The fastest F1 car of all time?

[Juzh]

Literally the first time I've seen speed trap numbers being put into question. Ever. No one even thought about it before all this.

[Juzh]

I can only conclude that you have never seen a speed trap and do not know how they function. A speed trap consists of two points of reference a distance apart. The time between those two points is measured and the average speed between them is calculated. By definition, you cannot measure speed at a point as it is a measure of distance over time, thus if the distance is zero...

Yes, I'm aware of that. However, I don't think it takes very long to measure a certain speed with a level of equipment available to the F1 and is a moot point in this case.

It's really not a moot point. Let's say the trap is 1 metre long and the cars are traveling at 100m/s. The car will take 0.01 seconds to pass through that trap. Suppose the clocks are accurate to 0.001 seconds. The actual time shown for 0.01 could be anywhere between 0.0095 and 0.0105. That's ~95-105m/s or 342-378kph. Quite a range. There's timing gear that does down to 0.0001 and that would be accurate to within 1.8kph in this example. For this reason, 1m is a very short speed trap. So where is the speed trap? Your assumption that it's the start/finish line cannot be right as it's larger than a line over the track. For all we know, it's doing the measured speed on the fourth row of the grid.... or pit in. We don't have enough evidence to make an assertion.

For all we know, they ARE doing speed measurements on the finish line. Finish line is at the pit entry. As I've shown in the screenshot.

If systems were inaccurate teams would be all over it.

[Tim.Wright]

Thank you Tim.
Could you elaborate a bit more?
A slip angle sensor is basically a camera scanning the tarmac under it, isn't it? Can the distance to the tarmac be measured with appeopriate precision? How, a laser sensor? Is this practical during competitive running like Q3?
Also, the GPS based doppler sensor sounds awesome, is it literally working on the frequencies of the satellite signals? I was under the impression that GPS had a poor temporal resolution, so that lots of interpolation would be involved. Maybe this is wrong?
So I guess the question then is... Which of those sistems are installed in actusl competition race cars... during the competition itself?

Slip sensors have a different operating principle to a camera but their strong point is their distance measurement accuracy and for this they are often used for braking distance tests where wheels speeds are too unreliable. I don't remember the exact specs off the top of my head but if you search for kister slip sensor you should be able to find a data sheet. I think the error was well less than 1km/h at its max speed. Obviously if your distance measurement is accurate then your speed will be accurate. Their accuracy is independemt of their height from ground as far as I know. They require some correction at high roll angles but this is only of concern on bikes rather than cars.

Gps doppler is something Ive used a few times but never really understood. All I know is its using doppler calculations rather than differentiating its position because positional accuray and signal to noise is very poor from raw gps signals.

Pitot-staic tubes are probably the worst for velocity measurements. First off they need a correction for their position because air speeds up as it flows around the car. Secondly it will obviously contain any ambient wind speeds which happen to be blowing across the track at the time.

[Ogami musashi]

It's really not a moot point. Let's say the trap is 1 metre long and the cars are traveling at 100m/s. The car will take 0.01 seconds to pass through that trap. Suppose the clocks are accurate to 0.001 seconds. The actual time shown for 0.01 could be anywhere between 0.0095 and 0.0105. That's ~95-105m/s or 342-378kph. Quite a range. There's timing gear that does down to 0.0001 and that would be accurate to within 1.8kph in this example. For this reason, 1m is a very short speed trap. So where is the speed trap? Your assumption that it's the start/finish line cannot be right as it's larger than a line over the track. For all we know, it's doing the measured speed on the fourth row of the grid.... or pit in. We don't have enough evidence to make an assertion.

Did a bit of research. Back in 2003, two loops were set 30 meters apart (1). This is also what appears to be 9 years later (2). Apparently in drag racing they are set to 23 meters (3). In any case it is recommended to have at least 10 meters of separation for a vehicle going faster than 100km/h to have a correct accuracy (4).
It also appears all timekeeping system are down to 0,001 of a second (5,6).

So, the noise/signal problem you talk about is not really relevant there since 0,001 second over a third of a second (30 m/100ms-1) will only result in a 358/361 km/h range.

Now of course the length of averaging is 30 times the one you proposed so the car may have gone faster (or slower even if that is far less likely in a straight line) and its rate of acceleration may have changed.

