rm -rf

Actual wheel size measurement:
When I used wheel magnet speed sensors, I did a rollout to get a pretty accurate wheel size:
Start with the valve stem at the bottom center. Put a tape marker on the floor -- sight over the handbar down to the valve stem for consistency. I liked to roll two full revolutions if I had room. Mark the ending point of the valve. Now measure the distance with a tape measure. Divide by two to get one revolution. google "inches to mm" and do the conversion to mm. That circumference of the wheel in mm is the usual number to put in.

You are supposed to press down with your weight on the bike, but it didn't seem to make a lot of difference.

~~~
GPS accuracy:
From a recent ride using my Garmin 1030. This map is from the useful "My Tourbook" PC software, that holds my rides history.
The dots are one-per-second recorded locations from the Garmin. They are color coded by speed. This was about 17 mph, on flat roads in farmland.

With heavy tree cover, roads with steep hillsides that reflect GPS transmissions, or downtown buildings, the path drifts more, and accuracy is less.

The map: Outbound from the east, with a left turn heading south. returning from the west to east. You can see which side of this two lane country road we were on! The scale on the map, 0.05 miles, is 264 feet.
(I was quite surprised the first time I saw these maps. It's amazing that the GPS can do that complicated math once a second and be this accurate.)




Wheel sensor with GPS
I have the Garmin sensor that straps around the hub and uses accelerometers to count wheel revolutions. It knows instantly when I'm stopped, and it's more accurate on low speed, steep climbs too.
The Garmin 1030 is set to "automatic wheel size", it recalcs wheel size from the GPS data as I ride. I think it's accurate, and it's simple too.

steelbikeguy

There are a number of those sorts of "jogs" in the roads around here too. The roads are all laid out in a 1 mile grid, except for where rivers and such interfere. I have a lot of respect for the folks out in the field with transits and the measuring chains, taking careful measurements and notes. Nothing is perfect, and the errors add up. Eventually things need to be tweaked a bit to compensate for the inevitable errors.
Such is life.

In regards to tracking one's location over long distances, I'm also impressed by folks who track the stars to know where they are. I'm amazed that aircraft are often equipped with windows to allow the navigator to take a fix on a few stars as a back-up to whatever their navigation system is (or was). Even spacecraft have been equipped with star trackers to know their orientation (and control it).

Steve in Peoria

rm -rf

I can't tell if that's an error or one of the grid corrections needed for N-S square grids on a spherical earth. (It does look like they either miscalculated or detoured around something.)

~~~
Canals!
I want to know how they surveyed canals in the 1820s with distance chains and simple transits. Obviously, the level is extremely critical, over long distances, so that water fills the canal and flows correctly.

RCMoeur 

The standard parallel corrections are taken 8 miles south of here at Baseline Road, which is the base line for the G&SR B&M system used to survey most of Arizona. So calibrating your speedometer on one of these north-south streets might not work well.

There was a local news story recently about someone pronouncing the road's name similar to that of a petroleum jelly product. Beats reporting on actual news, I suppose.

Here in the Valley of the Broiling Spherical Fusion Reaction, the canals generally followed the alignment of canals originally dug by the Hohokam in the 14th century - who of course used electronic total stations and digital terrain modeling to set their centerlines.

Steve B.

There are a lot of examples of surveys that once thought to be correct are later found out to not be. One of the more egregious is the so called "Four Corners" of the states of Colorado, Utah, New Mexico and Arizona. Where a nice tourist attraction exists that lets you place a hand leg in every one of those states. Unfortunately the actual corner boundary was found to be about 1800 feet to east. Nobody feels like moving the monument.

RCMoeur 

As a resident of the state in the southwest quadrant, a slight correction: yes, the intent was to have the benchmark and monument at 37 degrees N and 32 degrees west of Washington DC. But the survey lines converging on the monument are the actual legal state boundaries, slight deviations and all.

Now if you want survey shenanigans, talk to New Mexico about how Texas carved off a slice of their eastern edge - and they're still riled up about it.

Pompiere

A long time ago, before GPS and cell phones, I ran SCCA road rallies. The rally would always start off with a few miles for an odometer check. The route instructions would tell you the start and stop and the official mileage so you could calculate a correction factor for your car. That number was used for the rest of the rally to adjust your speed and mileage calculations. It wouldn't be very hard to do something similar to get a correction factor for your bike computer. Just use a GPS or phone to measure a course of a few miles, preferably on a straight piece of road, then calculate a correction factor for the bike(s) in question.

Reynolds 

I compare my computer distance to distance measured on Google Earth and correct accordingly. Not sure about GE actual accuraccy though...

Steve B.

Or ask the US State Dept. about the island that Canada stole in the Gulf of Maine/Bay of Fundy. I love this geographic silliness.

gugie 

You work with any young'uns? It would seem that schools don't teach signficant digits anymore! "The electronic display said 12.385", then you look at the accuracy of the sensor and it's +/-0.5%...

