# Tech In Plain Sight: Speedometers

In a modern car, your speedometer might look analog, but it is almost certainly digital and driven by the computer that has to monitor all sorts of things anyway. But how did they work before your car was a rolling computer complex? The electronic speedometer has been around for well over a century and, when you think about it, qualifies as a technlogical marvel.

If you already know how they work, this isn’t a fair question. But if you don’t, think about this. Your dashboard has a cable running into it. The inner part of the cable spins at some rate, which is related to either the car’s transmission or a wheel sensor. How do you make a needle deflect based on the speed?

## Mechanical Solutions

Early versions of the speedometer used a governor pulling against a spring. The faster it rotates, the more the two weights of the governor pull out against the spring, and the needle moves with the weights.

As an aside, this sort of centrifugal governor is also known as a fly-ball governor, and similar devices were commonly used to regulate the maximum throttle on steam engines. The arms of the governor would be fully extended once the engine reached its top speed, which lead to the term “balls-out” becoming used to describe a machine operating at its upper limits.

Another type of mechanical speedometer had an escapement like a watch. The time mechanism would move the needle back, and the rotation of the wheels would move it forward. The net result was a needle position that would increase with speed.

## The Magnetic Approach

However, most cars use a magnetic type speedometer — although it doesn’t work in the way you might imagine. There’s no reed relay or Hall effect sensing the magnetic field. Instead, there is an aluminum cup attached to the speedometer needle and, nearby, a magnet that spins on a shaft moving at some ratio of the car’s speed. There’s no direct connection between the two.

Being a non-ferrous metal, aluminum is not generally something we think of being affected by magnets. Under normal circumstances that might be true, but a moving magnetic field will induce eddy currents in aluminum. This forms a field in the aluminum, too, and the spinning magnet tends to drag the cup, thereby deflecting the pointer.

A spring similar to one you might find in a mechanical clock or watch pulls back the pointer so the needle hovers at the point where the force of the magnet pulls against the spring. The pull on the spring has to account for the gear ratios and the size of the tires to accurately reflect the vehicle’s speed.

If you want to see an entertaining teardown of an old speedometer, [Tubalcain/Mr Pete] has you covered in the video below. He also shows how the odometer part worked, too.

## Modern Times

Of course, these days you are more likely to pick up a pulse using a Hall effect or some other part of the vehicle and just count the pulses in the car’s computer. In fact, the pulses might be encoded at the source and travel over something like a CAN bus to get to the computer.

It is also possible to pick up speed from other tracking information like GPS, although that might not be as accurate. But if you have, for example, a mobile phone app that shows your speed, that’s probably what it is doing. The obvious way to do that is to take position measurements periodically and then do the math. However, more sophisticated systems can actually measure Doppler shift to get a more accurate reading.

We see a lot of bicycle speedometers for some reason. Eddy currents make induction cooktops work, too. Even tiny ones.

## 29 thoughts on “Tech In Plain Sight: Speedometers”

1. Canuckfire says:

I have also seen fly-ball governors on top of elevator cars to drive the braking system.
I am not sure how common they might be, but I have seen quite a few.

It makes perfect sense for that application too:
They can operate completely independent of any other system, are mechanically simple, and with gearing would reliably activate the braking system with mechanical movement of the car and intertia alone

2. macsimski says:

on a side note: the linear motors used on conveyor belts at the airport work the same way: a aluminium (aluminum) plate is attached to the bottom of a cart perpendicular to it and passes though the poles of an electro magnet, pulling on it for a short time until its half way in.

and again for lineair motion control on a roller coaster to break the train on the end of the track. no wear on the mechanism as nothing touches

3. Paul says:

Also the operating principles of the eddy brake in brake dynos, and the Klam Retarder in trucks.

4. PPJ says:

” But if you have, for example, a mobile phone app that shows your speed, that’s probably what it is doing. The obvious way to do that is to take position measurements periodically and then do the math.”
Is there an app that calculates speed by integrating accelerometer reading?

