Realistic off-vehicle testing of transmission components

A new automatic transmission test rig with controls developed by COM Inc. (Dexter, Mich.) uses two dynamometers-one to drive the component under test and another simulating real-world load conditions.

By Jim Montague, news editor November 1, 2001

To achieve realistic, off-vehicle testing of transmission components, COM’s two dynamometers drive the component under test and simulate real-world load conditions

T esting automatic transmission components is challenging because it is costly to devise a method for instrumenting critical components such as torque converters in an operating transmission. On-vehicle testing can determine component life after the transmission is taken apart, but provides little or no information on critical operating parameters such as the torques, hydraulic pressures and temperatures that they see.

A new test rig with controls developed by COM Inc. (Dexter, Mich.) uses two dynamometers-one to drive the component under test and another simulating real-world load conditions. The test unit is instrumented with strain gauges and thermocouples that, when connected to a telemetry collar, provide real-time information from the rotating part. This allows engineers to control the slip in torque converter clutches, evaluate different component designs, and immediately see how performance varies.

A key factor in making this system affordable is its use of Microstar Laboratories’ (Bellevue, Wa.) data acquisition processor, which runs real-time operations independent of the non-real-time graphical user interface (GUI). This approach increases reliability, and-using COM’s Finite State Machine software-saves the $20,000 to $30,000 that would otherwise be required for a programmable logic controller (PLC) system to provide safety monitoring and machine control.

Microstar’s Universal Test Controller (UTC) data acquisition processor runs real-time operations independent of the non-real-time graphical user interface.

Present testing methods

Currently, there are two basic methods of testing critical automatic transmission components, such as torque converters. Both leave something to be desired. The first method is simply running full vehicle tests on either a dynamometer or proving ground. This approach provides valuable information in evaluating the durability of the overall design. However, it is a relatively poor tool for component-level design because of lead-time and the fact that it’s very costly to provide the instrumentation needed to measure operating parameters on individual components.

The other approach is to test transmission components in a laboratory, normally by driving one side of the component, while the other is fixed to the machine. The dynamometer typically is cycled through a series of loads. The problem here is inability to correlate the loads seen by the component with a real world vehicle. While general information can be obtained, such as the ability of different friction materials to withstand certain operating temperatures, laboratory testing has limited ability to evaluate the relative performance of alternate design concepts.

Constant slip test rig design

To overcome these testing limits, an automobile manufacturer asked COM to develop a constant slip test system that would accurately simulate real-world transmission conditions. The basic idea was to use two dynamometers, the first simulating the automobile driving the transmission, and the second simulating the resistance of the driveline, tires and road.

To recreate actual conditions seen by the component, it had to be instrumented, so that its operating conditions could be clearly measured under various scenarios. COM’s solution involved instrumenting components attached to both dynamometers with sensors; then using telemetry collars to transmit these signals to a data acquisition device. Sensors used in the test rig measure rpm, torque, strain, temperature, pressure and other parameters. This makes it possible to perform measurements that haven’t been accomplished before, such as measuring the amount of torque that causes a clutch to break away and begin slipping under a real-world control profile. This method also addresses a major need of automotive OEMs to be able to change designs, such as using a new friction material or a different plate geometry, and immediately measure the change’s impact on the operation of the torque converter.

While there were several challenges to designing the test rig, control of the two dynamometers was perhaps the most critical. Dynamometer inputs are based on measurements taken in road tests. Certain measurements can be easily taken on an automatic transmission, such as the rpm and torque going into and coming out of the transmission. It’s also possible to measure the pressure and temperature of the hydraulic fluid used to operate the torque converter. And, the torque and rpm applied to the input of the transmission are used to determine the control signal for the driving dynamometer. However, determining the signal for the second dynamometer is more complicated. A mathematical model developed by the automotive OEM is used to transform the road testing data to determine the operating signal that will duplicate these parameters. The automotive OEM wanted an easy-to-use GUI to operate the test rig.

However, aware that Microsoft Windows is not a real-time operating system, the OEM needed some other method to provide safety and limit checking on the rig to ensure operator safety and avoid damage to equipment and components.

An alternative to PLC control

Instead of the traditional OEM approach based on a PLC, a digital controller used for applications such as on/off control, timing, logic, counting and sequencing, between the graphical interface and the machine.

Operating systems like Windows NT use many cycles on the PC platform, and when they take control of the CPU, they hold onto it for a long time in machine control terms. A failure of the operating system could leave the machine in an unknown and possibly dangerous state. When measuring or controlling a real-time process, computing power has to be available when needed. However, incorporating both a PLC and a personal computer would have driven up the machine’s cost.

COM’s engineers searched for an alternative that would keep the machine affordable. They had previously used data acquisition processors (DAPs) from Microstar Laboratories in several other projects, including dynamometer control systems for full transmission testing.

The capability offered by DAPs is an onboard microprocessor, which runs a multitasking, real-time operating system optimized for high-performance, real-time data acquisition and control applications. DAPL, the onboard operating system used in DAPs, simplifies communications by providing more than 100 commands optimized for data acquisition and machine control. This intelligence on the DAP board is designed to extend the power of the Windows user interface by executing all processor intensive routines in real time and performing data reduction, so that the software on the PC can handle more demanding applications than usual. Onboard intelligence tipped the balance in favor of choosing a DAP for this project.

Finite state for real-time control

Next, COM’s engineering team configured the DAP to provide a software control loop for the test rig. The Windows interface provides the basic dynamic commands to the rig. Each of these commands goes through a finite state machine developed in DAPL by COM’s engineers. The DAP communicates with Windows through a watchdog interface. Should Windows fail to respond to the watchdog timer, the finite state machine takes over and brings the test rig to a safe stop.

Actions of the finite state machine are determined by a simple ASCII script file, which performs safety and limit checking in a manner similar to a PLC, but at a much lower cost. Though upfront development costs reduced savings on the initial machine, John Ritter, COM’s system development vp, says that now that the finite state machine has been developed, use of the DAP to replace PLCs on future test rigs will save $20,000 to $30,000 per unit.

As a result, the automotive OEM’s engineers will for the first time have the ability to tweak the design of torque converters and other components and immediately determine the effect of their changes on key operating parameters, such as breakaway torque, temperature, strain, and others under real-world operating conditions. For example, the component could be instrumented with temperature probes to measure previously undetectable hot spots that could cause premature failure.

‘One of the keys to the success of this project was our ability to add a real-time processor to a PC in order to keep the mission-critical functions in a separate processor from the GUI,’ says Mr. Ritter. ‘The use of the Micostar DAP and iDSC 1816 modules have been so successful that we plan to use this approach as a basis for our Universal Test Controller (UTC) Product.’

Jim Montague, news editor jmontague@cahners.com