Fundamentals of Data Acquisition - part 5
Data Acquisition Software: What You Need
INTRODUCTION: The most important feature in data acquisition software is comfort. The user must be comfortable with the software. Awkward software is like a pebble in your shoe. It may not slow you up, but it makes you think about the pebble every step instead of allowing you to concentrate on where you are going.
Awkward software makes you think about software rather than analysis. Good software allows you to concentrate on analysis while using the software as a familiar tool. Bad software forces you to think how to get the software to display what you need to see rather than concentrating on the analysis. It is the same as a balky shift in a race car. Shifting gears should be an automatic activity allowing the driver to concentrate on driving the car. A balky shift forces the driver to think about the transmission and distracts him from thoughts of apex and passing.
The first requirement in comfortable software is familiarity. Any piece of software will feel awkward while you are learning. You must devote a certain amount of time to learning a new piece of software before deciding whether or not it is comfortable, just as a new pair of shoes must be broken in. Do not expect to do this at the track. Spend several hours working with your software before forming an opinion. Get some data and study it as if it were your data. Try to see what you can learn from the demo data and if the software helps you in your analysis.
I know people who use PI, Edge, CDS and Corsa software. Most of them swear by the software they use, even if they sometimes swear at it. This is because they are familiar with that software. I prefer my Debrief2 software because I have been using it almost daily for the last eight years. I know the program very well and am VERY familiar with it. In addition, when I discover something new I want to do, I change the program. You will not have this advantage. My disadvantage is that I become accustomed to awkward portions of the software. It is not until someone complains about having to use too many keystrokes, or I have to document it, that I make it simpler. I am investing a considerable amount of time in this because the GAP (Graphical Analysis Package) will be used in many of my new programs.
If you invest several hours with a program and still are frustrated, then the problem may be the software. If the software does not act as documented, then you have a serious problem.
CHEAT SHEETS: The easiest way to learn a program is with a cheat sheet. This is one or two pieces of paper, preferably laminated, which give you keys to the use of the program. You may know what you want to do - the cheat sheet lets you quickly find out which menu you need. To move the cursor to the minimum speed point in Turn Three, should you use the Cursor menu (Alt-C) or the Pointer menu (Alt-P) or the Turn menu (Alt-T), or the mouse? The cheat sheet answers this question AFTER you have learned some of the basics and UNTIL you are totally familiar with the program. But I still use the cheat sheet for certain obscure commands.
WHAT YOU NEED: You should be able to display any variable over any interval on any portion of the screen and with any color. This allows you to arrange the data in any form you desire so as to highlight any features or relationships you are seeking. The following graph has Speed in the Top Half, Engine RPM in the third quarter, and Long G in the 4th quarter. There is also a track map in the upper right hand corner and a turn map alongside.
This plot has Distance as the X-axis. Graphing by Distance is best for comparing two laps of different times because it matches points on the track.
Above the graph is a track map and a turn map. The track map, in the upper-left hand corner, displays the entire lap of Mid-Ohio. A track map is very important to identify individual turns in the data. Without a map you can waste a lot of time trying to match points on the distance graph with turns. The turn map to the right of the track map displays a portion of the track in a larger scale. In this case it identifies the bend in the Mid-Ohio "straight". In an actual turn the turn map will clearly identify the entrance, apex and exit of the turn.
The cursor (vertical line on the distance graph) is the same point as the cursors (square boxes) on the two maps. As you move the cursor with the arrow keys or the mouse the map cursors will move. You can also click the mouse on the map to select a point in the data.
The above plot also displays numeric values for the variables that are NOT graphed. This allows you to check another value, such as Lateral G, without adding that variable to the graph. This is a very useful.
The Data menu (Alt-D) allows you to select the variables you want to display on the plot.
The Location menu (Alt-L) allows you to choose where a variable will be displayed. You can choose the entire graph area, the top or bottom half, the upper, middle or lower third, and any of four quarters or five fifths.
The Color menu (Alt-C) allows you to choose the color to display any variable.
(*** Need graphic of Data menu ***)
(*** Need graphic of Color menu ***)
(*** Need graphic of Location menu ***)
You can auto-scale a variable or select specific minimum and maximum values to plot. If you are displaying speed for an IndyCar on a big oval you may want to show 220-250 mph. For a street course you may want to show 40-150 mph. If you are looking for the effect of a rev limiter at the end of the fastest straight you may want to display 140-150 mph while showing RPM from 7000-8500.
(*** Need small range graphic ***)
This graph shows detail of back straight speed and rpm. Both variables are scaled to show greater detail.
The View is also zoomed to show only the back straight rather than the entire lap. This permits detailed analysis of the straight.
Long G shows how acceleration declines as speed increases. The rate of RPM increase also declines.
You should also be able to choose the time interval you are displaying. You may want to begin with a complete lap and then zero-in on a specific turn. You should also be able to display an entire run consisting of several laps. This is useful for looking for over-revs, excess temperatures, or low pressures.
You must be able to graph data by time or distance or any other variable. A graph by time is useful for comparing shift times. A graph by distance is necessary for comparing two different laps. Graphing by distance compares two laps at the same point on the track. Graphing by other variables, such as Speed or Lateral G, allows better analysis. A plot of Long G by Lateral G shows how the driver is using the traction circle.
The ability to smooth data permits you to eliminate noise and concentrate on trends. However, you must beware of software which smoothes data without telling you. This can hide important details you may want to look at.
(*** Need smooth graphic here ***)
The graph at left is a smoothed version of the Speed variable from the previous page. Some noise from the production speed sensor has been eliminated and the overall trend is evident.
The trick is knowing when to smooth data and when to examine the detail.
There is often useful knowledge in the details.
Other useful features include statistical analysis. The capability to compute and display the mean and standard deviation of any variable over any interval can save you a lot of calculator time.
The graph at left presents statistical analysis of Speed over an interval of the back straight. It shows that speed ranges from 71.118 to 99.617 with an average of 88.283 mph. The analysis also shows that Speed increases 1.961 mph per second or 0.015 mph per foot, or 15 mph per 1000 feet.
Below is a Histogram plot. It shows how much time is spent at each value.
Histrograms are very important for shock velocity analysis (shock tuning).
Engine tuners also like histograms
of engine RPM. This tells them what
sort of torque curve they need.
Other important features include the ability to compute Math Channels. This means combining data channels via some formula that produces interesting information. For instance, Normal Steer can be computed by multiplying Speed times Steering. This produces a math channel that can be compared to Lateral G.
Links to Tech Reports
Links to part four Fundamentals - part 4