How do centrifuges work (that is, the mechanism)?

Factual Questions
1

How do centrifuges work?
I’m not asking like “how do they make the solids settle out at the end?”, I get that, that’s intuitive and easy, everyone understands that. You spin something long enough, the centrifugal force* will settle things out.
*and yes, I know the difference between centrifugal and centripetal force and that centrifugl force isn’t “real”

What I’m asking is, how does the MACHINE work? Specifically for the lab-type centrifuges that can spin up to 20,000 rpm and higher. Surely it isn’t a mechanical drive with a gear up ratio. But going up to 20,000 rpms needs some special mechanism, both for the forces, and to go faster in the first place since electric motors are limited by the frequency of the AC current (60hz = 3600 rpm max, since that’s as fast as the poles will change, no?)

I was looking stuff up and found centrifuges with hydraulic drives, but those were industrial centrifuges handling lots of material.

Ultimately, I need to make/have made centrifuges that can spin 20,000 rpm +, handling liquid, but not tiny lab amounts, larger amounts, like 3-8 gallons.
Not that I’m an engineer, and not that it needs inventing since this stuff already exists, but if I were designing it, I would guess the best way would to increase the frequency of the current and create a few more phases with a special transformer set up (tesla coil? but I don’t even know how a tesla coil even works I only know what it does) so that an electric motor could just by itself eventually spin up to 20,000 rpm. Or I’d “gear up” with “gears” made of magnets on the rims instead of teeth, so they never actually touch. Obviously they’d have to be highly exactly machined, as would any design realy I guess, dealing with those speed.

So how do these ubiquitous lab tools usually work?

2

A DC motor has no speed limitations, except for mechanical ones.
I have a Dremel tool which goes to 30,000 RPM.

PCB drilling spindles go to 200,000 RPM or more. These use high-frequency AC.

3

OK, so is that how it works, just a rectifier and a DC motor?

4

This ultracentrifuge uses a “direct drive induction motor.”

I’m not a motor expert, but I suspect it’s the same type of motor used on the high-speed drill spindles.

5

induction motor seems to be the same or similar to 3-phase motor. Same basic concept, but without permanent magnets, and the magnetism provided by induction. Again, in this situation the rpm is limited by the frequency, but looking all this stuff up apparently that indeed what they do is change the frequency with electronics to change the speed

And “direct drive” makes it sound like , well, direct drive. That is, just a motor connected to a shaft connected to the centrifuge’s spinning thingy

6

Drills and Dremels used to be DC motors, but these motors are heavy and inefficient. More popular recently are brushless DC motors. These are very similar to synchronous AC motors, in that a controller must switch current to the field windings on/off at the correct times. The difference is that in brushless DC motors the field current is pulsed in an all-or-nothing fashion, and in synchronous AC motors, the field current is sinusoidal. The result is that brushless DC motors inherently produce some vibration, whereas synchronous AC motors can theoretically be glassy smooth. In either one, there’s no sliding electrical contact (as opposed to the brushed commutator of a DC motor), so they can be very efficient.

brushless DC motors are usually used in small PC fans, with the controller being an integral thing; they’re pretty cheap. Synchronous AC motors are usually larger, and often have an external variable frequency drive to manage the field current. A good example? The Tesla Model S uses a synchronous AC motor to power the wheels. at its top speed, the motor is spinning at 16,000 RPM. The speed may be limited by its size and/or power (the car is doing 130+ MPH at that point). 16,000 RPM means only 266 Hz on the field coils, which is a pretty low number. It should be possible for a VFD to put out much higher frequencies than that, so as long as your motor can physically withstand it, 20,000 RPM should be achievable.

7

I don’t think that’s possible. I might have got the math wrong, but by my calculation, assuming a 50cm radius (to accommodate an 8 gallon container), your 8 gallon sample will ‘weigh’ about 7500 tons, when spun to 20,000 RPM.

8

Your design specifications are wrong. As in, they’re asking for the wrong thing. Your specifications should be specifying a particular acceleration, not a particular rotation rate.

9

If you’re going to participate in discussions here, please use correct technology.

That would be the “spinamajig.”

