I know this is off topic for the OP and I apologize for this. I do think this is somewhat relevant since you need to know how you determine how far off the speedometer actually is and what you can expect to be accurate. And besides, you did ask.

Here is one link for you and there are many if you are looking for this type of technical information. Different manufacturers use variations, but in general this really is how it works.

I don't want to go all nerdy, so if you like that sort of thing, Google up "Kalman filter GPS" or maybe get a copy of this book and have some serious fun.

For this discussion, I'll try not to make too many typos, or over-simplifying assumptions, but I can't describe the process in detail here. My goal is to explain why GPS receivers calculate speed as they do. Have some popcorn while you read.

My statement on why position change over time is not used does not depend on the current state of Selective Availability. Of course, I am aware that SA has been turned off since the first gulf war. This does not correct the over-estimation of instantaneous speed when calculated as position change over time. Even small errors in position fixes with that algorithm introduce too large an error in instantaneous speed for it to be useful.

To see why, imagine your car is motoring along and INSTANTLY moves a few meters at random (the position fix error associated with the GPS). Think about what that change would do to speed as the discrete steps are integrated. Instantaneous speed is inflated using this method because the total distance moved will appear to be greater than the actual distance moved. The higher the sample rate, the more the distance in the same amount of time and therefore the more the speed is overestimated. This error would also be magnified at lower speeds where the actual position change between samples could be much less than the position error of the fixes themselves.

You may say - "Well, just increase the position fix time interval to reduce the error component." Hummm - Sounds like this might be a good plan. Assuming you don't really need "instantaneous" speeds, what might this do? Using longer time intervals so that greater distances are covered between position fixes, position change is large with respect to expected error in position so that position change over time becomes more accurate. But what if the car changes course between samples?

Unfortunately, any change in course will cause speed to be underestimated, as the distance the car covers will always be greater than the position change. To see why, take it to the extreme. Imagine a sample position fix interval of an entire day. You start in your garage with the first fix and drive around all day, returning to your garage for the second fix. Even though you drove all day, you kept changing course so your position change is 0 miles and the calculated speed is also zero MPH. If you didn't change course at all during the day, on the other hand, the actual and calculated speed would be the same. If cars didn't turn, this would work great.

This is not to say that the change in position over time is not used at all, but it is not the primary factor in the speed algorithm. A GPS uses PLL's (an implementation of the Kalman filter I talked about above) to lock on to the frequency of the signals from the satellites. It knows with great accuracy the phase of the signals. Phase shift (the Doppler effect) found by the PLL is used as the primary component of the instantaneous speed value.

 

I'm no expert, but I'm going to throw a little elementary physics wrench into the idea of checking your speed with a frequency shift. When a train goes by you with the horn running, you can hear the difference in speed as it approaches and as it departs, but not as it's right next to you. So wouldn't that mean that in order to get an accurate speed reading via a frequency shift you would have to be heading directly towards or directly away from a satellite? Also, assuming a satellite was right on the horizon behind you, would that signal make it through all the atmosphere to reach your receiver? This is the one thing that makes GPS speeds less accurate in that it's difficult for a GPS receiver to calculate your altitude if there isn't a satellite closer to the horizon, which then translates into speed errors because the distance between one spot and the next can't be accurately calculated based on the change in altitude if you're climbing or descending.

At least that's the discussion I've heard based on what will hold up in court should you happen to wish to introduce into evidence a video of your GPS showing your rate of travel preceding a traffic stop.

On top of that, GPS receivers do not get instantaneous time, at least not my Tom-Tom, it updates that number on it's little display every few seconds. So my speedo could be reading 25, but the GPS will show a speed somewhere around 15 because it hasn't received a consistent series of data showing my average speed over a few seconds. And yes a change in course would show an inaccurate speed, but you'd be hard pressed to prove in court that your GPS was showing the right speed when the officer has recent tuning logs on his radar gun showing your speed was different. I'd be surprised if any GPS company would admit to having their speed number show up with more accuracy than could the position data, and not only that but also show up as an instantaneous speed not an average speed over time.

That said, I've found that my '85 MR2 matched the 2k Celica in it's speedo reading both with stock wheels and tires as being higher than the actual speed by about 3%. I've since changed out the wheels and tires on the MR2 and it is spot on with larger wheels and tires.

