Picture of How to make an ice ball maker
Do you want an ice ball maker like those sold online, but don't have $800+ to spend?  Then make your own!

Don't know what an ice ball maker is, or how it works?  Then check out the last page of this Instructable for a video of it in action.

You'll need access to a machine shop, and specifically a CNC machine.  TechShop provides all the equipment you need for a monthy fee.  I made it at TechShop, and you can too.

Ice ball makers work by melting a large block of ice into the proper shape.  It accomplishes this by having two large blocks of aluminum (aluminum has high heat conductivity and capacity), each with a hemisphere carved into a side, slide together to form a single continuous chamber on the inside.  What's left is a perfect sphere.
Remove these adsRemove these ads by Signing Up

Step 1: Materials and Equipment

Picture of Materials and Equipment
The basic material is two large blocks of aluminum.  You'll need to pick a size based on the size of the ice ball you want.

You'll need blocks that are in total at least 10 times the volume of your sphere.  This is so the aluminum properly melts away the rest of the block.  My ice ball maker produces 70 mm diameter ice balls, and the aluminum blocks are each 125x125x75 mm.  That's a 13-to-1 ratio, and works pretty well.

You'll also need stainless steel rod.  These serve as the bottom vertical guide shafts.  I used 3/8" SS rod.

Last, you'll need plastic rod of a larger diameter.  This piece slides over the stainless steel rod.  I used 3/4" teflon rod, but teflon turns out to be hard to machine.  I hear that Delrin machines well.

The only extra material is whatever you think will make the final product easy to use.  I added rubber feet and a cabinet knob for lifting the top.

For equipment, you'll need a CNC mill and a lathe, and basic tooling.  The inner hemispheres will be cut with a ball end mill, and you'll want as large a diameter here as possible.  I used a 1" ball end, but smaller diameters will work as well.  You'll also need a tap and die set.

1-40 of 94Next »
praylax5 months ago
Can u sell me one of these please
Franzplanz1238 months ago

waste of aluminum

have you ever tried freezing waterballoons?

mikejs3 years ago
New goal: get a 3D scan of my head, and then mill that into some aluminum. Perfect ice faces, every time. Thank you!

theres an idea

Brian9961 year ago
How long do you have to wait between moulds?
Nuclearpwr2 years ago
I am really on a mission to make these really cool ice balls...hope you can help me out here....

I do not have access to any machining equipment and even if I did, I am not a machinist...would you consider making one for me? I would of course pay you for the material, machine time and your time.

Thanks again,

Marko in Mass
Nuclearpwr2 years ago
You are my hero. I've been trying to make clear ice balls for a while and your post does it all...clear ice and the ice ball machine...and a whole lot cheaper than the on-line units.

Question : What size piece of clear ice do you need to use in the unit? Does the size of the clear block of ice matter...assuming it is big enough to fill the sphere...and does the shape of the clear block make any difference.

I found a small mini-version of the on-line model and when I use mine, sometime I get flat spots on the ball...and I think it might be because I am using irregular shapes of clear ice...pieces just broken off a larger pc of clear ice.

Thanks again and great instructional video.

ToolboxGuy2 years ago
Way cool ! Given that you have a CNC at your disposal, I am surprised that you didn't add any runoff channels/gates, to keep the countertop dry. Since you don't want to corrupt the shape, just make an outer ring around the shape, and an outlet channel.
I would love to make one of these for ice formed whiskey glasses, say four at a time, or perhaps shotglasses in quantity. Shape then freeze until the party. Have to make grips/indents so it doesn't become too slippery.

Perfect for pool parties - and very little to clean up, at least for those of us who drain our glasses.
spcutler (author)  ToolboxGuy2 years ago
That's a great idea! I was originally thinking in terms of having a small hole in the bottom that the water could flow through, but that would, as you say, corrupt the shape. And I'll admit, drilling tiny holes in a nearly-finished project always scares me! Your idea with the channels solves both problems. I was already planning on making some more, and I'll be sure to try that out if I do.

One thing to watch out for when refreezing--because there is still a film of water on the surface, you will tend to get marks corresponding to the container you put them in. So it looks a bit imperfect, though after 30 seconds in the glass the marks will melt away.
spcutler (author)  ToolboxGuy2 years ago
That's a great idea! I was originally thinking in terms of having a small hole in the bottom that the water could flow through, but that would, as you say, corrupt the shape. And I'll admit, drilling tiny holes in a nearly-finished project always scares me! Your idea with the channels solves both problems. I was already planning on making some more, and I'll be sure to try that out if I do.

