In my '05 MCS, I've used the measured-mile facilities on I-5 several times. They have a five mile section that I use to verify the accuracy of my odometer, and with my 205x55x15 tires, the odometer agrees with the measured mile to within a percent or so.

Enter my new Garmin Streetpilot GPS, which offers a display of speed, distance traveled, etc. When I used it on the measured mile section, the GPS agreed with the odometer on the distance, to within maybe 25 feet.

However, the GPS and the MINI speedometer disagreed by around 5% - the speedometer read about 3 mph higher at 60 than the GPS-calculated speed (and 2 mph high at 40, and 4 mph high at 80) .

Now the good news is that this has probably saved me a ticket or two, but I'm wondering if anyone else has the same experience?

Are all the MINI speedometers calibrated to read higher than the odometer, or is this just a quirk of mine?

 

Oops - sorry - I just realized that there are other threads on this topic. Interesting though, that the odo reads correct, but only the speedo is off...

 

I also drive a stretch of I-5 that has a speedometer test, but I have never used it. However, according to my GPS, my R53 speedometer reads about 1 to 1.5 MPH fast at any given speed. I have not, however, tested to the accuracy of the odometer.

As a side note, Honda recently settled a lawsuit over this issue. It seems that some of their cars were reading about 3% high. Honda's claim was that their cars were within the error standards (set in the '70's when the odometer was a straight mechanical device). As far as I understand it, these days, odometers use a calculation to determine distance and speed so there is no reason to have such a high error rate.

 

It seems that all BMW speedometers read high, sometimes as much as 5% with stock tires, and there is no easy way to re-calibrate them. The BMW boards suggest that there is some regulation in Europe that says that speedos cannot read low and there are hefty fines associated with that so they set them high to give them some wiggle room. Sort of like the regulation for airplanes that says that the only time that the fuel gauge needs to be accurate is when the tank is empty. So I figure this extends to our Minis too.

For what it's worth when I changed the runflats for Pirellis the speedometer became accurate and the odo moved a bit downwards.

Rich

 

First lesson in GPS: They can be off by as much as 100 feet or more. GPS satelites are owned and operated by the US military, and have a built in error which military and "official" GPS units have the ability to filter out. The don't want the bad guys to be as accurate as our own military.

Never rely on a GPS 100%.

 

You are right that GPS is not positionally accurate (unless you work for the military) but the MPH/KPH readouts have usually been within 1 MPH in my test using a radar gun and TomTom GPS.

That said although you are correct that the errors can be as much as 100 feet or more depending on the unit you are using, the GPS units used for race data logging do a really good job these days and my guess is that they get the turns within 5 feet or so judging by the data sets I have played with.

Just my experience.

Rich

 

The induced-error feature you're talking about is called "Selective Availability", but it's been turned off since May of 2000. The consumer GPS units are now every bit as accurate as the military units (if not more so). It's true that the government still retains the ability/authority to re-activate "Selective Availability" if they need to, but unless we start getting attacked in our own country using GPS-guided weapons, I wouldn't worry about it.

EDIT - if your positional error is routinely greater than 20 feet or so, you may have a problem with your GPS antenna or receiver. Even without WAAS, the horizontal positional error on the three units I've owned is normally around 7-18 feet.

 

And my GPS was showing within-8-ft. accuracy while I was checking the odo and speedo.

 

Your experience is totally normal. Even though the speedometer and odometer are fed from the same sensor, the two systems can interpret the information differently. That's why the speedometer can be calibrated without affecting the odometer.

The old mechanical systems were the same way. The same spinning cable fed both the odometer and speedometer, but while the odometer used a fixed system of gears, the speedometer used a spinning magnet to move the needle, and the distance between the magnet and the needle could be varied to calibrate the speedometer.

Your GPS velocity error can be as low as 0.1MPH, but in almost all circumstances, it shouldn't be worse than 0.8MPH.

 

Since this thread seems to have uncovered some very knowledgable GPS gurus, can someone point me to a test or technique that can use to see the absolute accuracy of my unit. I'm assuming I would need to read out the UTM data and compare it to a known spot on the earth... are there convenient markers that the public can easily find?

I had a freind once tell me about clubs that do weekend games where finding hidden objects using GPS units was a key part.

YD

 

The sport you're friend's talking about is called "geocaching", and you can find a bunch of information on the web about it. Unfortunately, none of the geocaching techniques will help determine your unit's accuracy, because the location of the caches are only as precise as the GPS used by the person that placed the hidden object.

Also, it's important to know what *kind* of accuracy you're interested in measuring. There's "circular error probability", which just measures accuracy in latitude and longitude, and "spherical error probability", which includes altitude error as well.

You're probably just interested in circular error probability, so a typical accuracy rating would be something like "circular error of 5 meters 95% of the time". That means that 95% of the time, the GPS will give you a reading that's within 5 meters of its true location.

Testing it yourself will depend on what you can find for a reference point, and how your unit displays latitude and longitude. For instance, my Nuvi only displays lat/long to the closest .001 angular minute, which means that it's incapable of displaying different lat/longs for two points that are closer than about 8 feet apart. The internal computer may store the location with more precision, but the display only shows the position to the closest thousandth of a minute.

If you can't find a surveying reference point or something similar that's been accurately-recorded down to the "gnat's ***", you can still get a fairly reliable measure of your unit's accuracy, regardless of where you are. This is because all of a GPS's reported positions tend to be evenly distributed around the unit's true position (In other words, you won't get readings that are consistently too far North or consistently too far West).

