Hey everyone, I hope you've been enjoying your summer with some tasty beverages. Things have been ebbing and flowing in Party Robotics land over the last couple of months. We ran out of several things some weeks back, but now we're stocked up on everything again, and the best news is that Bartendro Dispensers are now 10 bucks less! Did we change anything about them? Nope! As we've been refining our process, and understanding our costs better, we've decided that we would pass the savings on to you. It might not seem like much, but every little bit helps when you're trying to build a big bot.
For those in the Bay Area, we've also officially signed up for the Cocktail Robotics Grand Challenge, happening in about a month at the DNA Lounge in SF. Come root for your favorite robots. The event is on a Sunday from 5pm to Midnight and tickets cost $10 in advance.
When making a cocktail dispensing robot, peristaltic pumps are clear winners. You can read more about why in our previous post
. We've designed and manufactured several of our own peristaltic pumps, and now understand the design nuances and trade-offs. We've spent hundreds of hours researching, dissecting and testing both home-made and off-the-shelf varieties. Why are some pumps $300 and others $100? What does that extra money get you? Which ones are good enough?
Like most engineering problems, this one is multi-variable and application specific. But generally, from what we've found, there are at least 8 things to understand when looking for a great pump.
1. Flow Rate / Number of Rollers
2. Occlusion (fixed vs variable)
3. Ease of Maintenance / Tube Replacement
4. Tubing Catch Mechanism
5. Motor Selection
6. Encoder Capability
7. Material of tubing, rollers, pump housing etc.
While I could practically write a book on the subject now, I'll distill these ideas into a few sentences each:
To maintain a seal, the minimum number of rollers is 2, although there are exceptions to this. The more rollers there are, the slower the flow rate, but the less "spurty" the flow. We used to think that an odd or even number mattered, but they don't, flow rate and flow smoothness are the factors.
Occlusion. This one is tricky and really the heart of the whole design. As a positive displacement pump, we rely on every revolution of the motor being a known volume. Occlusion refers to how much squeezing is happening to the tubes. So, if a tube's wall thickness is 0.125", sandwiched together it's 0.25." The space between the roller and the internal wall of the pump needs to be smaller than this, say 0.242." That means there is an occlusion of 0.008." There is a rather large range for this number and that is determined by a host of things, namely tubing material, pump housing material, pump roller material and type of motor. Friction is playing a large role here and cannot be discounted. Some pumps have a spring loading down the "shoe" or surface that the tube is pressed against. This insures that over time, as the tube wears, or as lower tolerance tubing is used, there will be enough occlusion for proper operation and no leaks.
Some pumps make tube replacement a breeze by not requiring extra tools. Those pumps typically require more moving parts making them considerably more expensive. A nice option, but not mandatory.
That sneaky friction guy that we just mentioned is a real pain when you discover that the tube tries to walk out of the pump by inching with every revolution. There needs to be something in place to prevent the tubing from moving during operation.
There are DC, AC and stepper motor options with variable controlling complexity. DC motors are typically the easiest to use.
With the exception of one, none of the pumps we've seen support encoders. Encoders count the number of motor revolutions, for accurate monitoring and control. Optical and magnetic are the most common.
Tubing Material will likely be its own post one day, for now, just know that options are vast and usually liquid dependent. We've done our homework to make sure you get the best results. For rollers and pump body the concerns are coefficient of friction, wear life, cleanability and cost of manufacturing. Metals like stainless steel are easy to autoclave and will last a very long time. Plastics are lightweight, usually give more visibility and are more economical.
Which brings us to cost, an important item to consider especially if you intend on multiplying by 15 to build a big cocktail robot. Medical pumps have several things going for them like decades of testing, FDA approvals, numerous adjustments and guaranteed life-critical operation. We're just having fun here, so if there are a couple of extra drops of vodka in our drink, we'll live! Cost is also, as always, volume dependent. For example, one Watson Marlow pump (on the left) is $239 and that's not including a motor or control electronics...just the pumphead. In volumes of 100 they cost much less at $135, but still way more than is reasonable. There are a couple of low quality peristaltic pumps out there in the $25 dollar region
, but you are loud, slow and don't have encoders, so your dispensing won't be repeatable.
More great reading about Peristaltic Pumps
at the Wikipedia page.
The pumps we've evaluated are made by (from left to right) Watson Marlow, Masterflex, Anko Products and Welco.
Care to share any of your peristaltic pump experiences? What other methods have you used to dispense liquids of varying viscosity and chemistry?
are not new, they are pumps that mimic peristalsis, a biological mechanism most commonly depicted as the muscle contractions of a swallowing esophagus. The vast majority of these pumps serve the medical and pharmaceutical industries. This does not bode well for hobbyists because the pumps are priced at an average of $300 each. When attempting to build a cocktail machine that dispenses over a dozen liquids, the cost becomes prohibitive.
"Why use a peristaltic pump in the first place?", you might be thinking. Well, for many reasons. The first and most obvious reason is maintenance. Since liquid does not pass through any moving parts, it becomes easy to swap out or clean hoses when necessary. In the same vein, reliability. Syrupy liquids like grenadine and Bailey's tend to gum up mechanical pumps and valves pretty quickly. Sanitation then becomes another issue when liquids become trapped in the crevices of a mechanical assembly and become impossible to clean. Finally, accuracy and simplicity are the final components that put the nail in the coffin on other approaches. While gravity valves and simple diaphragm pumps are a possibility, there is still a big issue with dealing with different densities and viscosity of liquids. Since peristaltic pumps are positive displacement, meaning they dispense the exact same amount of volume with every revolution of the motor and gearhead, it simply becomes a software problem to account for the number of motor rotations for accuracy. This is much easier and cheaper than fashioning some sort of flow rate meter.
When designing the pump, several factors need to be taken into account such as the occlusion of the tubing, the ease of assembly/dis-assembly of the tubing, maintainability and modularity. The picture shows the parts for the first batch of peristaltic pumps I made. There is a plate that the motor mounts to, a hub with sleeve bearing rollers, a shoe which is replaceable and provides the right amount of occlusion, and a support for the hub that prevents lateral motions from causing it to run to eccentrically. All in all they can be built for under $100 given the right amount of tools and resources.