One thing to watch out for when refreezing--because there is still a film of water on the surface, you will tend to get marks corresponding to the container you put them in. So it looks a bit imperfect, though after 30 seconds in the glass the marks will melt away.
jayhitek2 years ago
Can you post the G Code?
eyong22 years ago
Hi im from malaysia and Im trying to make an ice ball maker using your steps shown above. May i get more information from you like from "Step 3 Machining the Hemispheres", what sort of sizes of the end mill and ball end mill should i use for the cnc with the DFX file u attached?

spcutler (author)  eyong22 years ago
My opinion is that it doesn't matter too much, but the larger the better. I used a 3/4" end mill for the roughing and a 1" ball end mill for the final pass. These days, I would use a real roughing mill instead of a standard end mill since there is so much material to remove.
jayhitek2 years ago
I will buy one from you.
Well its awsome
what would happen if you tryed to use strofoam or somthing like that?

Great stuff. You wrote that aluminum has high heat conductivity and capacity. I was considering milling this in Steel. Do you think steel is just wrong for this?

Many thanks.
spcutler (author)  superpositron2 years ago
I would avoid steel for a few reasons. The two big ones are, as you said, heat capacity and conductivity. Per gram, steel has about half the heat capacity as aluminum, so the unit would need to weigh at least twice as much to work as well (it could be a little smaller, though). It also has only a fifth of the heat conduction, so it would work very slowly.

The other big problem, as you might expect, is corrosion. Even without anodization, aluminum doesn't corrode in any significant way in the presence of water. Steel will of course rust if you have no protection on it.

Steel is also harder to work with, and isn't necessarily cheaper when you take the heat capacity into account.

You could use stainless of course, but that's far more expensive, even harder to machine, and only fixes the corrosion problem.

The only other material that I would really consider is copper, and that would be a situation where cost isn't the dominant factor. It's impossible to beat aluminum in bang-for-the-buck terms.
Is there a reason you specify a 10:1 ratio of metal to ice?
spcutler (author)  illusionistpro2 years ago
Warning: geek out alert.

The short answer is that you need enough thermal mass to melt through all the ice, and 10:1 is about the right amount.

Long answer:
To melt ice, you need to get through what's called the latent heat of fusion. For ice/water, this is 334 joules/gram. The density of ice is 0.917 g/cm^3, so this comes to 306 J/cm^3

This heat comes from the aluminum (in the short term). Aluminum has a heat capacity of 0.9 J/K-g, so at a density of 2.7 g/cm^3 we have 2.43 J/K-cm^3. Room temperature is 25 C, and we can only go down to 0 C before it stops being able to melt ice. Therefore, the aluminum can supply 61 J/cm^3 of heat energy to melt the ice.

306 J/cm^3/61 J/cm^3 is just about a 5:1 ratio. But that's a bare minimum: the ice ball maker works very slowly when it gets close to 0 C, and the ice will start at a temperature <0 C (depending on how cold the freezer is), and there are some extra holes and such that take away some capacity. So in practice we need a little extra leeway, and 10:1 makes for a nice round number.
firesirt2 years ago
this is awesome. I think I might try something like this, only casting the aluminum rather than cutting it.
spcutler (author)  firesirt2 years ago
Good luck, and let me know how it turns out! My first thought is that the surface wouldn't be smooth enough. But on the other hand, the surface of the ice ball melts fairly quickly and would probably leave a nice surface in short order. So it might actually work really well. Have fun!
r_harris23 years ago
Very interesting project. You know, this could be done on a lathe, even a manual one. You could either make a custom cutter, or build a radius-cutting attachment. Then mount the block on the lathe faceplate (or 4-jaw if you have a big enough one). I think a block of this size could just barely be mounted on a mini-lathe's faceplate.
blastin063 years ago
I think this is an awesome project and I commend you on all the work and learning you did. With that being said, for future possibly since you have to use a CNC anyways why not use a CNC lathe and turn multiple steps into two or three. Using a three jaw chuck you could use a round stock, shave the sides, face off the front and also cut the sphere shape. This all could be done in one step and most likely with out the use of a CAM package. Leaving the guides to be drilled on a knee mill.
spcutler (author)  blastin063 years ago
Unfortunately, I don't have access to a CNC lathe. TechShop has several CNC machines, but a lathe is not one of them--they just have manual lathes (with DRO displays).

If I make another I may try out using round stock. It's actually pretty hard to make a perfectly square block, even on a good mill!

