Some one on another forum asked about some comparison of data so here's a simple example. I figured I'd post it here just to keep everything in one place. Keep in mind that it's not as easy to draw conclusions looking at the data days/months/etc... after the event. The best comparisons are always drawn trackside while the run is still fresh in your mind. This way you know what the conditions were, traffic that would through off your data or line, etc.. The best practice is to take notes on your recently run session when you first get off track. I haven't started taking notes yet but intended to start doing that now.

This data is with GPS and accelerometers only. It's from that TracQuest example (not me) I mentioned earlier where you can see the improvement made over the weekend. No additional inputs used, less than 5 minutes to get the unit situated in the car and loose wires secured. More inputs help but unitl your used to looking at the data it might just add to your confusuion.

2 fast laps from 2 different run sessions:

The cursor on the graph is indicated on the track map so you can see where you are on track relative to the data. Note the time slip graph at the bottom, this makes it easy to see where the time differentials have occured between 2 runs.

Looking at the are indicated area in more detail, as to why speed is being picked up in this area and how much:

Speed over the top of Madness is only 1mph different but sector exit speed is is up 6.5 mph!! Why?

Looking at the line from each run:


You can see the second session is taking a wider exit line coming across the crest of the hill, which has allowed him to say in the throttle instead up pinching the line which was causing understeer and throttle lift.

Note: This example is with the DL1 software. I don't have MidO data from the same day on the G2X. The software varies slightly with the G2X on how you would look at a similar situation but the tools are all there.

One thing that none of these <$1000 units will do (nor do they claim it can) is let you compare line data from different sessions on different days, due to the GPS drift. Here's data from both of these runs at the pit exit.

You see how the grouping for the 2 days (and the pit in line) do not overlap. This is due to that GPS drift. I have found good results comparing lines within in a given session (as in the example above) but the greater the time diffential between the runs, the more chance for drift. I'll be doing some more to study the GPS drift this week.

 

what about a group buy for this?

 

All you can do is ask. Call Al or Tim, depending on which way your leaning. Low volume stuff like this is tough to give much discount.

 

I found another good example for you all. Here's my all my data from the S2k during the TracQuest event that I instructed for at MidOhio. You'll notice a large variation in lap times as this is from a school so traffic varies quiet a bit through out the session.

It's also interesting to note that the fast time for the weekend (which happened to be with a passenger, no less) does not have a single fast sector time.

Here's 2 of the channels of data from that lap.

Even with just the accelerometers and GPS antenna you can pick up all the shift points that happen when not under braking. To pick up the ones under braking, the rpm pickup becomes valuable.

So if all I had was a lap timer, I'd have to think that the things that I did during that fast lap were key to getting that lap time. However when you have data, you can see items like what I mentioned about there not being a single fast sector time during that lap.

Another thing that I did during that lap was downshift to 2nd in the braking before the turn over the hill into Thunder Valley (sector 6). Without the RPM pickup, I can't see that but what I can see is upshift going down the hill into Thunder Valley (1).

But you can see from the data that it over slows the car, despite it "feeling" faster and giving me that fast lap time. The turn entry was slightly faster (2) but another thing that the downshift to second did was make it easier to slightly slide the rear near the crest of the hill and you can see this slip in the lateral g's (3). It wasn't much of a slip and if I had a steering angle sensor, I doubt you'd see much steering correction. These 2 items combined for a 0.3 second time loss in the sector, so it's better to stay in third for that turn. That speed also carried through for the next sector which was faster as well.

It's a slippery slope but once you have data, you won't want to be with out it.

 

Jeremy,
Please correct me if I am wrong.
so DL1 doesn't come with standard display which costs $600-900 additionally? If that's the case, the choice is obvious to me.
What would be your choice of additional inputs you would consider in DL1?...vid sync would be beneficial as well to learn from data?
I just can't think of much..
thanks.

 

[QUOTE=jonam,Jul 24 2005, 09:16 AM] Jeremy,
Please correct me if I am wrong.
so DL1 doesn't come with standard display

 

I got a question about these GPS gizmos.

