zacster

I like my Wahoo system that will use the GPS data it is already collecting to measure the wheel circumference so you get accurate speed and distance from the sensor. Using GPS alone does not give immediate speed, using a sensor alone with fixed circumference will not take tire pressure/wear/variability into account. Combining the two is pretty accurate.

dsbrantjr

It is not difficult to do. Set your computer to an approximate rollout distance. Ride a known distance. See what the computer calculates vs the known distance. Do the proportion calculation (simple math) and adjust the rollout distance. The calibration will be near-perfect.

gregf83 

Until your tire pressure changes
The point was the OP’s solution doesn’t address this issue.

wphamilton

There is another issue, in that those magnetic reed switches - the cheap ones used in bike speedometers - are imprecise compared to optical switches and also due to debouncing. Putting one on each spoke, or every other spoke, could have one still noisy from the bounce at an appreciable portion of the time segment where the next is due to activate. You won't achieve the precise instantaneous measure that you're going for with that setup. It doesn't matter in normal operation because you're averaging several wheel revolutions in a second, and it averages out to give a pretty good speed.

Granted that reed switches are more accurate than many Hall Effect alternatives, and have other advantages such as low power operation, pure simplicity, reliability and so on which makes them ideal for a normal bike speedometer. If for some reason you want instantaneous speed with better precision you need to look at other sensors. OP's idea isn't really unusual for that purpose.

SBinNYC

You could achieve the result of a quicker low speed response by adding magnet(s) to additional spoke(s).

SBinNYC

There's a fundamental problem with using accelerometers. No instrument can distinguish between inertial (due to changes in motion) and gravitational acceleration. That's part of Einstein's General Theory of Relativity.

Inertial navigation systems work by having an embedded gravitational model (based on current position) and subtracting gravity from every reading. The problem is that the equations are unstable.

That's not important for short trips like ballistic missiles. However, for longer trips the system must be recalibrated. That's the reason that the astronauts needed to take star measurements periodically during the moon trips. Ditto for submarines.

SBinNYC

The reason for the delay in registering speed changes isn't due to the system's slow response to sensor input. Measuring a rate of change by differentiation means that high speed changes due to noise are magnified more than low speed changes due to a trend. The cure is to include only long term changes by filtering out high speed changes. Any low pass filter introduces a delay in recording long term changes.

wphamilton

The equations are nasty, but there is an easy way to bypass them using the constraints of a bicycle's motion. We're far adrift from the bike mechanics bike speedometer application though. The main point, I think that you'll agree, is that unless you're looking for an accurate instantaneous speed, during a fraction of a wheel's revolution, a reed switch is about as good as it gets without heroic efforts.

If you do want that instant speed for some reason, looking at the raw signal from a bike magnetic reed switch I don't see any way you're going to feasibly achieve it by adding extra magnets. You'd need to source a specialized high-precision reed switch, take some care with either the electronic filtering or digital debouncing, and by the time you've done that maybe an optical switch is easier and cheaper after all. I like his idea, and even the interrupter scheme has been utilized when people wanted that kind of data, but as he speculated environmental factors (and complexity) are problematic for everyday use. I think that the pinhole method, where a disk blocks the laser until a small hole passes by, has been found in practice to be superior though more elaborate.

alias5000

Go to the engineering side of position sensors for synchronous motors and you'll find position sensors with an absolute overkill resolution and accuracy (the gate ones go to 24bit on 260degress AFAIR). Using a well-designed and well-placed magnetic field, plus 2-axis hall-effect sensor can give you the position information with a resolution of better than 1deg resolution. Differentiate and low-pass filter that and you'll be able to observe the impact of your pedal strokes on your bike speed. Call it 'the masher meter'.

SBinNYC

Wholeheartedly.


I merely wanted to point out the perils of using accelerometers for estimating velocity and position.

wphamilton

Which you did and for the record correctly as far as I know.

fietsbob

a range finder that targets a fixed object, getting nearer, from a moving one,

much like the radar speed gun the police use to get citation data on speeders..

can be the same devise, as the Constable in their fixed location

pointing the thing at the approaching vehicle and measuring their closing speed distance change rate.





.....

FBinNY 

Too much work for no benefit. At 15mph the wheel,turns at 3 revolutions per second. How much more closely defined do you want it? Even at 5mph, it's still one reading per second.