:smiley:

10

direct drive motors are used.

so much is an all depends on what you are doing. are you separating biological material, inorganic material?

the solution is a result of what you want.

11

I think he could be spinning minerals out of solid rock at that spec.

12

My rough calculations above came out at 223,600g - which seems a trifle excessive.

13

Glad I wasn’t the only one who thought that.

14

XKCD disagrees.

15

re: the gravity calculation

Dang,

so the only way they do that high-speed centrifuging is in small batches with small lab type equipment? You guys don’t think you could make a larger machine to do large (not huge) batches?

The material would be liquid/organic; water/liquid with suspended solids

16

Mangetout’s calculation posits a 50-cm (19.7-inch) spin radius. That’s very large compared to most tabletop centrifuges. They also don’t usually spin that fast. Most of the ones you see listed on Amazon spin at 3000-5000 RPM. Of the ones that get up into your territory (16,000+ RPM)), If you pick this 18,000 RPM model for example, and scroll down to the specs, you’ll see that its max g rating is 25,718. I won’t do the math, but suffice it to say that the spin radius is far less than 19.7 inches; based on the picture and the listed dimensions of the whole unit, I’d guess the spin radius is more like 3 inches.

As was suggested upthread, you’re specifying the wrong thing. You should specify the desired acceleration (this will be dictated by what you’re trying to separate out from your samples), and then choose a spin radius and RPM to achieve that (or simply shop for a centrifuge with that G-rating).

17

There are larger-capacity laboratory centrifuges and ultracentrifuges. Having shopped for centrifuges recently, I can tell you that a fairly ordinary bench-top centrifuge can spin 4 L at 4000 x g, or 600 mL at 24,000 x g (depending on the rotor). Moving up from there, I see you can get ultracentrifuges that can spin 800 mL at 180,000 x g, or high-capacity lower speed centrifuges that spin 12 L at 7000 x g.

Do you really need a very hard spin for the “suspended solids”? It doesn’t sound like you’re going to do anything with precise density gradients, so as long as the solids are denser than water a few thousand x g should be plenty.

ETA: My numbers come from flipping through a pile of Thermo centrifuge manuals, and the brochures I haven’t yet pitched out.

18

Thank you Machine Elf

I haven’t been specifying a g because I can tell you that the process is always done at at least 10,000 rpm, but usually at 20,000 rpm or faster because the separation happens faster. In industrial processes, time is usually the favored variable to shrink, so I’m thinking doing the process commercially you’d opt for the faster speed/time.

The question from there is, given the g forces you guys describe, what kinds of forces does that come out to mechanically? I’m thinking I’d turn it into a pressure calculation, among other ways of calculating it out

then from there, can a vessel be constructed to contain those forces? We already know that it isn’t some high science to create vessels that can handle a lot of pressure, it’s a regular thing. Don’t scuba tanks hold like 2,000 psi or so? Anyway, even just plate steel could maybe handle it

Some of the calculations may be hindered by the fact that the vessel may have an odd cylindrical/sloped shape, to hold the liquid the way centrifuges that hold vials do, where the vial is on a slant

WHen I’m less lazy and have more time, I’ll do some calculations

Here’s a better question. If all the liquid is in one vessel, which is probably ideal (though it may not end up being so, for the later extraction process), once you get up to speed, will there be differential currents due to the differential speeds (interior is slower than exterior) that would interfere with separation? So far this process is only done in small batches with lab centrifuges, where entire vessels are loaded, so in each vessel is separate and is it’s own little gravity chamber and the differential speed of spinning is of no consequence since there are walls

BTW points to the gentleman who can figure out what my “idea” is. It should be somewhat obvious, if you’re relatively culture savvy and follow trends and stuff

19

In the process another fraction has to come out that represents less than 1% of the liquid, and it is necessary to do it at at least 10,000 rpm. I think it takes two hours at just 20,000 rpm. Already, that’s a long time (for commercial production). Maybe you could do it for like 15 hours at 10,000 rpm and it would work,

20

A gallon of water weighs about 10 pounds at one g. 3 gallons at 25,000 g would exert a force on the bottom of the container of about 75,000 pounds.

Taking things to the ludicrous extreme, 8 gallons at Mangetout’s calculated