" in OD.

To mark the gear at every 90 degrees, I used the same 4 slips of paper, but started from the RH side instead of the LH side, so the worm gear will move the mating spur gear in the right direction.

1st and 2nd Worm Gears

Power carving the teeth

You can use this link to watch me power carve the teeth using a Dremel.

Made simple gage to measure thread depth

2nd Worm Gear Installed

New Worm Gear in Action

You can use this link to watch a video of me hand turning the 2nd worm gear.

Run-out on spur gear

I made a jig with a center hole for the 3/4" spur gear shaft, put in a sanding drum on the drill press, the rotated the spur gear by hand, and ground off the high spots. I was surprised that I had as much run-out as I found using this method.


As I do the final assembly, I noticed my 2 pick-up fingers are note exactly in the right position anymore.  I made these with the hand crank temporarily in the front.  Now, I have the spur gear in back, and I'm guessing I have the location of the driveshaft just a smidgeon different now.  I may have to play with the location of the 2 spur gear supports and move them slightly............versus messing with the 2 pick-up fingers.   I probably should have started with the spur gear drive in its final location versus out in front.

Everything installed except hand crank

Back Up a little bit

Because everything builds from the crank that drives the mechanism, I decided to start over with respect to mounting the 2 bearings for the spur gear. I experimented with moving the bearing block to the right, to the left from the centerline where I had the temporary mount. I even shimmed it up a little, but no position would allow all 3 functions to work (pick-up the ball at home position, transfer the ball from 1 link to another, and drop it off at the top). So I mounted the 2 bearings on the same centerline as was on the temporary block and I will adjust the brass mechanism to match.

Almost There

Got it all built and it worked 3 times in a row............then the upper linkage arm broke at the solder joint :(

I think what happened was the angle changed on the top link, and I made the unload wood chute and stuff to the model in that condition. Darn, I thought soldered joints would be stronger than that. I ended up changing the angle of the top link to its middle pivot by about 10 degrees from the design, and bingo, everything started working ok !!!!!!!!

Model is Finally built :)

I am really pleased with how the worm gear works, I was afraid cranking the model might throw the marble off the lifting links, but it did fine.  Also, the 20:1 speed reduction ratio I picked is really good.  You really have to crank it to get a suitable lifting motion from the mechanism!!

Here is a link to the model in action.  Found out my S6 Samsung cell phone makes higher resolution video than my old Canon Powershot A610 circa 2004.

Updated Sketchup Design

I decided to keep my baseboard piece, but I will have to add on 1 inch on the left hand side.

Tasks Left Before Stain/Painting the Model

Add on 1 inch to the LH side of the base.

Make new angled support for LH track with 1 hole that I will use for carrying the model, my left hand finger will go into that hole.

Make new home position chute.

Make back vertical support from brass instead of wood.

Make short brass square and bolt it to horizontal frame, to keep RH vertical support from moving.

Make roof for marble drop chute.

Knocking out last tasks before stain/painting

Add on 1 inch to the LH side of the base.  COMPLETED

Make new angled support for LH track with 1 hole that I will use for carrying the model, my left hand finger will go into that hole.
COMPLETED.......... [I put 2 screws in old base plate, and 2 in new 1" wide strip, this helps to hold the strip against the old base, plus

I installed 2 horizontal screws as well. The screws should let the main base expand and contract ok with temp/humidity changes]

Make new home position chute.  Complete

Make back vertical support from brass instead of wood.......COMPLETED

Make short brass square and bolt it to horizontal frame, to keep RH vertical support from moving.
---Boy, the travel of the mechanism is extremely sensitive to angular changes in the vertical uprights on the RH side. Just a few degrees is enough to shift the interchange of the marble!!  The vertical piece always moves to the right, which makes sense if you look at the model, because the force is pushing it that way.  I should have used 2 bolts smaller than 1/4" each to keep it from rotating. I'm going to try using a wood triangle block on the RH side to see if that solves the problem.