There's also a technical problem. Bicycle speed peed varies slightly as you pedal, especially at low speed and when climbing. Since a constantly changing display would be impossible to read, you'd need a buffer to take a rolling average and send it to the display at intervals. Just for the sake of readability, you probably choose to update only every few seconds, so whatever instantaneous accuracy you hoped to gain would be lost in the buffer anyway.

flanso 

How about a helmet-mounted Pitot tube?

alias5000

But then you would have to differentiate between ground speed and air speed
Thinking about it, maybe that would even be a pretty valid information!

fietsbob

+1) add air speed data to ground speed?

in a boat fighting an incoming tidal flow in a narrow passage ,you can be going nowhere for hours until the tide turns.





...

alias5000

The closest analogy to biking would be on indoor trainers. Air speed vs ground speed would mostly be dominated by fan speed. We could upload that to Zwift to supplement the wattage readings!

KD5NRH

Find an old ball mouse and pull it apart; most of them that I've dealt with used a pair of these sensors. Some with better accuracy than others, but all had issues with dust or lint getting into the guts and blocking one or more slots in the discs.

70sSanO

The biggest problem with designing a device that focuses on speed, and therefore mileage, accuracy is competing with the modern computers that provide a mountain of data including route, altitude, heart rate, etc., and a multitude of various splits, intervals etc.

I use primitive magnet computers, I really don't care if my average speed over a certain number of miles is off by some fraction. If it were that important to me I'd just get a Garmin and all the other goodies.

John

_ForceD_

"Different" doesn't mean better.

Dan

Kimmo

I like the old Avocet idea, just because it's so much neater than an ugly magnet; higher resolution is just a bonus. Pity it didn't survive into the wireless era.

As for the accuracy, it doesn't depend on the resolution, it depends on how accurately you've entered the circumference. Some speedos only take cm rather than mm, which is crap, but otherwise, it's up to you.

oldschoolbike

ABS works pretty much exactly as the OP describes, except that a little strobe wheel (looks like a gear with flat-topped teeth) has to be provided. The light is bounced of the tooth tops. Unlike the saw-stop, which is more like an airbag with it's explosively single-purpose and sacrificial function, ABS systems provide a multi-purpose data stream. Even my 2003 car uses the data for ABS, speedometer, anti-spin traction control, and cruise control. Newer cars probably do even more with the data. My motorcycle's speedo sensor is in fact a re-purposed car ABS sensor, looking at the final drive crown gear.

One lesson from car and motorcycle systems is that the dirty optics and background light problems can be solved. There are very effective electronic and/or software solutions for identifying even very weak periodic signals against a strong and/or noisy background. Once you find the signal you can "lock on" to it and react if its period changes. An example of finding a weak periodic signal would be the smart phone apps that can detect and display your pulse rate by looking at the light scattered from your fingertip when you put it over the illuminated video light LED. Maybe if you duct tape your iPhone to your forks.......

All that said, a bicycle speedometer does not need the quick response of an ABS system, so a single magnet is good enough. It's still a great thought experiment to ask "what if" or "why not" use optical signals.

Hoopdriver

I'm still using my old Avocet 20. With 4 magnets, the response is great

harshbarj

Thoughts?

Sure, move on. The system we have now works good enough and is relatively simply and free from interference for the most part. Even at a slow speed (about 8 km/h) your wheel is likely rotating at least once per second. Who needs to know their speed at a rate more than that? We also have GPS based systems. I myself use my phone and find over my 8.1km commute it is virtually always 100% accurate and updated faster than my old bike computer did.

Remember you did ask for our opinion.

+1

SethAZ 

On an old bike that had a magnet-based speedometer with a Cateye bike computer I simply divided what my Garmin 305 told me was the distance on a very long ride (like 40 or 50 miles) where the gps error would be very small, by the distance reported by the Cateye computer to arrive at a correction factor. Multiple the "distance per revolution" factor set up in the Cateye computer. Et voila! The Cateye suddenly agreed to a remarkable degree with the gps, and was probably about as accurate from then on as it was possible to be.

If someone really wants to fix what ain't broke and come up with a device for counting spokes so they get a higher-resolution instantaneous speed measuring ability, they could rig up a small magnet wrapped with some fine wire in the manner of an electric guitar pickup. If you mounted that close enough to the spokes you could count the pulses in the induced voltage through the coil caused by the steel spokes passing through the magnet's field. You wouldn't have to put anything on the spokes at all.

Of course doing any of this is stupid, because those tiny magnets on the spoke passing by that little wire coil zip-tied to the side of the fork do just fine, and are as accurate and instantaneous as anyone willing to settle for such low-tech solutions really needs it to be.