How to Measure Blender Speed: Methods

How can you measure the speed of a blade spinning at tens of thousands of RPM without fancy lab equipment? On this page I will describe how to measure blender blade speeds at home with two simple methods. The first method is a bit more involved, so if you get bogged down in it, you can skip ahead to the second method, which is easy if you have access to a laptop computer. You can read about the results in my companion Vitamix speed measurements post.

First Ideas (not tried)

My first thought was to use a high-speed camera, but unfortunately I couldn’t figure out how to affordably get access to a fast enough camera. (There are cameras that shoot at tens of thousands of frames per seconds, but they cost tens or hundreds of thousands of dollars.) My next idea was to shine a laser at the blades and detect the flashing reflection with a photodetector. That would be a lot cheaper than a high-speed camera, but it seemed like the setup could be tricky. Then I thought of something easier.

Method 1: Arduino-powered LED Strobe

I realized that if the blades are illuminated with a light flashing at the same frequency as the blade rotation, the blade will appear to “freeze” due to the stroboscopic effect. Serendipitously, I had recently received an Arduino kit as a gift, and the very first Arduino tutorial involves blinking an LED light. (For those not familiar with Arduino, it is an affordable and relatively easy to use electronic microcontroller1.)

The room has to be pretty dark for the LED to be the primary illumination, so I did it at night with the main room lights off. At first I hard-coded in a blinking frequency to see if I could get the blades to “freeze.” Sure enough, by adjusting the variable speed dial on the blender I was able to find a speed where the blades appeared static. After this proof of concept, I invested $4 in a rotary knob so that I could easily adjust the strobe frequency.

With a rapid prototyping breadboard, this was possible to put together quickly without any soldering:

Arduino strobe setupI did not have a display screen, so I used the serial-USB connection to a laptop to display the frequency.

There is one more wrinkle to this method. The blades will also appear static at multiples of the true frequency. That is, if the light flashes at half (or one third…) the frequency of the blades, the blades will appear in the same position. Additionally, from above, the blades have 4-fold rotational symmetry, so the blades actually appear frozen at multiples of quadruple the frequency. This may be starting to sound like a mess, but fortunately there is a way out. We can start to make sense of it by finding all of the frequencies where the blade appears static.

As the blade is spinning and the strobe is flashing, there are a variety of visual patterns. I found it useful to focus on two patterns as viewed from above: when the blade appears with four prongs as it does when it is not moving and the case where it appears with eight prongs. The eight prongs appear when the blade rotates one eighth of a revolution per flash. Then at half that frequency the blade rotates a quarter revolution per flash, and the four prongs appear. The four prongs appear at subsequent slower multiples as well.

To get around the problem of rotational symmetry, I used the fact that the blades are not rotationally symmetric when viewed from the side. (I could have instead just marked one of the blades, but the side view worked well enough.) On first glance the Vitamix blade has two-fold rotational symmetry when viewed from the side, with two of the prongs pointing straight out and two angled up. However, of the prongs pointing straight out, one points down at a slight angle:

Vitamix Blade Sideview(The first time I noticed this angle I thought that the blade had been bent with heavy use, but once I saw the same angle on other blades I realized that it must be at that angle by design.)

This asymmetry means that when the light is flashing at quadruple and double the blade frequency, the lower part of the blade appears as a doublet from the side view (due to the superposition of the differently-angled prongs). At the true rotation rate there is only superposition of the same prong, so the doublet disappears. This allowed me to confirm that in fact the four-pronged pattern that occurred at one eighth the frequency of the eight-pronged pattern was the true rotation rate. (The doublet is also absent at slower multiples of the true rate, so it is necessary to look for the fastest speed that gives a singlet side-view.)

With this method I found the maximum speed of the Vitamix Pro 750 to be 380 Hz, or 23,000 RPM. Vitamix claims that the Pro 750 blades move at 270 MPH, and if you take the distance that the tips of the 4″ blade travel per revolution, 380 Hz comes to 271 MPH. Not bad!