OOOPs............the wood triange block did not work!! Turns out part of the cycle the forces push the verticle left, and another part push it right.  Bit the bullet and bolted to main horizontal rail roughly at 45 degree angle.........nice and solid now!   Only did front, and not back verticle, this should be good enough.........COMPLETE

Make roof for marble drop chute. Turns out that 3x3 inch is too small to saw in the table saw, each side at 10 or 15 degrees.  Gave up finished it on the horizontal belt sander.  COMPLETE

Updated Sketchup Drawing

YouTube Video on How to Make a Wood Worm Gear

I searched and did not find any existing videos, so I made one. You can use this link to watch this video.

Finished Model with 1 glitch

The interchange point is not working quite right. Figured out there is enough slop in all the linkages involved, on RH side of slop, doesn't work.........but on LH side of slop it works perfectly !!  I want the model robust so it keeps working over time, so have to figure something out.

I tried in oak to add 3/16" in length to the short link attached to the top of the vertical bar, and this I made a new brass link. On this link, I drilled (2) holes 1/8" dia in each of the 3/4" long 1/2" o.d. pieces, and used 1/8" bar in the holes. This worked better than the bigger 1/4" brass rods, plus the drilled holes help support it.

I got that new link done, and found the top pick-up arm solder joint had moved again !!!!!!!!   I resoldered it on both sides, but I may end up doing something else on that joint.  I could use a 3/4" or 1" OD piece that is 3/4" long for the pivot, then drill a 1/4" hole in it to better support the single 1/4" arm going to the pickup bucket.  I had to move the marble wood drop tower 9/16" to the left for a better and more robust drop off point.

Brass looking Paint

The solder joints look bad, so I found some paint that mimics a brass look. It really did a good job of making the model look nice.  I shined all the brass linkages using steel wool.

Sketchup Warehouse

I uploaded my design to the Sketchup Warehouse. You can use this link to access or download the model.

Finished Model

Youtube video of model in a

Closing Thoughts on this Project

I initially thought making the 8 linkages and getting them to work would be the hard part of this project.........boy was I wrong.  The hard part turned out to be making a brass rod track that looked nice, and would do a loop and 2 horizontal circles.  I learned that making track is not one of my easily found talents !!!!!!!  I had to revert back to making wood track, which is something I can handle fine.

I learned that basic physics say the drop height has to be a minimum of 2.5 times the vertical loop radius, ignoring friction.  Another person said in the real world, it is more like 3.5 times the loop radius.  My crude attempts failed miserably at 9 inches drop height for a 1.5 inch radius loop, or a ratio of 6:1. I don't know why my efficiency is so bad?

If I made another one, I would start with the crank in it's final position. This model is extremely sensitive to the crank position is why.

Rather than 1 drilled and tapped 1/4-20 UNC bolt in each pivot point, I should have used 2 of the #10 bolts in each one, to prevent them from rotating in service.

While researching stuff for this project, I found 1 trick I did know, but had forgotten.  Whenever I make wood spur gears, I should put them on a dowel the same size as the axle hole, and make the OD perfectly concentric to the axle after I have scroll sawed the teeth. No matter how hard you try, the axle hole is always a hair off from true center when you mark and drill it.

Drilling and tapping brass, at least 1/4-20 UNC threads is relatively easy compared to steel.

I was surprised how weak my solder joints were. They really won't take any bending or tension forces at all.

I did not try to solder the 1/4x1/2" brass bars together, I chose to use drilled and tapped bolt connections. I am not sure my soldering iron could get both parts hot enough for a good solder joint.

The 2 pick-up fingers are very difficult to hold in place for soldering. I got some modeling clay from mcmaster-carr and that helped.

My method of making a wood fixture for each of the 8 links worked well with respect to making sure they are as perfect as possible.

Found out my Samsung S6 phone takes better video than my Canon Powershot A610 camera.  Now  I just need to get a holder for a tripod  for my cell phone.

I learned there is such as thing as a right-hand and left-hand worm gear !! Make sure you pick the right one if your model input and output directions are important.

I also found that Klinspor Woodworking sells larger diameter power carving attachments than I have. They have a 1/2" and 3/4" diameter version. Mine is only about a 1/4" diameter. I think I will order and try some of the larger ones.

Another thing that threw me for a loop was that I built the 2 pick-up fingers so their interchange point worked great, but a couple days later they did not. Turned out the solder joint had twisted out of position on the upper link!!  