I thought the accuracy of GPS based systems was about 10 meters. Which is 30 feet. So if you are racing around the track, 30 feet off line in Turn 9 at Willow either puts you into a wall or into a ditch. How does the software figure out the "driving line?". It seems like there has to be some "margin of error" that could affect this data.

So here's my question:

If I average say 90 mph around the Willow Springs track, that equates to 90 x 5250 = 475200 feet per hour. There are 3600 seconds in an hour, so each second you are traveling 132 feet. So if the GPS is only accurate to 30 feet, that means it could be up to about .25 seconds off the laptime (which is huge), but your unit is producing laptimes that are only .05 off? How does it do that?

-Doug

I'm impressed by how accurate your timing is

 

Don't confuse absolute accuracy with relative accuracy.

Absolute accuracy is: mark the apex of T1. Come back next year, and go to those exact coordinates. How close are you?

Relative accuracy is: mark the apex of T1. Walk a meter north. Check the new coordinates. Are they one meter off from the old set?

GPS does "drift" with atmospheric conditions. The accuracy also depends on the location of the satelites (too close together and they can't triangulate well, too close to the horizon and the signal has to push through a lot more atmosphere). But usually those have time scales of hours, not minutes. So from one lap to the next is should be quite accurate.

I suspect the software also has some error-checking functions that look for bad data.

 

I've seen a detailed explanation somewhere about the lap times. I need to find it again. As Tim mentioned, in actual use, they are seeing a max of 0.05 secs variation versus beacon based systems. That's good enough for me.

It's important to note that neither company is claiming detailed line analysis capabilty and that they both quote an accuracy of 3meters. However, in actual practice, there is some "relative" analysis that can be done.

 

One of the confusing areas when discussing GPS is exactly what Doug asked, which was a good question and was answered well by Mike....Of course I can't give away all our engineering secrets, can I :-)

Distance is related to velocity as much as it is positional. As Mike noted, if you move the vehicle to a given point over a period of days or years, the three meters number comes into play. If you were to move and return the vehicle to a given point over a number of hours, the accuracy drops well below the meter range.

The newer GPS boards we are utilizing are much more sensitive to satellite signal, than what we were using just a year or two previous. They acquire the satellites incredible quick, usually under 30 seconds upon cold start and just a matter of seconds thereafter. Also, they are able to locate and lock upon more satellites. I usually see an extra two or three satellites, compared to old boards.

JL is right on with his data analysis review. It's like I sometimes joke with those interested in our system...I tell them they should just tape over all the gauges on their street vehicle and guess how fast they are going, until the police are nice enough to use their data acquisition (radar gun) to let them know. The human body as a data sensor is good...you can feel minute changes to the vehicle..but the human body as a data recorder is not so good. It is tough to remember everything in a 1.30 lap and many things you think happen, does not.

Concerning the addition of sensors. The G2X does have one dedicated port for the input of a single digital signal (wheel speed). However, there is an additional port that utilizes our Vnet canbus technology. This port can accept the input from any combination of over 25 of our digital and/or analog Vnet sensors. For example, we have developed a steering/throttle/brake package. This consists of a small 3"x4" Vnet Quad box. The box has a cable routed to each sensor. Each cable can be shortened by the user, for their particular application. However, there is only one Vnet cable that returns from the box to the G2X logger. All of the data from the three sensors is returned via this Vnet cable. Other sensors would also share this common cable and could be connected in any order, in any location throughout the vehicle. This works much like a power strip. You can plug your drill, saw, etc in any order on the powerstrip. However, there is only one cable to the wall outlet.
In addition, we can interface with four aftermarket EFI units, via this Vnet input..with the data appearing in our software just the same as our sensor.

We are trying our best to get our demo software and manuals up on our website..along with individual item pricing, better breakdown of how all this works, photos, etc. If anyone would like a demo run CD and manual, please let me know.

Thanks