Thanks for the support!
My compliments on an excellently designed and well made part... but actually making a "perfectly square block" is rather easy. I have been a machinist for 30 years and can do it on even the most humble mill. Good setup and one little trick is all you need. As long as your vise is flat to the table travel in the x & y axis, it's solid jaw parallel x axis, and the spindle & solid jaw are perpendicular to the table it can be done in 7 steps with 3/4 ball with a flat on it and a parallel. I have won many drinks proving it. Maybe I should do an instructable on it if there is any interest. Or maybe just for the fun of it. Just not sure this is the right forum to post on.
I'd *LOVE* to see an instructable on this! The only mill I have access to at my friend's house is definitely a very humble mill.
spcutler (author)  imatoymaker3 years ago
I would love to see an instructable on it, and think it would be quite popular with folks like me who like to do machining as a hobby but don't have the decades of experience to pick up tricks like what you described.

I managed to do the first 4 sides without much trouble but the final 2 sides were more difficult. I ended up fiddling around with an L-block (is there some better name for that?) to get a side perpendicular to both axes, but I'm sure there's a better way.
skrubol3 years ago
Why is squareness of the material important? To me, the whole operation of milling the face flat, cutting the hemisphere and boring the alignment holes should be able to be performed without re-mounting the workpiece. As long as you can hold the piece down securely without interfering with any of the cuts, the shape of the piece shouldn't matter.
spcutler (author)  skrubol3 years ago
You are almost certainly correct, and I mostly blame my inexperience.

One small problem is that the CNC machine is only available in 4-hour increments. Although with an optimized process I probably could have finished a full block in that time, with me learning as I went it took much longer than that. Given this, and the fact that I needed the blocks to be square in the end anyway, I decided to make that the first step in the process.

If I do it again, I'll probably look into doing the squaring, planing, hemisphere boring, and drilling all in one pass.
Well that makes sense.. Making something square is challenging for a novice machinist, so it's a worthwhile thing to try to do even if it's not necessary.
big-jamie3 years ago
that is absolutely gorgeous, the mould and the finished ice ball. i really wish i had access to a cnc machine =(
rimar20003 years ago
Beautiful ice ball!

A doubt: the holes in the photo of step 5 seems out of phase. It is to say that joining the hemispheres, the holes don't match. Does you changed them?
spcutler (author)  rimar20003 years ago
Thanks! The holes are actually in perfect alignment, probably less than 0.0005" error. Getting this right was one of the most important steps so I drilled the holes for each half without removing anything from the vice. If the the holes look out of alignment, it is probably a matter of the photo's perspective.
To be fair you did tap the holes which gives you some intrinsic misalignment. If you do another I would suggest using a press fit. Also you mentioned using an edge finder, if you want better accuracy you can swing the block in with a dial indicator, this is commonly done in mold making. Lastly a spot drill will give you much better results than a center drill, center drills are typically just for lathes and due to their shape don't do well in CNC mills, granted your aren't drilling a lot of holes, but you break one of those suckers and it'll ruin your day. I was wondering if you drilled the holes in a separate op or if you were able to get them into your program, I know the tormach doesn't have too much z travel, but even just spotting them in the same setup can give you much better accuracy.
spcutler (author)  koebwil3 years ago
I forgot to get to your last question:
Although I didn't incorporate the hole drilling into the program, I was able to use the Tormach for both the hole drilling and tapping. I did this separately from the hemisphere cutting, but that is not a big deal--the hemispheres could be a couple thous off and no one would notice. But each pair of holes did need to be the correct distance apart and so I did those in one step.

If I do make another, I'll see if I can incorporate the drilling into the program. I was using peck drilling since the holes were so deep and there was a lot of volume to remove, and it became quite tedious (and error-prone) after a while.

You're right that there isn't a huge amount of Z travel on the Tormach, but it was sufficient for my purposes, even with my 2" height gauge.
spcutler (author)  koebwil3 years ago
True, the tapped holes seems to have some intrinsic amount of wobble. I didn't realize this going in because of my inexperience, but it seems you're right that a press fit would work better.

Thanks for the advice on the spotting drill. It seems that there is some debate as to whether center drills are appropriate for this application or not, but in any case spotting drills are a good choice here.
Pardon, I am an old fool man.

I saw the pieces simply rotating around a vertical axis, the holes do not match, but I realized now that as they are perfectly symmetrical, you also have to rotate them horizontally.
1-40 of 94Next »