The disadvantages of this method are: it requires an Arduino, you have to carefully sweep out all the frequency multiples to figure out the true rotation rate, and it is impossible to do if there is anything in the blender because you can’t see the blades. Even blending something clear like water does not work because air bubbles get whipped in and make it impossible to see. My next method avoids these disadvantages.

Method 2: Recording and Analyzing Sound

I wasn’t sure how well this would work, but it turned out surprisingly well. We all know that blenders make a lot of noise. What if some of the sound directly corresponds to the frequency of the rotation of the motor and blades? As a first hint that this might work: if you pay attention to the sound as you change the speed of a blender you might notice that the pitch changes. Blender speeds of tens of thousands of RPM correspond to hundreds of Hz, solidly in the audible frequency range (middle C on a piano is ~260 Hz). In fact, Itzhak Perlman has an entertaining rendition of Twinkle Twinkle Little Star played on his Vitamix.

To quantify pitch, we can record it onto a computer and then analyze the frequency spectrum. This might sound daunting, but it’s quite straightforward. If you have a laptop, it probably has a built-in microphone that should work. Frequency analysis can be done with a few clicks if you download the right software. We want to calculate and plot the frequency spectrum to look for the dominant frequency.

I used the free, multi-platform, open-source software Audacity to record the blender sound and then analyze it. You can select a portion of the sound recording and go to Analyze→Plot Spectrum to view the frequency components. Audacity will give you a plot that looks like this: 750 empty Max Speed spectrumIf you mouse-over the peaks, it displays the frequency. Note that dB on the vertical axis is a log scale, so that peak is even more significant than it appears. (In fact, because of the way Audacity displays the plot, the narrow peak is actually even higher than shown—it’s just that it is so narrow that the top didn’t get plotted.) This is a spectrum of the maximum speed of a Pro 750; it matches the strobe measurements!

Audacity will also display a spectrogram of a sound recording. A spectrogram is essentially the frequency spectrum plotted versus time, so it allows us to view how the frequency changes over a full sound recording. I made these recordings by setting the blender to the lowest speed, waiting ~5 seconds, then hitting two wooden spoons together to indicate the speed was about to be changed, then adjusting the speed to the next highest setting, and so on for all speed settings. The whack of the spoons shows up as a vertical bar in the spectrogram because it doesn’t have a strong characteristic frequency. Here is a spectrogram of an empty Vitamix Pro 750 stepping through all of the speeds: Vitamix-Pro-750-empty-all-speeds-spectogramAt most speed settings the darkest frequency line is the true frequency. There are also harmonics, with the most significant one at double the true frequency. You can also see the speed-independent electric hum at 120 Hz.

One strange thing is that the Pro 750 and the 6300 that I tested both had hardly any change between speeds 1 and 2. All other speed increments increased the velocity the same amount on those models. The 7500 speed increments were the essentially the same for all speeds (including 1–>2), except for 9–>10, which was a bit less.

These sound measurements also allow us to make measurements while the blender is actually blending something. To keep everything simple and standardized I just used water. With 8 cups of water the spectrogram of the increasing speeds of the Pro 750 looks like this: Vitamix-Pro-750-8-c-H2O-all-speeds-spectogramYou can see that the spectrum is noisier (particularly starting at speed 3—I believe this is the onset of turbulence), but the blade rotation rate is still apparent. You can see the major characteristic of the Next Generation Models: at increased loads the low speeds stay the same, until they reach the maximum speed for that given load and the speed levels off.

With a blending load, the Standard models have the same maximum speeds as the Next Generation models. The difference is that on Standard machines all of the speeds decrease when the load is increased. For more details and discussion of these differences, see my Vitamix speed results post (which includes plots of the speed measurements).

Summary

I made measurements of blender speed with two simple setups. One using an Arduino-powered strobe, and the other with a laptop to record sound. If you would like more detail, or if any of the writeup is unclear, please do not hesitate to ask questions!


