No one demands more out of a car than Michael Andretti of Newman Haas Racing (NHR). He has more victories and has led more laps than any active Champ Car driver in history. With that reputation to uphold, it’s not surprising that his quest for the perfect car setup is relentless. Just when team engineers think they have found that magic combination of camber, toe, springs, and aerodynamics, Michael will want still more grip of the car to the track.
An essential part of tuning a race car is repeatable car preparation. Each test run is a carefully controlled experiment. Car setups must be quantifiable and reproducible. Any magic must be confined to what the driver does on the course, not to what the engineers and mechanics do in the pits. To this end, NHR has a rigorous component testing program that uses a number of test and measurement systems.
Increasing competitive demands have made open-wheel race-car suspension systems more complex. The dynamic platform on which the vehicle rests must satisfy many competing objectives, including vehicle attitude and height control, weight distribution, weight transfer distribution, and tire force variation or grip. These fundamental parameters have follow-on effects that ultimately contribute to the overall acceleration performance of a vehicle around a closed circuit.
NHR and other Championship Auto Racing Teams (CART) purchase their vehicle chassis and powertrains. Then, the individual teams must perform the mechanical integration and vehicle dynamics optimization (setup) needed to create a high-performance racing machine.
While teams make a variety of performance-enhancing modifications, car setup primarily is concerned with suspension design. Generally, a race engineer seeks to achieve the best total tire adhesion possible under the constraints of an imperfect road profile and the necessary turning maneuvers.
Primary springs play an integral role in the system (Figure 1, see July 2000 issue). They must be stiff enough to prevent the chassis from touching the ground, but soft enough to absorb the shock of road irregularities. They largely determine the wheel rate, roll rate, and load transfer responses to acceleration, braking, and turning forces. Springs also can affect static cross-weight or the distribution of load between the two diagonals of the car.
Cars are four-legged beasts that allow weight to be unevenly distributed around the four corners. Cross-weight is used to bias the car in anticipation of cornering load transfers. This is particularly useful on ovals where all of the turns are the same direction and about the same radius. The key objective is to find the distribution of tire normal forces that will maximize the lateral forces produced at the tire contact patch in response to a cornering force.
Determining the best spring is a matter of finding the right balance among ride rate, ride height, roll rate, and other factors. Generally, it is desirable to leave cross-weight alone when making spring changes. This leaves the load biasing decoupled from the spring-driven handling metrics. Then, a race engineer is free to adjust either variable independently.
The only way to make a spring change without affecting cross-weight is to bias the strut length using a matching spacer at each corner. That is precisely what teams do. When a spring change is made on a car with a given load distribution, all four of the loaded strut lengths (compressed spring + shim) remain the same. If one of the lengths were to change, weight would be jacked to another corner.
Creating a matching spacer for each spring in a race team’s inventory is a big job. For example, there are more than 600 springs to cover all five cars at NHR. To determine these spacer lengths, the team traditionally relied on manual measurements using a press, load cell, and calipers. The inevitable problems of a manual process resulted in spacers that were hardly ever the right dimension.
This forced the team to pre-fit springs during the evenings of race weekends. Pre-fitting involves installing the springs and shims and adjusting the shim dimensions while the car is on the scales. This proved to be a thoroughly unpopular activity with the mechanics.
Shortly before the final race of the 1999 season, NHR decided to build an automatic spring tester that would enable the crew to prepare spacers in advance of the race. The design objectives set for the machine were fourfold:
- • Minimize measurement error.
- • Perform faster and more frequent testing.
- • Characterize linear and nonlinear springs.
- • Conduct unattended endurance testing for devices whose behavior changes with use.
Figure 2 illustrates the system block diagram. Suspensions are made up of precisely defined component dimensions and stiffnesses. The spring tester presented an opportunity to characterize a range of devices besides primary springs. Bump rubbers, push-rod load cells, secondary springs, and third springs needed to be accommodated.
Bump rubbers exhibit nonlinear, progressive characteristics that vary with use and temperature. While coil springs may only need a few data points to characterize, bump rubbers require many across the entire range of deflection. To this end, a multichannel, PC-based data acquisition board was selected for load and length measurement. Figure 3 shows the completed system.
The tester uses a pneumatic cylinder for force generation and a large-capacity load cell for force measurement. Length is sensed by a linear transducer located rearward of the spring under test. Springs with a free length of up to 9 inches can be tested. The length transducer can resolve to 0.001 inch.
In previous NHR data-acquisition projects, machine software was developed using PC-based, high-level language development environments like Visual C++. This time, NHR chose The MathWorks’ MATLAB® as the application platform.
While MATLAB is an easy-to-use, flexible computational tool, the challenge has always been to get raw data into the workspace. Generally, test data needed to be converted into MATLAB data files. The MATLAB Data Acquisition Toolbox solved this problem by providing a seamless interface to a number of third-party data acquisition boards.
The toolbox allows live data to be streamed into MATLAB for run-time analysis and visualization. The data acquisition board setup and data transfer are accomplished using MATLAB objects. The toolbox is completely integrated with MATLAB, so data collection and processing are done within one environment, using the MATLAB language. This approach favors applications that require significant post-processing and custom visualization.
Like most test and measurement applications, the machine software controls the test sequence, captures the data, and after post-processing, delivers the results. Since a technician operates the machine, a graphical user interface (GUI) was developed to facilitate running the same test and analysis on multiple springs.
Custom interfaces were designed to automate the entire spring-tester process. Each spring test is archived into a relational database for recall and reporting. Finally, custom graphics are used to visualize each result.
Software Architecture Setup
Setting up a spring run requires a chain of GUIs that identify the spring, spring type, and testing parameters. Next, the data acquisition board is configured for acquisition by using the analog-input, analog-output, and digital I/O objects in the toolbox.
Test Sequence Control
The toolbox has a number of options for customizing acquisition. The simplest is to acquire a fixed number of samples. Time and channel value-based triggering can be added to acquire data before and after an event. In this application, a load cell-driven state machine determines the acquisition period. This is accomplished by setting the samples-per-trigger property to infinity to continuously acquire data.
While acquiring the data, the input to the state machine is fed by using the peekdata function. The state machine doesn’t need to run in real time; it just needs to ensure that the load range criterion for the test is met. The length and load data that results from the acquisition is acquired in real time and simultaneously transferred into MATLAB for processing.
Post-Processing
You can create custom plots to view and analyze the output of the system. A sample result from the spring-tester machine is shown in Figure 4.
The spring-tester machine has improved NHR’s car preparation process. Team engineers and mechanics no longer have to worry that a spring change will alter the balance of the car. The crew is free to focus on other pressing performance and reliability issues.
About the Race Team
Newman Haas Racing was formed in 1983 as a partnership between racing entrepreneur Carl Haas and actor Paul Newman. The team has won three championships, 54 race victories, and 60 pole positions.
About the Author
Michael Hegel has been an R&D engineer at Newman Haas Racing since 1998, developing embedded systems, data acquisition systems, and modeling tools. Prior to joining NHR, he was at the Ford Motor Company, working on Formula 1 projects with Benetton (active suspension), Sauber (driver displays), and Stewart Grand Prix (drive-by-wire). Mr. Hegel received a B.S.C.P.E. and an M.S.E.E. from Michigan State University. Newman Haas Racing, 500 Tower Parkway, Lincolnshire, IL 60069, e-mail: [email protected].
Published by EE-Evaluation Engineering
All contents © 2000 Nelson Publishing Inc.
No reprint, distribution, or reuse in any medium is permitted
without the express written consent of the publisher.
July 2000