Andy_K and I measure our rides together by how many beers it takes to rehydrate afterwards. It's a good system.

steelbikeguy

My recollection is that significant digits was a bigger deal in the era of slide rules and vernier micrometers and all of that other analog stuff. Getting to 3 significant digits might be as precise as things got, so there wasn't much meaning in any digits beyond the 3rd one for any calculations using those measurements as an input.

Granted, it's a bit esoteric/archaic now, and really doesn't affect most folks.
Still... I know folks that will ride around the parking lot after a bike ride, just to watch their computer/gps/garmin/etc. roll over to 50.00 miles, and I just sorta shake my head and sigh.

Steve in Peoria

DiabloScott

As the engineers in the real world say: "Measure it with a micrometer. Draw it with a crayon. Smack it with a hammer."

Pompiere

We had a variation "Measure it with a micrometer, mark it with chalk, cut it with an axe."

Kontact

When I started having multiple road bikes I stopped buying computers. But then I got into a riding with que sheets, so I "replaced" my computer with the minute hand on my wristwatch: 3 miles = 9-10 minutes, depending on hills.

bikingshearer 

There is an event custom made for you right here.

You're welcome.

Mr. 66

The Seattle area has a big survey mishap. It's known as the Renton 'S' Curves on the I-405. They were building the freeway from opposite directions and had to quickly re-engineer the freeway into the curve to compensate.

I myself don't use any bike computers or guidance aids.

mpetry912 

this was a good discussion, altho it should probably be in the "lighting and electronics" section of BF. I especially liked "rm -rf"s track plot showing the 1 pps position fixes. that was new to me. Thanks.

Since the original question was "how accurate should your GPS bike navigator be" it's probably worth a discussion of the GPS satellite navigation system and how it works. This is a fascinating topic.

First off, there is NO position information coming down from the GPS satellites, of which there are 24 on orbit, plus some spares. this is an important point. Of the 24 satellites, you can see at best, 8 or 9, and your receiver will pick the 4 or 5 satellites with the best signal;. The Block IIF and Block IIIa satellites currently on orbit ONLY send a very accurate time signal, along with some spacecraft ephemeris (position) data. The satellites, which orbit at 12,600 miles above the earth, send their time signal on 4 (Block II) or 5 (Block III) frequencies in the range of 1.175 - 1.76 Ghz ranges. There are several levels of time signal (referred to as the TOT) embedded in different GPS signals, but the one that we civilians get to use is called the CA code, which stands for "coarse acquisition", and when processed by the receiver, the best possible accuracy to be obtained, assuming perfect signal reception and the best possible satellite arrangement (multilateration or PDOP) will give a position accuracy to about 4.9 meters or 15 feet. Of course there are additional GPS signals, some of which are encrypted for military use, that allow position accuracy considerably better than that :

Now it gets even more interesting because your little Wahoo or Garmin unit is doing some clever stuff with the received data. The 1pps position estimate is based on the receiver's ability to process all the time signals and derive an estimated position. The first thing it does is "snap" the position estimate to the receiver's street map data, allowing you to see what street you are on ! This would be very hard to do with satellite data alone ! The second thing it does is derive an estimated speed based on the distance between the most recent position fixes. Now you can see where you are, how fast you are going, and how far you have traveled.

Several posters above have noted that wheel sensors are more accurate indicators of speed and distance, and yes this is true, if correctly calibrated.

What is interesting is that the accuracy problem is shared between the satellite signal and the receiver's map and processing of the received signal. Your little Garmin or Wahoo unit puts a lot of navigation power on your bike for just a couple hundred bucks.

Some pics of the GPS satellite constellation and a Block II satellite are included below

if you want to learn more, the wikipedia page is very helpful. I wrote some of it. Global Positioning System - Wikipedia

/markp

Fredo76

My little CatEye speedometers are not complex or interesting enough to distract me from the scenery, or anything else. They provide occasional useful information, so I'm keeping mine, thanks!

Fredo76

Too bad you can't do a correction factor for your GPS, because a well set up speedometer is more accurate. It's like measuring a shoreline - how fine a measurement do you want? They get longer with smaller units. The bike speedo tracks your actual path, and the error can be set to less than 0.5% (1 cm circumference resolution compared to over 200 cm circumference).

jdawginsc 

2 beers adds 4 miles...

gugie 

Which makes s 41 mile ride 45, rounded it’s 50.

I like it.

sd5782 

So has anyone ever checked the old mechanical cyclometers that mounted at the axle? I know Schwinn of course offered ones, and there are/were many others. It would be interesting to check with the modern devices we have now. I’m still sticking to my story of a 600 mile circle Ohio ride with 3 centuries as recorded by the Schwinn unit in about 1980.

Steve B.

[QUOTE=mpetry912;22867731]

Since the original question was "how accurate should your GPS bike navigator be"

The original question made no mention of GPS. I think most of those replying assumed that was the target of the discussion.

mpetry912 

you're right.

maybe I will delete my post !

/markp

Inusuit

All I want from my very basic Garmin during the ride is how fast and how far. My Apple watch delivers more datum to my iPhone - speed splits, heart rate, elevation changes, map, etc.