So, place your GPS somewhere it can get a good satellite lock, and record the latitude and longitude every 30 seconds or so, until you have 100 readings. (You really only need to record the last few digits of the latitude and longitude, since those are the only numbers that will be changing).

When you're done, throw out the reading with the highest latitude, the reading with the lowest latitude, the reading with the highest longitude, and the reading with the lowest longitude. Then, measure the difference between the highest and lowest latitudes of the remaining 96 points, and the difference between the highest and lowest longitude. These two numbers will give you the length and width (in angular minutes) of the rectangle that contained 96 of your 100 data points.

It's a little complicated to convert the difference in longitude into feet, since the longitude lines aren't spaced evenly throughout the globe. But, if you're willing to believe that the "error rectangle" is actually a square, then you can just convert the latitude difference into feet, and assume that the longitude difference would be about the same. This is a pretty fair assumption, and is certainly valid, considering that we're working with the limited-precision locations from the GPS screen in the first place.

Fortunately, the lines of latitude are evenly-spaced across the globe, regardless of your location, so this conversion is easy. Each one-thousandth of a minute (.001') is a little over six feet. So, if all of your latitude readings were with .005' of each other, that means that your "error square" is about 30 feet across. That means that 96% of your readings were within about 15 feet of your true location. (Remember that we threw out the four most-extreme reported positions).

I made some assumptions and simplifications in this method, but nothing that's going to seriously affect the accuracy, and it keeps the math down to a simple minimum. Keep in mind that your result won't be exactly the same as the 95% Circular Error Probability for your unit, but it will be pretty close.

 

My understanding about GPS is that you cannot rely on it for instantaneous reading of speed. It cannot directly measure speed; it computes a speed from successive position readings and the amount of time between them, and takes an average over a number of readings. The GPS speed should be accurate if you are driving in a straight line at a constant speed.

 

The I-5 test section I have used is flat and straight as an arrow, and gets highest-accuracy GPS readings as there are no obstructions above the horizon. I used the snooze control to keep speed as constant as possible.

If the speed and mileage readings the GPS gave during this test were off by .2mph, I'd be VERY surprised.

So the issue that this brings up is that since there is a single source of wheel-rotation pulses feeding the speedo and the odo, and they differ, then there is plainly a programmed difference between the two.

I'll see if my local dealer is willing to recalibrate the speedo...

BTW, there is a small black box called the ERA (Electronic Ratio Adapter) that modifies the wheel-sensor pulses before they get to the computer, and allows you to adjust the speedo/odo to about .1% accuracy. http://www.abbott-tach.com/era.htm
I used one on an RX-7 - worked great.

 

You might have a hard time getting a dealer to adjust your speedometer, unless it's *really* out of whack. Someone here posted the technical bulletin from MINI, and the speedometer is allowed to read high by up to 10% of the actual speed, plus an additional 2MPH. So, if you're going 60 MPH, the speedometer is allowed to read as high as 68MPH and still be considered "in spec". The bulletin also warns that under no circumstances may the speedometer read *lower* than the car's actual speed - not even by a little bit. So, a dealer may be a little bit nervous about trying to get the speedometer "spot on", for fear of it reading too low.

And as for the Electronic Ratio Adapter, if I understand correctly, any correction it introduces will affect both the speedometer AND odometer by an equal amount. So, while you could use it to make your speedometer accurate, it would also cause the odometer to read low.

Oh, and your sigfile quote is usually attributed to Abraham Lincoln, not P.T. Barnum, although there's some controversy as to whether Lincoln even said it.

 

Scott, thanks for the explanation and information. You are obviously more knowledgable in this area than the average bear.

This is one of the great things about forums like this... if you ask the right question, there are usually several people who are true sources of good knowledge about most any subject.

I totally understood what you said, but I suspect I most likely will not do the experiment. I was just curious if there were UTM surveying markers that had the info availalbe and if I could find a couple and compare those numbers to my Nuvi. Just an intersting experiement for my level of interest and entertainment.

thanks again

YD

 

Unfortunately, surveying is one of the areas I know very little about. But even if you find a surveying marker with a known position that's accurate to within the displayed resolution of your GPS, you're still going to have to watch your GPS for at least a few minutes to see how much the displayed position varies over time, and then convert that variance into feet.

But if you do find such a location, and want to take the measurements, watch your GPS display and keep track of how much the latitude (the North/South part of the readout) changes over the space of a few minutes. If your GPS displays lat/long in decimal minutes, the latitude will look like this - N 36° 56.549'

You'll probably only see the last two digits change. Each .001 change is about 6 feet, so if the decimal minutes part of the GPS' latitude reading changes from .540 to .550 while you're watching it, and the decimal minutes portion of the actual position of the marker is .545, that means that your GPS was always within 30 feet of the true latitude, and that's a fair approximation of the Circular Error Probability for your particular unit.

 

Yes, by changing the source, the ERA would adjust both readings by an equal percentage.

I mentioned it because there are uses where one might want an accurate speedometer at the cost of making the built-in odo read a few percent off. One could also switch an ERA in or out of circuit, and have the choice of a corrrected speedo or not.

Personally, having to compulsively calculate the true speed every time I look at a speedo makes me nuts. I'll correct it if I can figure out how to get into the diagnostics and adjust the speedo, or pay a dealer to do it for me.

 

This whole reading over thing is actually a UN deal. Here are the details:
http://www.unece.org/trans/main/wp29...egs/39rv1e.pdf

Just be glad we still don't have the old frederal mandated 85MPH speedos!