1. iliis says:

I’m sure there is something, as it’s not that hard to implement. But it would be of little practical use, as these sensors are pretty noisy and your integral will drift away quickly. If you combine it with the camera though and do visual-inertial odometry then you would get very good results, but that is much harder to implement well. Luckily there are AR frameworks that do the heavy lifting.

2. CityZen says:

They can use a “multiple sensor fusion” approach, allowing them to work even when not all of the sensors work (such as when you enter a tunnel).

5. craig says:

There are aftermarket GPS speedometers in the market now that claim to be very accurate. And they don’t change as tires wear etc.

1. TG says:

Interesting. If they’re cheap and reliable it might be nice to have on vehicles with larger than stock tires, I should look into that. It’s tiresome doing the math whenever I pass a speed trap

1. Drone says:

@TG said: “Interesting. If they’re cheap and reliable it might be nice to have on vehicles with larger than stock tires, I should look into that. It’s tiresome doing the math whenever I pass a speed trap.”

Working from (sometimes unreliable) memory: If your speedometer/odometer sensor takeoff is from a point not common to a particular wheel or axel (i.e. the transmission gearbox, which is often the case), then the accuracy of the speedometer/odometer will be dependent on the diameter of the driven tire or tires. Larger diameter tires result in lower than normal speed readings and total mileage, and vice-versa.

2. The Commenter Formerly Known As Ren says:

I have three GPS devices in my car. I often turn all them on, except for short trips.
Why? They were cheap, and I’m a nerd. But I do like to check their speed readouts with my speedometer.

6. SeattleSipper says:

I bought a replacement speedometer for my boat. The original unit (c. 2000) had a paddlewheel at the stern of the boat connected to a wire that ran up the length of the boat to the speedometer in the dash. The replacement just takes 12Vdc (no velocity sensor data) because it has an internal GPS unit. The sole purpose of the GPS is to determine location so that the velocity can be calculated and displayed on the speedometer; location data is not displayed. The size and appearance of the two speedometers are identical, and the new GPS unit is cheaper than a replacement paddlewheel unit. The display on the new unit is mechanical (electro-), but I suspect a future generation will use a digital display.

1. Paul says:

GPS receivers can provide velocity natively — there is no need to use successive position measurements to calculate it.

1. CJay says:

The velocity is in Knots too, the name of the protocol, NMEA, gives a bit of a clue too I think.

2. The Commenter Formerly Known As Ren says:

Interesting, yesterday I saw parts of a pitot tube speedometer for boats.

3. GEO says:

Most sail boats will have the paddle wheel and a gps speed. The paddle wheel gives you the distance log for the boat and the speed through water, where the gps gives you speed over ground and distance made toward the waypoint. The difference between the two will tell you if you are going with currents or fighting them.

The GPS will also give you “velocity made good” which is the amount of speed actualized toward the waypoint. In many cases it is worth it to point 30 or 40 degrees off the waypoint to sail at a better angle and higher speed, than it is to struggle along pointing at a waypoint on a bad point of sail.

7. zog says:

Typically the spring in the speedometer doesn’t tie to the gear ratio, size of the tires, or other such aspects. Some cars drove it directy from a wheel like the original VW Type 1 (aka Beetle) which went through a hollow front axle stub and through a square hole in the bearing dust cap but more typical was driving it from the transmission output shaft – different gear ratios were used there to calibrate the speedometer by standardizing the rate of rotation it saw. You could never get it exact but close.

In the first case, the speedometer would need to be calibrated to the wheel size. I know of “speedometer shops” that could tune and even certify speedometers (police cars typically used a “certified” speedometer – even my 2001 ex-cop Crown Victoria stated that on the speedometer) but I’m not sure how they actually did it. Replacing that spring seems like a nightmare.

1. CityZen says:

I think you are agreeing with the actual point. The spring is calibrated such that a certain rotation rate corresponds to a certain speed displayed. The input to the speedo must match that expectation as well. So if you change tire diameter, the reading will be incorrect. If the input comes from the transmission prior to the final drive, and you change the final drive ratio, the reading will be incorrect.

