wross996tt

So you are basing error estimates against another measurement system with unknown measurement error...sorry, but this just doesn't fly mathematically.

turbolou

Since the speedo is electronically controlled through the computer. Some one with the correct software should be able to recalibrate the error. that would be the best option.

Beej

Hmmmm:

a) I think your argument should be that it does not fly scientifically, rather than mathematically, as there is no error in my maths!

b) As for the error of the GPS system, GPS accuracy is widely known to be within 1% or better,

c) To check on b), the GPS/speedo readings were also checked against the tachometer reading at the various speeds. If you look in your Porsche drivers manual, there is a graph that shows engine speed vs car speed in each gear. This data can also be calculated simply for any gear if you know the gear ratio, the final drive ratio, and the rolling diameter of your tyres. I also did those calculations, and found that they correlated with the drivers manual, subject to a 1-2% error due to the fact that my tyres are 315/30/18s which have a slightly larger rolling diameter than the standard 295/30/18s.

As a check for this, drive your car in 6th gear sometime at say 70mph steadily, and note the engine RPM on the tacho - check that against the graph in your drivers handbook and a reckon you will find a pretty significant speedometer error.

So, bottom line, the GPS measurement accuracy was also "calibrated" if you like against the engine speed vs car speed calcs, and found to be accurate, so the error there was in fact known (to within a fair margin for the purposes of this exercise anyway).

I have absolutely no doubt that the error in the speedo in my car is pretty much what I determined posted above!

Any thoughts on my idea to go for 295/35 rear tyres to cancel out the 5% error?

As for getting the speedo re-calibrated via a software flash, the dealers cannot do it (I asked them), and none of the flash tuners I contacted about this can do it either Very frustrating, as previous cars I have owned this type of re-calibration has been a doddle!

Cheers,

Beej

wross996tt

Well actually, by error I assume you meant standard deviation. These are not additive, so your math is wrong.
You mean widely accepted not known. The second issue is precision (repeatibilty for example) you have no estimate of this.

Lets say theoretically you ran identical conditions...would the speedo report the same output? No...this is precision error.

I believe this will change the mean not the variance.

Calibration only deals with accuracy. Accuracy is the deviation of the mean from a known "true" vaue (its true because some gov't agency says so...here in the US it is NIST). Calibration does not effect the precision (variation).


cheers

Beej

Ok this is being made*way* too complicated! What we are talking about is an observable, consistently measurable ERROR in the read out of the cars speedo relative to the cars true speed. Eg, I observe that at what I now believe to be a true speed of 100km/h, my speedo reads 108km/h, always, every time.

I, and based on these forums, a lot of other 996TT owners around the world, seem to have noticed that our speedo's are in fact very optimistic. I wanted to measure the error at various speeds in order to understand whether it was a proportional (%) error relative to the true speed or whether there was both an absolute and proportional component.

Using a mobile GPS AND the tachometer readings at various speeds, I can consistently show the error in my speedo is APPROXIMATELY +5% of true speed + APPROXIMATELY 3km/h. Every data point I have checked has backed this theory up to within a reasonable (less than 1km/h) margin.

Every heard of the 80/20 rule? You can get 80% of a result for about 20% of the effort compared to the effort it takes to get a 100% result?

The stuff being raised about precision vs accuracy of instruments, statistical variation in measurements of the same observation (standard deviation and so on), calibration against NIST standards etc etc, is not relevant for my purposes as those issues are negated by the use of the word APPROXIMATELY above. Ie, the measurements I made are for all intents and purposes perfectly repeatable with a level accuracy of interest to someone wanting know the speed of their car to within say 1-2km/h! I don't need NIST to tell me the standard for defining whether a car is in fact traveling at 100km/h or not! If I wanted to take my approximate formula for my cars estimated speedo error the the next level of accuracy, then the precision/accuracy of measurement issues raised would become relevant.

So I guess this must be "clear as mud" now for all the folks interested enough to read this thread??

Cheers!

Beej

wross996tt

You're kidding right....OK I get it it's a joke. You are the one complicating it with worrying about 1-2-5% error. Lighten up....just drive your car.

Beej

LOL! - Check the maths again: try 8% error @ 60mph; quite noticeable, and I and many others find it very annoying! Which was the point of this thread in the first place!

Anyway someone asked for the data, I got it, now I'll

Cheers!

Beej

SimonK

Speedos in cars and bikes are firstly inaccurate for legal reasons. Manufacturer does not want to get involved with speeding tickets being blamed on them. This is why they design speedometers with a slight over reading error. Secondly, they can’t be accurate because of physical limitations. Tyre’s physical properties change all the time, loss of thread and heat in them, this all accounts for different speedo reading. You can install electronic speedo correctors and these are readily available to purchase from the web.

As far as GPS; the accuracy is solely dependent on how many satellites your GPS device is receiving at the time. Navigation is possible with as little as three but your GPS device can operate up to 10 and more. The accuracy is greatly affected by the number of satellites your GPS device sees. The more it sees the higher the accuracy. Even GPS is not 100%.

As someone said it’s a good thing in a 911 turbo as your ticket will always be for a little less. :-)

slant911

I said it in a previous post, but I'll say it again because it was an odd occurence I observed while running the mile shootout last year here. My speedo is consistantly off by a few mph. However, I ran the mile at a clocked 170.5 MPH (radar gun and speed trap at the mile mark) and my speedo read about 171 as I crossed the line. So my thinking is that it gets more accurate as you go faster. Maybe at 200mph it's right on

Of course their equipment could have been off by a bit too, but I wouldn't think that the speed trap equipment, the radar gun, and my speedo were all off by the exact same amount (relatively).

Zuffenhausen911

My 996 is is the shop today for scheduled service. I told the representative about my speedometer being 4 mph over. Hopefully they can give me an answer. I will post the answer.

jbossolo

You guys crack me up with all this mumbo-jumbo back and forth over math and % errors!
Just drive the friggin' thing until the wheels fall off!!!
Just my 2 cents (or is that 1.9999999998 cents due to measurement error?)

CaptPegLeg

I've had lots of guys that fished with me, bring handheld GPS with them and compared it to the Northstar (big antenna, bigger price) unit on the boat. They were never very close. I don't think they can be, nor were they designed to be, all that accurate. FWIW

viprklr

I used my handheld Garmin GPS to measure my speed at various marks on our way back from Florida. It show consistently at almost all speeds that the car's speedometer read 1-2mph high over the GPS. I would tend to believe the GPS over the cars. Now the kicker may be that the car has aftermarket 19" wheels rather than the stock 18" wheels. I would tends to doubt that the previous owner or installer re-calibrated the speedometer in the car to take the larger wheels into consideration.

Adam Bowles

I think the speedometer intentionally reads high. I have a Brockway R920, which is a digital multi-gauge that plugs into the OBD2 port. When my speedometer reads 80 mph the Brockway reads 76 mph. Since the Brockway gets it's info straight from the ECU, and the two speeds differ, Porsche must have intentionally made the speedometer read high.

Last930

The owners manual for my '79 930 has a chart showing speeds for each gear/rpm, which shows about the same % difference as reported on the 996TT here. Is there a chart in the 996 manual?