This is here that we maybe have something. Wake evolution with speed and its interaction with the following car is non linear. It can be seen in onboard vids that the speed increase in F1 is not constant. Even if the acceleration rate is low, the variance can be significant.

Combined with what Tim Wright said about onboard sensors, there may be some credibility to that 378 km/h figure (in a slipstream context).


(1)http://www.nytimes.com/2002/09/26/techn ... -tick.html
(2)https://www.quora.com/How-can-Formula-O ... o-accurate
(3)http://my350z.com/forum/drag/142624-tra ... -feet.html
(4)http://www.tagheuer-timing.com/resource ... 1-2012.pdf
(5)http://www.ris-timing.be/tagheuer/Motor_Racing_EN.pdf
(6)http://www.stroi-intl.ru/en/happen/circ ... rmula_one/

[andylaurence]

So we have a 3kph tolerance and probably a 30 metre window where that speed happened. Is the start or finish off the speed trap at the finish line? If either. Assuming the latter, then the speed achieved is likely around 15 metres before that. Lots of tolerances add up. Add in a 0.001 tolerance at the start of the trap and one at the end, you suddenly have a tolerance of 0.002.

What's key here is that the speed of a car can be measured in many ways and they all have different tolerances, which may vary on circumstance. It's quite possible that the tolerance on the FIA speed trap allied to the tolerances on the location and that of the speed measurement of the data logger on the car has led to a large disparity in figures. It could alternatively be the PR team making a number up.

[Ogami musashi]

So we have a 3kph tolerance and probably a 30 metre window where that speed happened. Is the start or finish off the speed trap at the finish line? If either. Assuming the latter, then the speed achieved is likely around 15 metres before that. Lots of tolerances add up. Add in a 0.001 tolerance at the start of the trap and one at the end, you suddenly have a tolerance of 0.002.

What's key here is that the speed of a car can be measured in many ways and they all have different tolerances, which may vary on circumstance. It's quite possible that the tolerance on the FIA speed trap allied to the tolerances on the location and that of the speed measurement of the data logger on the car has led to a large disparity in figures. It could alternatively be the PR team making a number up.

Sorry i don't understand your first part. if you have an averaging either you have a constant speed or a superior minus inferior speed hence there's no indication as to when the speed was reached (this is why we try to measure speed with as little distance as possible).
As for the tolerance, the 0,001 is for the decoder so the tolerance of the loops are already factored in it seems (and probably some software related tolerance too).

Now from what i read and my experience as an official sprint kart time keeper, the FIA doesn't bring anything, the loops are part of the track infrastructure (and that's precisely one criteria of homologation for having minimum level competitions) so there nothing really added to the decoder tolerance (yes you can always add some probable errors like wear of loops, interference etc.. but that's nothing special). And even if you do 0.002. you just add two km/h in the range (357/362)

Now the way i understand it:

-the 366 km comes from the speed trap which is on site and the averaging leaves an error margin that is composed off: tolerance_error+averaging_window_error. the minimum error would be 2-3 kph with tolerance error alone. From that we would need to know the rate of change of acceleration to know how the speed was averaged over the 30 meters.
-The 378kmh smees to come from onboard sensors and/or GPS (datalinked to the pits) and thus according to tim are very precise. In fact according to (1) the precision of GPS (which is bettered by optical sensors) is 0,001 kph.

The way i interpret it:

Even with the averaging the speed trap measure doesn't lead to completely distorted measurements. The first reason is that the error tolerance over signal is low due to the window chosen. Next the averaging window can't hide too high variation simply because we are dealing with cars at high speed and thus with relatively low acceleration and the time (0,3sec) needed to span the window is relatively small to have lots of changes in rate of change of acceleration.

However, during the window, such change (in acceleration rate) may happen due to slipstream conditions and what is interesting is that if such a change really occurred (and what hidden by the averaging window) then the next few meters to the initial brake point may have seen a rather great acceleration of the car. If three cars were in slipstream this may be the reason.

Of course the most important point is to know where the speed trap loop is centered on.


We do not know much more but at least it seems the averaging is not the reason and hence we can have confidence in the speed traps (hourray!) for what they are: a measure of speed at a location along the track. How the speed will evolve after it is another story.









(1)ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=425848