The Manhattan Project: a Mechanical Cocktail Mixer

The seed for this project was the desire to mechanically measure specific amounts of liquid in a way that was both adjustable and easy to clean. In the past, I have used measured pour spouts on cocktail machines, and they work fine, but they are not adjustable and occasionally they just stop working.

I started by thinking about the Japanese Shishi-Odoshi ("deer scarer") bamboo fountain feature, which measures liquid against a counterweight and pours when the liquid reaches a certain level. The system is wonderfully simple, with one big caveat: the liquid never stops pouring.

So I began prototyping on a piece of plywood with a plastic cup. I new that the receptacle had to be lightweight, so that the liquid would make up the majority of the total weight and be easier to fine tune the "cut off" point for the pouring. I discovered that the pouring mechanism would actually need four parts:

1. The bottle of liquid above with a flanged pin valve, which releases liquid when pushed up. I made a stainless pin and used a little piece of silicone sheet as a gasket.
2. A receptacle below. The receptacle needs a hook to latch onto the counterweight arm. I made a little four-bar set so that the cup would pour where I wanted it.
3. A counterweight arm with a little ledge that the locking arm can lock onto
4. A lock arm that pushes the bottle's pin valve up when activated, and locks onto the counterweight arm.

When the cup is lifted, it pushes the Lock Arm up, releasing liquid. As the Lock Arm lifts, the Counterweight Arm lifts also and the two lock together. The hook from the Counterweight Arm is now holding the cup up. When the cup fills with enough liquid to overpower the Counterweight Arm, it pulls free of the hook and pours, while the Counterweight Arm pulls free of the Lock Arm and the bottle is sealed again.

It pretty much works. A little leaky, but overall a great first step.

After the initial Proof-of-Concept prototype, I wanted to make a more formal test. I ordered some parts, including a bottle from McMaster Carr, and began modelling the pouring mechanism in Solidworks.

The bottle, Mcmaster #4787T64, was a Kimax lab glass bottle, and made me think of a Mad Scientist Chemistry Experiment type theme. So I started planning how I could make the final product fit this theme. I liked the images of chemistry kits with different components clamped to vertical stands, and lots of busy parts crossing from here to there. This theme lends itself to producing the machine in mostly stainless steel, which would make it easy to clean and maintain, so I ran with it.

I found another McMaster part, Quick Clamp Sanitary Tube Fittings (#45195K44) that would work great for clamping to 3/4" diameter tubing, and could have parts welded to it. You can see it modeled in the CAD screenshot above.

I decided that the machine would make Manhattans, mostly just because I like Manhattans. The ingredients are, roughly: 2 parts Bourbon, 1 part Sweet Vermouth, and a dash of bitters. I knew I would use two "Pouring Modules" for the Bourbon and Vermouth, so I began trying to figure out how to deliver the bitters, etc.

I had a general idea that I would need to attach 4 modules to the frame: two Pouring Modules, a Bitters Module, and a Mixing/Stirring Module. I wanted the machine to be operated from a single crank, and I decided that the process of making a drink should take about 1 minute. Given that people generally turn cranks at about 60rpm, I wanted to make the crank input drive a worm, with a 60 tooth worm gear that would activate each part in turn.

I began constructing the frame while I was still conceiving and designing parts. I had access to a nice Omax 5-Axis Waterjet table, so I began cutting parts out and ordering more hardware. I bought 3/4" and 1" diameter 304 stainless tube and bent corners and welded up the frame (see image), including the corner at the lower right where the drive parts are. I welded threaded rods to the the Tube Fitting Clamps to attach modules to the frame. The clamps allow parts to be easily moved around and adjusted.

I 3D printed many of the small components for the Pouring Modules on an Objet Connex 500 in ABS-like material (for strength). I used stainless rod epoxied into 3D printed tee fittings to create the Four-Bar arrangement that would pour the receptacle.

Both the Counterweight Arm and the Lock Arm have weights attached, as well as other small metal wire parts to reduce friction and wear when they rub against each other. I drilled and tapped the 3D printed parts for threaded rods to hold the weights. I ended up soldering brass knurled nuts onto threaded rods to hold both the weight and the little hook onto the Lock Arm.

Because the Quick Clamp brackets that hold the Pouring Module onto the frame are off-center, the funnel bracket below also had to be constructed off center. You can see in the images how I jigged these parts up for TIG welding. They are made of 1/8" diameter 304 stainless rod.

The receptacle is a 3oz. disposable plastic cup held in a little 3D printed bracket that also has a hook to latch onto the Counterweight Arm. I chose the cup because I needed this part to be food-safe but also as lightweight as possible, as well as removable for cleaning.

I began with a stainless cocktail shaker. I wanted a stirring wand to sit inside the cup and stir the ingredients with ice. Then I wanted the ingredients to pour out of the bottom of the cup.

I purchased a large diameter bottle filler (like this one: http://www.amazon.com/Plastic-Spring-Tip-Bottle-Fi... removed the tube, and turned a mating countersunk piece in aluminum on the lathe.

Then I drilled the bottom of the shaker cup, and made a tool to flare the hole for the countersunk fitting. I added a thin piece of silicone as gasket between the cup and the countersunk part, and when assembled it was watertight! Perfect.

I created a bracket for the cup which allowed it to be easily removed for adding ice and cleaning. The bracket also has a 3D printed part that holds the tubes from the funnels in place just over the cup, as well as a bearing sleeve for the stirring arm, which is held in place with a thumbscrew.

The whole machine is operated by turning a crank at the lower right corner. The crank turns a worm, the worm drives a worm gear, the worm gear is attached to a shaft. On that shaft sit four cams and a partial gear. The cams are just clamped to the shaft, so they can be rearranged and adjusted easily. When each cam comes around the top of its path, it pushes a lever. The lever pulls a cable, which rides inside a tube to one of the modules, where it pulls whichever part activates that module.

The first cam lifts the Receptacle on the Bourbon Module. The cup latches onto the Counterweight Arm and begins the pour.

The second cam lifts the Receptacle on the Vermouth Module. The cup latches onto the Counterweight Arm and begins the pour.

The third cam advances the carraige on the bitters module, pushing a few drops out of the pipette into the Stirring Cup (I didn't get any good pictures of this part... sorry).

Then the Partial Gear begins turning the stirring assembly; a sprocket and chain make the stirring arm rotate inside the Shaker Cup for a while.

Finally, the fourth cam pulls a little lever under the Stirring Module that pushes up the nipple on the bottle filler. The finished Manhattan pours into a cup below. A small pin behind the worm gear retracts and releases a little bell, announcing your completed beverage.

Here is a rough video.

The Manhattan Project was also shot by a photo crew from Popular Mechanics and featured in their article about the Pier 9 Workshop at Autodesk: http://www.popularmechanics.com/technology/how-to/gadgets/inside-instructables-kooky-creative-warehouse-wondershop#slide-2

There are a few improvements that I plan to make, such as using thicker cables between the cams and modules, and fine-tuning the pours, but by and large the machine is wonderful and makes a great Manhattan.

In the future I do look forward to creating a "Garnish Delivery" mechanism (Manhattans usually have cherries, which seems plausible, even though I prefer orange peels).