I think I heard about people doing this a long time ago when more people used basic cyclocomputers. My question is why does it matter? I don't know about anyone else, but I'm not looking down at my computer to check my speed WHILE I'M RIDING.

 

I check my speed while I'm driving my car too. You should try it. It's good to know speed .

 

That wouldn't necessarily make it more accurate.

If the computer knew there were two sensors and took the average of the readings to try to get a more accurate number, it might work. But the computer isn't doing that.

Without that, I think the uncertainty on the two readings would be just be additive.

 

That wouldn't necessarily make it more accurate.

If the computer knew there were two sensors and took the average of the readings to try to get a more accurate number, it might work. But the computer isn't doing that.

Without that, I think the uncertainty on the two readings would be just be additive.

It would increase the resolution because the computer is taking a reading every 1/2 turn instead of every turn. It definitely would do that. Since climbing is very slow, like 3-6mph, it would register a truer reading if you accelerate or decelerate while climbing.

 

I can kinda see it but let's think this through.

There are two ways the measurement could be calculated:

A) Revolutions per unit of time * circumference = distance per unit time, which can then be converted to MPH w/ simple units conversions.

In this case, yes, 'resolution' is important, for example if your time window is 3s, and you got 2.9 sensor hits in 3s, then the computer would assume 2 revolutions and the speed would be inaccurate (to the low side).

B) Circumference / (Time between current sensor hit and the prior sensor hit), then unit conversion to MPH

This seems like a better way to calculate it, because it avoids the '2.9' problem in method A.

In this case, extra sensors would not make it more accurate, in fact, they might make it less accurate because there are uncertainties in each measurement of time (the magnet might not trigger the sensor with the exact same precision each time, there might be a +/-1% or something). One longer revolution with a single time interval measured between the two sensor hits, might be a better sample than a shorter sample every 1/2 revolution, because time uncertainty is a larger in proportion to the unit time.

For example, let's say your wheel is rotating 1x per second and the uncertainty in the time measurement is +/- .05 sec.

With a single revolution that's MPH = Circumference / 1s +/- 5% (then convert), with two sensor measurements per revolution, your calculation is MPH = Circumference (halved this time) / .5s +/- 10%.

See what I mean? The error is a larger percentage of the time measurement here.

Either way, time uncertainty is a problem with both single and double sensors, so my guess is that the system would average the reading over the past 3 or 4 revolutions at any given time, which, would just wash out the effect of your double sensors really.

If I'm wrong, I think any accuracy gain you might get (perhaps +/- 1-2% max) is nullified by the uncertainty in the circumference of the tire anyways. Despite measuring the circumference on paper or with the tape-on-the-ground method, there will be real-time uncertainties in the actual circumference, like how far the tire deforms under your weight, your PSI on that day, how high or low the knobs ride due to the hardness of the surface, etc.

Those may seem small but when you're dealing with, say, a 26" wheel, 2.7" - 3.2" of tire height (varying) = ~ 90.5 - 92 inches of circumference, that's +/- 1-2% error in your MPH anyways.

 

Okay, so, that's blown up.

The way my rims are laced, I can't get two spokes on the same side of the rim exactly 180* apart, so, it won't work on my rim. Oh well.

 

Not surprised.

Sent from my SM-G900V using Tapatalk

 

why stop at 2? One magnet per spoke, divide wheel circumference by the # of spokes and get supreme accuracy. Or just accept the fact that there is a normal level of error in all systems and roll with it.

 

Because the lowest entry that can be made for the circumference is 1000mm.

 

Right, it's trying to extrapolate a measurement on the order of ~100 inches / sec into one of 64k inches / 3600 seconds. That's just inherently noisy.

 

@ride, assuming the measurement mode is B above (circumference / time between this sensor hit and the last one), here's an analogy:

Go out to your garage, and measure the width of your two car garage door, with a 1 foot ruler. You'll notice, each time you place the ruler, there's some uncertainty in where you place it relative to your last placement. Measure all the way across, 3 times, and take the average of the readings.

Now, do it again, with a single tape measure. You'll notice you measurements with the ruler were not consistent with either themselves or the tape measure method.

It is much the same with two time measurements per revolution, each starts from the last, so the total uncertainty is the sum of the uncertainty of each measurement.

Make sense?

If the used measurement method A above, then more resolution is better. But we just don't know, we'd have to look at the programmers algorithm.

 

From a prior thread:

Note that with a wheel magnet (or cadence magnet) sensor, the magnet has to pass the sensor twice to give data.

So if you take off like a batoutofhell, the lag should be less because the head unit is waiting less time for the second ping compared to a gentle start.

I can confirm that you can put two magnets on a wheel (in line with the wheel centre, find the right spokes) and set the wheel diameter size to 50%, you get much more responsive data on Garmin units when starting and stopping. Especially on slow-moving bikes with big wheels, like fatbikes. The reason I tried this (a few years back) was when overlaying GPS data (i.e. speedometer) on my Gopro video, the lag was way more apparent then it would ever be riding, and the two-magnet setup cut this lag in half.

I'm not sure if the new Garmin magnet-less sensors transmit one ping per rotation or a bunch. If it's the latter I might buy one, since two magnets are twice as likely to get knocked out of alignment as one...

You are right that it will not increase distance measurement accuracy in a meaningful way, and it will not have any discernible effect on the speedometer display once the bike is up to speed.

 

From a prior thread:

You are right that it will not increase distance measurement accuracy in a meaningful way, and it will not have any discernible effect on the speedometer display once the bike is up to speed.

^interesting post.

 

That's great, it's great to see the experimentation here. But given there are two possible algorithms for determining realtime speed, we cannot know which GPS has which algorithm, even within a brand. Findings on a particular computer or GPS don't necessarily translate to others. Only the programmer knows.

A) Revolutions per unit of time * circumference = distance per unit time, which can then be converted to MPH w/ simple units conversions.
B) Circumference / (Time between current sensor hit and the prior sensor hit), then unit conversion to MPH

 

I find speed very useful on the single speeds.
After going down anything, I want to know when my speed is approaching 22kph so I don't resume pedaling too soon.

On a geared bike, I rate cadence higher, but on the single speed, speed over cadence.

 

I'm a pretty data driven person. Speed is the last thing I am interested in when riding the MTB.

On my GPS I have feet gained (so I'll know what I have left to climb for pacing purposes), elapsed time, mileage (for pacing), HR, HR average (for pacing). The only time I care about speed is on my roadie, when I am on a bike path. Speed and cadence aren't useful at all (no cadence sensor). If I had power, I would only use normalized power.

 

Right there with you on HR. After my back surgery I used one for every ride and every run for the next three years. I've sort of learned to gauge my exertion level based on perceived effort, but I don't really track that data anymore. If I want to know exactly how fit I am I ride with other people, when they leave me behind I know I'm not fit enough yet.