1. A name-brand Arduino board goes for ~$25 on Amazon.com and serviceable off-brand ones can be cheaper. For starting out, a kit with a bunch of different components is nice, so you can just start tinkering.


Did you enjoy this write-up?

Enter your email below to receive updates when new posts are published so you never miss new content. Emails are usually once per week or less.


Comments

How to Measure Blender Speed: Methods — 21 Comments

  1. What a tremendously nerd-stupid article. Today you can walk into your nearest RadioShack (or similar) and pick up a $50 RPM meter advanced enough to be able to measure pretty much any cycling mechnical feature.

    • When I started these measurements I did not know such meters exist. A few people who I showed this writeup to have suggested I could have used a cheap tachometer, but I still favor my methods for a few reasons:
      -I used materials I already had and didn’t have to buy a tachometer that I wouldn’t have much further use for.
      -My DIY methods were more interesting and fun than just reading a measurement off of a meter.
      -As far as I’ve seen, the cheap tachometers take an optical reading, and that would not be able to measure the motor under load as I did with my sound measurements.

      • I agree. This article is a good way to show how to think about problem solving. Picking a meter is so easy, it does not give you any added value. I would say the point of this article is not the goal (finding the speed) but the path (finding a way – actually two – to do it on your own).

    • How needlessly mean-spirited. :/ John, you might be kind of a jerk.

      Next time I want to measure the speed of something, I’m going to joyfully use the equipment on hand (a computer with a microphone and free software) rather than spending $50 on a single-purpose device.

  2. Your solutions are so hi-tech. Just use piece of twine and let it twine on the shaft of rotor for 1-2 seconds and then mesure it 🙂

    • Good luck with that approach at 23,000 RPM.
      The thinking and evolution of the technique here was excellent. A very good teaching exercise. Good job, Adam.

  3. Pingback: Do You Have Any Idea How Fast Your Blender Was Going?

  4. Excellent. I came to read the Arduino stuff because I need measurement AND logging for my bicycle. A bicycle computer is not enough for the various measurements I have been thinking about.
    Audio recording is my favourite subject. There are many things that would be really easy to verify if just the multimeters and logic analysers had speaker / headphone output. With fast logic signals you just need a frequency divider to bring the data rate down to kHz range. A long time ago I saw a crystal tester that outputs audio through a controllable divider. Anyway, interference based (IMO stroboscopy falls into that category) measurements are sometimes really useful as they can yield to higher resolution than direct detection. In radio astronomy a set of antennas form a large virtual antenna.
    All building and testing deserves lifting thumbs up! 🙂

  5. What a great article! While I had only a small interest in blending, I really wanted to understand the scientific method used. I found the author to be extremely clever at exploiting first principles. Kudos for writing such a great report. Keep it up!

  6. The spectrum in Audacity is found pretty easily, but I’m having trouble finding the spectrogram utility – any hints?

    BTW acoustic measurement may or may not be the simplest/cheapest way to measure things, but it will give you information that you can’t get anywhere else, and it can be done without ever seeing the blade, so ignore the trolls – great article.

    • Yeah, the spectrogram is not immediately obvious; I used Audacity for quite a while before I found it. Once you have a recording, in the upper left of the track there is a drop down menu titled “Audio Track;” click it and select “Spectrogram.” (The default selection is “Waveform.”) You will also likely want to adjust the spectrogram preferences, found under Preferences–>Spectrograms. Also note that you can drag the bottom (yellow) border of the audio track to enlarge the spectrogram. And there’s of course the zoom in, out, and fit buttons in the upper right of the main control window to adjust the time axis.

  7. Thanks for this post. Very interesting ideas.

    I used the Audacity method with a ZOOM recorder to measure speed on a late-2016 750 Pro Heritage model. Results are practically identical to your measurements.

    The behavior of the analog speed dial is peculiar indeed. There is no measurable difference from speed 1 all through speed 2. The motor starts picking up speed somewhere between 2 and 3. Wonder why that is.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.