I used brass for the mechanism because I thought wood would not let me get the pick-up interchange precise enough.  After doing it in brass, I am still not sure if I could have made it work with wood links or not.

If I made another one, I might try drilling a hole in the pivot piece, then inserting the rod into the hole and soldering it.  I tried this with 1/8" rod on a small link, and it worked great.  The 2nd pick-up arm seems to see the most stress, and would be a good candidate to use the drilling method on.

Project Cost

I usually don't track the cost of my woodworking items, because the cost is usually insignificant.  But on this project, brass is so expensive that I decided to tally up the cost. As expected, most of the cost was the brass items.

April 12, 2021 Update

I was short about 4 of the brass cap screws, so I used regular steel bolts, until I could order some more brass ones..........which I did.

When I went to change out the bolts in the back horizontal bar, I got into all kinds of trouble !! I had to remove the whole brass mechanism from the wood base, which is usually not an issue, but putting it back together I had trouble, and the arms did not line up at the interchange point again !!   Then the solder joint broke on the 2nd curved piece on the end.  Had to print out new pattern, make jig, and resolder.  Finally got it all working again.

Boy, this is the most sensitive model I have ever built !!  Because I used 1 of the 1/4" bolts to hold the little base pivot points to the bar, if you get these off a few degrees, the interchange and pick-up point start point change!!  I would have been much better to use #10 bolts instead of one 1/4" bolt!!

To keep the nuts tight on the moving joints, I tried a new trick, I used fingernail polish as my "Loc-Tite".  We will see how this works.

Then holding a rag under each moving joint to catch the excess, I WD-40'd all the moving joints.  I also put on another layer of Johnson's wax on the worm gear.

I'm calling it completely done, unless something mal-functions.


I continue to operate the mechanism each day, to make sure it works flawlessly.  Found that a couple of marbles hang up in drop tower.  Marble are sorted when they are made, so they not exactly 1 inch in diameter.  The horizontal distance between the bottom bar and sidewall was right at 1" and not 1-1/6" per the design.  Took tower apart and drum sanded a little thickness away using the Dremel.  Also tapered the 2 edges going into the bottom chute, so none can set on the lip against the plexiglass.  Hopefully, this completes debugging.

May 5, 2021 Update

I have taken the model quite a few places to demonstrate it.  I had a lot of friction on the crank handle, so using an artist's brush, I carefully lubricated the handle with mineral oil.......using a clean rag to soak up any excess.  This made it work smoother. I also brushed on some mineral oil onto the worm gear, and this reduced friction.  

I may have to repeat this from time over the future years to keep it running smooth.

August 2021 Update

The 3rd member going from left to right on the front view, broke the solder joint, so I had to repair it.  I learned that solder joints are not that strong in a mechanism like this.  If I did it over again, I would use 2 pcs of 1/8" brass rod with drilled holes to connect the members, instead of soldering 1/4" rod.

This model is so sensitive, I guessed it would need some "tweaking" after I repaired the 1 linkage, and sure enough, I had to do some tweaking including bending a link just a hair, and moving the marble chute a hair outwards where the brass mechanism drops it.

April 2022 Update

I have taken this model to several public exhibitions, where kids get to play with all my wood models.

About each time I take it out, the kids play with it for a while, then one of the soldered links breaks at the solder joint......... URRRRRRR

This last time was the link that is driven by the gear, and runs the RH side of the model.  It broke the solder joint in 2 places !!  I drilled and added some 1/16" brass pins to stiffen that member up.

Someday I may have to buy a MAP torch and silver solder the joints. My research says that silver soldered joints are about 5 times as strong as lead soldered joints.   I would have to heat and clean all the old lead solder off, maybe by wiping the hot piece with a rag, then silver soldering.

I still wonder if I could have used wood instead of the brass for the joints, and no issues with solder joints breaking!!

April 2, 2022 Update

I decided to go ahead and try designing/building a pawl that would prevent kids from turning the model in the wrong direction.  You can damage the model if you try to run it backwards.

After I got this design done, I thought, gee, I could have had the pawl rotate on the triangular support piece holding the vertical piece up (yellow), and then would not need the 2 axle supports and dowel.  Oh well, live and learn !!

Video of Pawl in Action

You can use this link to watch a short video of the pawl in action.