Posted 22 January 2002 on Porsche Pete's Boxster Board

My apologies for this long post but if you are interested to know how your Boxster's speedo- and odometer is likely to be off, how Porsche apparently intended it to be and how you can determine the speedo/odo error for your Boxster then take a deep breath and read on, or maybe print for later usage. If the text is too much, you still may find the graphs (see links) interesting to look at. If you just want a suggestion how to determine both odo- and speedo error the easy way, you could scroll down to the end of this post and read where this is summarized.

First let me give you a short introduction. Joris and I are working on what we call our "v-t project". Our goal is to generate representative acceleration curves (velocity versus time) for our Boxsters with relatively simple means. The collected data is based on video recordings of the digital speedo during acceleration. Subject of our trials are my (previous) Zoef 2.5l my98 manual and Joris' 2.5l my99 Tiptronic. We have the recordings and are busy translating them into actual graphs and 0-100 km times. Much more on that later! Btw, I plan to do exactly the same for my current S in near future. This should deliver a nice real world comparison of these three Boxsters (a pity we have no 2.7l available to complete the range).

Now for the speedo error. The recorded v-t curves as described above have to be corrected in order to obtain *actual speed* versus time. Therefore we have to know the error of the Boxster's digital speedo at each velocity. In other words we want to know the relationship between the digital speedo-readout (v_dig) and the actual speed (v) for both cars. We put a lot of joined effort into this and it became kind of a project within a project :-) We are now satisfied with the results and believe to have established the deviation of both speedo- and odometer with sufficient accuracy. The interesting thing about this is that it will not only learn how the meters are off for the Boxsters we tested but probably for the Boxster (if not all recent Porsche's) in general.

Way back in 1999 Joris already checked the accuracy of both his speedometer and odometer and found some interesting facts. Here's a link to his early post:

http://members.fortunecity.com/pboxster/speedo/post_joris.htm

Our recent measurements (where we seeked for high accuracy) confirmed independently the next important facts which are believed to hold true for all Boxsters:

a) if you plot actual speed (v) against the digital speedometer readout (v_dig) you will find a perfect straight line (linear relationship) which is expressed by:

v = a * (v_dig) - c

'a' represents the slope of this line and 'c' is a constant.

b) the slope of this line, expressed by the parameter 'a' in the above equation, will be smaller than 1 with a typical value of 0.96.

A factor 0.96 means that v_dig is to be corrected by -4% (and to gain the actual speed (v), the outcome has then to be subtracted by the constant 'c'). Be aware that this factor 'a' changes when the wheel circumference changes. For tire wear the effect is rather small though and is easily estimated. The circumference of a Boxster tire is typically 2000mm and the tread depth of a new tire 8mm. Let's say the tire is worn down 4mm which means that the new circumference will be 1975mm. This equals -1.25% which percentage adds to the speedo deviation. The -4% becomes -5.25% (factor 'a' is now 0,9475). Of course this was just an example to see the effect of tire wear.

c) striking is the fact that the line shows an offset expressed by the constant 'c' and therefore will NOT cross the point of co-ordinates where both v=0 and v_dig =0. You will find this offset 'c' to be typically 3-3.5 kph. When your actual speed (v) is zero your speedometer would still read several kph's but Porsche was smart enough to not display that value and below something like 5 kph the speedo will only display zero!

This offset value 'c' is to be considered as a constant that will not change for your car because it is programmed by Porsche. Therefore it is likely to be the same for all Boxsters or at least defined by Motronic versions?

d) the odometer error is reflected by the slope 'a' of the straight line. The reason why is well explained in Joris' post. Our recent measurements confirm this. So as an example: if the slope 'a' equals 0.96, your odometer reading has to be multiplied by 0.96 (which equals -4%) to obtain actual distance and when the slope (for instance after tire wear) is 0,9475 the odometer reading has to be lowered by 5.25%. Once you know the constant 'c', you only need to determine the odometer error to find the correct formula and with that the speedo error at every speed.

Now let's see how the measurements were carried out and then look at the actual figures. We did three types of measurements:

1) determining the odometer error

this was done by comparing the odo reading against the hectometer markers (every 100 meter) along the freeway. All our freeways have these markers and they are supposed to be very accurate. I drove a section of about 16 kilometers and compared the actual distance with the odo reading. I did this six times to increase accuracy. Doing this only once already will give you a reasonable value but you'll need to measure a sufficient distance for accuracy reason.

2) determining single speed errors

again we used the hectometer markers and while driving at a (very) constant speed on the speedometer (cruise control is a great help here) the time needed to cover an exact distance (I choose 1000 meter) is clocked (stop-watch!). The actual speed is easily calculated. I did this several times for several speeds (100kph, 60kph, 20kph). The 20 kph measurement on a freeway is not a good idea but I was lucky to find a low traffic road close and parallel to a freeway section where I could read the hectometer markers for about 500 meters, sufficiently for a reliable measurement. Maintaining exactly 20 kph on the speedo needs a little bit of practise. A low speed measurement is important to have the constant 'c' determined with increased precision.

Alternatively, if you have determined the odometer error very accurately, then what you can do is drive at a constant speed (v_dig) for a certain distance on the odometer and measure the time. The real distance can be calculated and when divided by the time provides the actual speed. This method is especially suited for low speed measurements when you lack distance markers.