1. Amenjet says:

I built speedometer calibration machines for use in automotive factories and the calibration was done by altering the strength of the magnet in the unit. A stronger magnet led to more turn of the needle. Every speedometer was put in a rig that had a vision system to record the angle of the needle at a set speed, and the calibration machine would apply demagnetising and magnetising pulses to the magnet and get the needle into the required position.

1. Amenjet says:

Actually, thinking about it, the pulse were demagnetising only, the magnet was over-magnetised and brought down to the required strength.

1. Just had to comment to say “thank you” for that info… I love to discover that random tidbit stuff.

2. spaceminions says:

You would replace the speedo gear if you replaced the axle gear(s) with aftermarket of a different ratio, or tires likewise if you cared about the difference.

1. The Commenter Formerly Known As Ren says:

I read your comment before I watched the video.
I’ve made a number of trips to La Crosse since 1998, but never saw/heard of an Autolite factory.
Now I know why.

8. OldTechGuy says:

I worked with a few types of speedometer configurations while working at General Motors in the US.

The earliest ones used a pair of hyperboloid gears; one on the transmission output shaft and one on the end of the speedometer cable that threaded onto the housing. There were several available, and one pick the combination the right number (or as close as possible) of revolutions per mile (1000/mile) to yield correct speed and odometer readings. The cable-end ones were easy to change; the other required dropping the driveshaft, unbolting the tail housing on the transmission, and depressing a small retaining clip. It was a pain to do.

Later, a reluctor was fitted to the transmission output shaft, with a potted magnet, coil, and core assembly that was inserted into the tail housing. The signal from this was fed to a small box called the DRAC (‘digital ratio adapter controller’) that could be configured to divide down the pulse rate from the pickup. This was used in vehicles that had a digital (numeric) speedometer display. I don’t remember if the engine controllers (at the time, single-cpu units) received a signal from the DRAC as well, and if they did, it might have been a nicely conditioned version of the high frequency pulses from the pickup.

In the last configuration I worked with, the signal from the reluctor pickup was used directly by the powertrain (both engine and transmission) controller, and the ratio could be adjusted in its firmware.

I last worked with this stuff around 1992, and I’ve been doing my ‘Swedish Death Cleaning’, so I no longer have a lot of my reference materials. As such, the above is solely from memory and may be fraught with some errors.

9. Any perfectly ordinary GPS receiver can get the time from the satellite network the same way it gets position. The only limit is position requires a minimum of three satellites, velocity requires a fourth. In practice you usually have at least six or seven in view if not more. It’s far more accurate than any system that works on the rotation of pneumatic wheels, because tire pressure will affect those readings. The main reason you don’t see cars using GPS for this is that they have to work in radio blackout conditions, such as in tunnels and mountain passes and cities where the GPS signal can get garbled by reflections.

1. Hugo Oran says:

But in tunnels the GPS don’t work. That’s why you need independent local speed measurement unit in the car.

10. zit says:

I don’t have a driver license, riding a bicycle for everything, so I’m not familiar with cars, but not long ago, I’ve been with a friend in his new Mercedes, and I was amazed by the speedometer giving also the fuel consomption : as a truck was behaving weird, my friend had to accelerate from 120 km/h, the consomption was something between 6 or 7 litres for a 100 km, to maybe no more than 150 km/h, and the fuel consomption raised to 25 litres for 100 km 8-0.
It’s a good thing to know when you pay for the fuel ;0).

1. That usually comes from injector pulse width… you know roughly how much gas is going into the cylinders per crankshaft revolution. The car already knows what gear is selected (and therefore how far it has moved during that revolution), so the math is straightforward to determine how far you’ll get for the amount of gas being pumped in.

11. Stephen says:

Here in Britain a man using the stage name Johnny Ball:
https://en.wikipedia.org/wiki/Johnny_Ball
presented a series of shows in the late 1970s and early 1980s presenting mathematical and scientific ideas accessibly for children. One of his demonstrations was how a magnetic speedometer worked. When I began reading this article I immediately remembered the demonstration on the show, and was delighted to discover that it was exactly right. He’s still going: he was on Numberphile recently.

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

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