 

More than elementary physics is required. GPS provides three seperate but related services - accurate position, velocity and time. Here, we are discussing velocity (well, really speed) and how it is determined. I'll give you a quote from the US Coast Guard NAVSTAR GPS User Equipment Introduction Manual.

 

I mentioned elementary physics to describe a situation that simple geometry could not solve. Indeed, your quote from the NAVSTAR user guide says: "Doppler is measured to provide an estimate of the relative velocity between the receiver and the satellite." It's used by military hardware to provide an estimate. And further: "they can be integrated over regular time intervals to produce deltarange measurements" which says to me that they are doing the same thing as checking position over time to get an average change position.

And since you ignored the rest of my message, I will re-iterate. It's unlikely that you'd be able to prove that consumer grade GPS receivers go to the trouble of doing frequency shift speed analysis when average speed over time is easier, requires less sensitive equipment, and less processing power, thereby increasing battery life. But if you find a non-military piece of hardware that does it, I'll be suitably impressed.

 

Sorry - I didn't intend to ignore any of your message. I read the whole thing, but only quoted the "elementary physics" part. More than a little snippy on my part and I will try not to do it again.

You are right - I am more familiar with limited audience devices rather than say a Nuvi 205W or something like that, but the issues are the same for any GPS based velocity measurement device. Maybe we should try something different. How about a shoe?

 

From your link, and I quote:

"Most receivers compute speed by a combination of movement per unit time and computing the doppler shift in the pseudo range signals from the satellites. "

It appears that the GPS velocity calculation uses both, not one or the other.

Thanks for the update, I was wrong to state that it's only position change over time, although that appears to be one of the measurements used to calculate velocity.

It also appears to be incorrect to state that only the Doppler shift is used to determine velocity, which is the claim made by others on this thread.

I stand corrected, but it appears that the real answer is more nuanced than many have thought.

 

@miniclubman,

Not only are you right about most units using both, but they can also use other hints to improve accuracy. Accelerometers, for instance are handy.

 

No.

What you say is true for one satelite, but we're working with a constellation of satellites, and the GPS locks onto at least 4 to get a good fix. In the case of one satellite you know the component of your velocity directly towards (or away from) the one satellite. Do that with another 3 satellites and you get 4 components of velocity on 4 different axes (not orthogonal axes). Now its a matter of geometry to work out what velocity vector would produce those 4 different component velocities.

 

Only Doppler shift can give you an instantaneous velocity. Any method which uses position vs time is giving you an average speed. I'm pretty sure the speed that a dashboard GPS unit spits out is only calculated on Doppler.

 

I'll skip the GPS discussion as I don't have one. I rely on the old fashioned seconds per mile. According to my chronograph at 80 mph indicated my speed is 78 mph actual. However I have just changed from 175/65/15 tires to 195/60/15's so I would think the discrepency should be 0.8 mph less according to Discount Tires tire size calculator.

 

Try reading the detailed explanation found in the link in reply #76 of this discussion. You only have to read the first paragraph.

 

You are correct, that the difference is 0.8 mph at 80 mph. Just remember that changing tire size changes your speedometer reading by a percentage, not a fixed amount. The size difference between 175/65R15s and 195/65R15s is 1% (1.07% if you want to be precise), or 0.8 mph @ 80 mph.

 

What? What's this? My Mini has a speedometer? Where?

 

I hear that the state police don't even bother to use RADAR on second gen MINIs, they just look at the speedometer as we drive by.

 

It is rather small, and hard to find, but it is there.

 

Ok, I've not been on for a while, but…

a) I am the one miniclubman replied to that started this bickering over GPS speed reading methodologies. Let me just say that the accuracy of GPS speed readings belies the fact that positional measurement is not a significant factor in its calculation. But rather than re-hash this issue on these pages, I'd like to direct interested parties to this post over here (which I wrote when I was still unaware of the doppler shift measurement method).

b) I asked the service guys at my MINI dealer about the speedometer accuracy issue a while back and they produced a service note MINI issued that says it's a known issue, and that they won't fix it unless it's off by a certain percentage (and even then, I assume the "fix" is just a unit replacement, not a software rescaling). I can't find the copy I asked for, but if I find that service note around here somewhere, I'll post it in here.

 

I know this is an OLD OLD thread.....