3) determining single speed errors by GPS

For both our 2.5l Boxsters we used measurements described under 1) and 2) in order to determine both odo- and speedometer errors. It was Joris' idea to borrow his son's handheld GPS (which is capable of displaying the average speed while driving). I visited Joris last Wednesday to show my new Boxster S and off we went in the S together with the GPS device :-) We measured at several speeds and were very curious to know how the line would fit. Interesting to note is that I took the winter wheels off my 2.5l and put them on the S so the measurements for both my 2.5l and S have been carried out with the same wheel circumference.

The next graph shows the result for my '98 Zoef 2.5l. The equation for the straight line was obtained as follows: the slope 'a' was determined by the odometer calibration which is the most accurate, see 1). Next the speed measurements at 20kph (5 times 400m each resulting in an average speed of 15.7 kph) were used to complete the equation:

v = 0.9575 * (v_dig) - 3.46 kph (Zoef 2.5l, my98)

In the graph I added the 60kph (3 points) and 100kph (9 points) measurements as well and see how well they fit. All individual measurements (17 in total) are represented by the red dots, most of them overlap though.

http://members.fortunecity.com/pboxster/speedo/g1kph.gif

For your convenience here is the equation and graph in miles:

v = 0.9575 * (v_dig) - 2.15 mph (Zoef 2.5l, my98)

http://members.fortunecity.com/pboxster/speedo/g1mph.gif

So the odometer reading must be corrected by -4.25 % (factor 0.9575) to get the actual distance !!

Because of the offset of the line, the speedo error is speed dependent and increases with decreasing speed. Now that the relation between v and v_dig is known, this error can be calculated for every speed. For instance at 30 kph (18.6 mph), the actual speed is 25.3 kph (15.7 mph) so the 30 kph on the speedo represents the actual speed increased by 19 % !! The next graph shows this speedometer error (%) as a function of speed. Just look at what speedo errors Porsche dares to introduce at speeds below let’s say 50 kph (30 mph)!!

In kph:

http://members.fortunecity.com/pboxster/speedo/g2kph.gif

http://members.fortunecity.com/pboxster/speedo/g2mph.gif

In the same way Joris determined the relation between v and v_dig for his Boxster and also found a perfect linear fit:

v = 0.9615 * (v_dig) - 3.0 kph (Joris' 2.5l, my99)

and in mph:

v = 0.9615 * (v_dig) - 1.86 mph (Joris' 2.5l, my99)

 So here the odometer reading must be corrected by -3.85 % (factor 0.9615) to get the actual distance. With the lower offset value of 3.0 kph, the speedo error becomes somewhat less, nevertheless at 30kph (18.6 mph) the actual speed is 25.8 kph (16.0 mph) so the 30 kph on the speedo still represents the actual speed increased by 16% !!

Where I find an offset value of 3.5 kph, Joris finds 3.0 kph for his Boxster. This offset is considered to be purposely programmed by Porsche but we are not yet sure whether the 3.0 and 3.5 is an actual difference between the two cars. The offset value is hard to determine with high precision and that's why this difference we find can still be caused by measurement inaccuracy. I really would like to know what exactly Porsche has programmed in our cars.

Last but not least: our first GPS measurements with my Boxster S. We measured 6 times at different speeds. The next graph shows the linear fit and the red dots represent the individual measurements.

in kph:

v = 0.9565 * (v_dig) - 3.0 kph (Zoef S, my02)

http://members.fortunecity.com/pboxster/speedo/g3kph.gif

in mph:

v = 0.9565 * (v_dig) - 1,84 mph (Zoef S, my02)

http://members.fortunecity.com/pboxster/speedo/g3mph.gif

The next graph shows the speedometer error (%) as a function of speed for my Boxster S:

in kph:

http://members.fortunecity.com/pboxster/speedo/g4kph.gif

in mph:

http://members.fortunecity.com/pboxster/speedo/g4mph.gif

 Striking is how close the slopes are for Zoef 2.5 and Zoef S. If Porsche did not change the conversion of wheel revolution pulses into distance this is exactly what you would expect when using the same tires as I did. So apparently there was no change relating this conversion between my98 en my02 and probably the offset value hasn't changed either. This confirms my believe that all Boxsters share the same conversion of pulses into distance (unless this is changed by the dealer) and with that, share the same annoying speedo and odo deviation (for comparable wheel circumferences).

Finally I would like to summarize how you could calibrate the odo- and speedometer the easy way with only two measurements:

a) determine the odometer error as described under 1)

b) carry out a speed measurement, preferably at a low speed like 15 mph and use one of the two methods described under 2)

Try to perform these two measurements as accurate as possible. Now you should be able to determine the equation v = a * (v_dig) - c

If you can make use of a GPS device see 3), measure at least two speeds, one low (like 15 mph) and one high (like 60 mph). The more the better. Fit a straight line through these points and find the equation v = a * (v_dig) - c

I like to thank Joris for his input, his ideas and the wonderful discussions we had and still have relating these kind of measurements and relating the Boxster in general. Joris, please do not hesitate to correct me where I was wrong, to give your opinion where it deviates from mine or to add any information you think is worthwhile. I want to thank my wife Candida too for helping me a great deal in carrying out the measurements :-)

Rene