DL1 Speed Accuracy

[Adrian Williams]

How accurate is the speed from a DL1?

Based on gearing calculation I'm getting graphs that are under reading by 3%, is there an adjustment in the software or is the gearing calculation wrong (I'm pretty convinced it's not)?

Adrian

[tristancliffe]

In your gearing calculation have you taken wheel slip into account, which is usually between 3% - 20%.

And the dynamic increase of the tyre circumference through rotation and pressure/load variations?

I'd have though speed within 3% isn't too far out compared with merely theoretical calculations based on static measurements.
_________________
Monoposto 2000 - Reynard 883 Toyota

[Support]

Hi,

Under normal/good GPS conditions, the speeds from the DL1s GPS system are certainly accurate to within 1km... typically much better.

Kind regards,

Andy
_________________
---------------------------------------------
Posted by the Race Technology support team

[stephen hooker]

My early series Mk1 DL1 GPS speeds are pessimistic by 5 to 10% under transient condtions, when compared to wheel speed sensors and rpm conversions. This means that cornering speeds indicated by GPS are inconsistant, and I generally disregard them. The GPS speeds are closer to reality under steady state conditions down straights, but how many races are won on straightline performance alone? The newer 10 Hz units may be better . . . anyone experienced with the newer units?
_________________
Stephen Hooker
Sydney, Australia

[tristancliffe]

[stephen hooker]

I know my % slippage from comparing front to rear wheel speeds on RWD car. Can also see brake lock-ups and slippage. I know it is GPS error wrt to true speeds. Hey, I have even been to different tracks with speed traps, and wheel speed always agrees . . . GPS does not always agree. I know of later units that appear to be more accurate than mine . . . but I also know of later units that are actually worse. Has anyone else used wheel speed sensors on the DL1's? Feel free to chime in!
_________________
Stephen Hooker
Sydney, Australia

[tristancliffe]

I confess I do not have wheel speed sensors, and nor have I been at tracks with speed traps (which are usually in silly places, like the end of straights).

However, I'd not trust the absolute accuracy of the speed traps anyway - there will be errors in them, and they won't be calibrated that regularly as it's merely a comparison tool and doesn't need to be accurate.

Also, when comparing front to rear wheel speeds bear in mind the undriven tyres will have some degree of slip.

I'm not saying it isn't GPS error, because all the measurements will have some error, whether it's dynamic tyre circumference, tyre slip, GPS signal and GPS drift correction (applied at the GPS end)... However, I'd have thought that anything within a few % is pretty good really.
_________________
Monoposto 2000 - Reynard 883 Toyota

[faraday]

You may be interested to learn that GPS speed is (was) the preferred reference in traction control systems in elite forms of racing.
By the very nature of racing, slip is usually present on all four wheels in transient conditions, but varies considerably.
The effective radius of tyres varies around the course and even tyre to tyre.
Not many racing cars have four wheelspeed sensors. Regulations frequently restrict this. The most common number is one, which is bordering on useless, unless it is the only source of speed for a data system.
Assuming that a reasonable attempt is made to measure laden rolling circumference, the greatest consistent source of vehicle speed error, is from ignoring the car's track.
Inside wheel vs. outside

Wheelspeed based systems selling for many times DL1 price have algorithms for "correcting" speed, but even when they have associated vehicle details such as track width, it is not used. The normal logic is to take the highest speed.
After all, we all like big numbers, don't we?

I am not suggesting that this is Stephen Hooker's problem, but it is very rare to find anyone in the paddock who has pessimistic wheelspeed callibration.
As Tristan states, all measurements have error.
I have certainly seen error in GPS speed from R-T and other products, but it is not consistent like Stephen sees. I have tested 10Hz and 5Hz and non R-T units simultaneously in the one car and not seen a difference in speed other than when for some unknown reason one gets it wrong.
This is in a very fast racing car with four wheelspeed sensors. If anything, the GPS errors seem more common at higher speeds (250kmh+). I would expect, as Stephen is observing, that the error would be greater at times of high accelerations because of the low update rate, but the 100Hz accelerometer information can fill in the gaps.
We need to take care in the mounting of our units. Soft AV mounts are not sufficient if the mounting surface is vibrating at high amplitude. The more racey a vehicle is, the harder it is to find a nice surface. Accelerometer data is of reduced value if it is full of noise.

I have seen some very large errors - 15-20% - in a few corners in a different car with no wheelspeeds. The driver was very inconsistent, but these speeds were physically impossible. I was lucky that we had revs to proove to the driver that he wasn't that inconsistent, even though he didn't mind thinking that occasionally he could out do Senna.
Computing speed this way demonstrated the strange GPS error. The GPS trace was smooth and had its minimum value where expected; quite believable, but...

In summary, GPS speed measurements, especially those processed with accelerometer assistence, are generally superior to wheelspeed, but there might be times when something goes awry. In my experience at racing circuits, this is obviously not due to low satellites or multipath, such as the urban canyon problem. When it has occurred with multiple units in the one car, only one has been affected. The obvious errors are too infrequent to point to a systematic problem.

If there is a consistent error, I would not accept it as part and parcel of GPS, but rather a fault in the individual device. Wheelspeed sensors have their purpose, but we should not need them to check on our GPS performance.

[Support]

Hi,

It would certainly be worth upgrading to the latest software as the speed processing is significantly better. It does a cross correlation between GPS speed and accelerometer data to align them within a few 1/100ths of a second before feeding them into the speed/position kalman filter... it's relatively complicated stuff, but the main point is that the speeds and positional accuracy are better than ever now.

I've done some testing under transient conditions and the speeds are really excellent (well within 1kph under a 1g transient) - certainly much much better than is possible with a wheel speed sensor. The very latest 20Hz GPS data is even better - so good in fact that the error is very difficult to measure, it's can basically be considered "right".

As you say at steady speed the error is tiny.... that's the easy bit

I haven't got any old DL1 here to test - but I would be quite surprised if there was a large difference.

Andy
_________________
---------------------------------------------
Posted by the Race Technology support team