I have a 2012 Clubman, but this is rediculous... I am being told my speedo is over by 10% and that IS OKAY ?!?!? REALLY and this is acceptable?! am I suppose to be okay with this?! because the dealership tells me to deal with it?! Seriously so frustrated with it I may sell it.


Julie

 

Just get an idea in your head about what the actual speed is when you are going at your typical highway speed, and compensate in your head. I know that my speedo is fast by about 3 mph in the 75 mph range, so I know that when the speedo tells me that I am going 75, I am actually going 73. So if I want to really go 75, then I bring the speedo up to 78. It's never perfect anyway, and I bet that they don't change it for different tire sizes either. It's always better to be going a little slower than the speedo says you are anyway, keeps you out of trouble!
I sure wouldn't trade my MINIs in because the speedo was off, there are too many cool things about the car for that to be a deciding factor.

 

I wish mine was just 3 mph, but it seriously various up to 10% which can be up to 7mph on 70mph, or 2mph on 20mph... hard to keep my miles on going thru a schoolzone (we have speed cameras for everything here in Washington State). Really not safe, and I shouldnt have to guess how fast I am going, mostly I didnt like the attitude of the dealer or the national people I inquired to both telling me they are good and can do whatever they want and with in the regs of the law and to get over it...

 

There are only two things that can be done that, together, might accommodate the excessive inaccuracy of the Speedo.
First, is to RECODE the software and reset the intentioned inaccuracy of the speedo. It is intentional to meet European laws that the Speedo cannot read slower than the vehicle is actually moving, so in order to make up for the range of inaccuracy of the instrument, engineers artificially set the the software to read about 3-4 MPH HIGH. The same software switch can be recoded by anyone who has the software tool and harness. there are threads on the tool in NAM. A BMW or Mini dealer might be able to do it if able and willing.
Second, is, at next tire change, move to a slighter smaller diameter tire. The standard 205/45/17 run flat is 24.3 inches in diameter. Use the Tire Racks Specs to view Tire Diameter sizes to work it out. A 215/40/17 is 23.8 In diamter, or 2% smaller. 2% of 70 MPH is another 1.5 MPG lower, combined with 3-4 MPH from Software gets you 4.5 - 6.5 MPH lower.

 

Mine is a Clubman with stock 15" tires, I do believe. The car is at the dealership now getting serviced and they REFUSE to do anything about the speedo issue. They told me that 10% is okay and that is just the way it is. I called MINI and they took my official compaint, but they said it is up to the dealer to deal with. Period. And the gentleman that I spoke to in the "resolution" dept said this was the first he ever hear of.............. LOL I see lots of complaints of this now that I google it. But he knew that they are within Federal guidelines, so obviously he knew about it already.

 

I was wondering about that. My GPS can show 55 but the mini shows 58. I figured this was a way to burn through the warranty faster. I believe you about odometer accuracy, but I'm gonna test it myself.

 

I am new to this forum so I hope this post will be seen by the right folks.
I have just purchase a 2006 S hatch back 6 speed and the speedometer reads 80mph while the GPS reads 70 MPH . Part of this is a tire size difference: the OEM size is 205-45-17 and I am running 205-40-17 so that is appox 2.1 MPH faster reading on the speedo but that still leaves 8 MPH of inaccuracy. Is there a fix for this? I have read the threads and some inaccuracy is acceptable but over 10% inaccuracy seem to be very poor engineering on the part of Mini( BMW).Any comments or help would be appreciated.
Thanks
Hoss

 

My speedo is off 3mph high with either 195/55-R16 tires or 205/45-17 non-run flat tires. I use a Go Point BT1 module that plugs into the Mini ODB-II port. Dash Command is one of the iPhone or Android phone apps that work with this Bluetooth adapter. I have programmed that app for a speedo correction of -5.0 and am now about 1 mph too high. I am experimenting with a higher correction to get that app to read the same as my gps.
The Go Point system has many display screens of a lot of information (boost/vacuum), coolant temo, power, torque, mpg in several ways, engine codes, TEchnical Service Bulletins,, skid pad values, and track logging that works on empty rural twisty highways using google earth.

I hope this helps. 2009 Cooper S Hatch, the only one with blue wheels in Nevada!

 

Just in case anyone missed it. The speedo solution is in the modifications section under sub-forum electrical. You can use NCS Expert to code the ecu that drives the speedo. But only the digital one. There is a setting for the analog speedo as well but changing that doesn't appear to have any affect.