All of the steering gear parts are high-precision components and the entire assembly has to be leak-free, both internally and externally, to function correctly. A bearing applies a preload to the working interface between the rack-and-pinion teeth that is critical to overall system performance. If the preload is not correct, the rack-and-pinion won’t function correctly.

In fact, part of the assembly procedure is to wear the bearings in by cycling the unit several times before the final test procedures are performed. This ensures that the inevitable burrs and sharp edges of the as-machined components have a chance to wear in and stabilize. Before this part of the assembly procedure was standardized, as many as 30 percent of the assemblies were routinely rejected during final testing, even though only a very small fraction of them had actual defects.

Steering Gear Verification

Numerous tests that relate to actual performance, and also to customer satisfaction, are commonly performed on rack-and-pinion units. While each automaker has its own specifications, the overall industry trend is toward tighter and tighter standards because of the critical nature of the components.

Delphi Automotive Systems Corp. (Troy, MI) assembles and tests rack-and-pinion steering gear units for a variety of customers at manufacturing plants located around the world. Recently, the company began using electrically powered, servo-driven actuators to replace the hydraulic systems traditionally used in power steering test cells. The genesis of this new approach was a function-test machine Delphi had purchased for an offshore application, but never actually used. Delphi wanted the machine modified for a plant in Australia that was producing rack-and-pinion steering gear assemblies for two different customers.

Everything on the machine was hydraulic, with analog devices including load cells, pressure sensors, flow meters and LVDTs, all under the control of a PLC. However, by the time the data were collected, translated and run through the PLC to generate a test profile, it wasn’t possible to tell the difference between actual system performance and noise. So at the end of the day, Delphi engineers still couldn’t tell good assemblies from bad ones.

One of the steering gears Delphi was producing in Australia had a functional test tolerance of only 75 newtons peak-to-peak, a figure never before specified in the industry. To meet the challenge, it was necessary to find a way to accurately measure forces as small as 7.5 newtons and as large as 7,500 newtons using the same equipment.

After examining various load cells, A/D converters and other alternatives, it became apparent that the PLC and hydraulics were not the best tools for that particular application, primarily because the PLC processes data in timed slices that have to be analyzed and synthesized into an output signal very quickly. To keep production moving, however, the existing test cell was modified to meet Delphi’s immediate needs, and the entire 10-station assembly and test system was shipped to Australia in late 2001, with the agreement that a superior solution would be quickly developed.

Pressing a Solution

Once the test cell was in Australia, Delphi decided almost immediately to upgrade it with new technology. In one of the key tests, called a returnability test, the rack is pushed from corner to corner by a hydraulic cylinder to simulate the forces encountered in actual driving. The test simulates what happens if the driver lets go of the wheel during extreme steering maneuvers. The steering gear must return to center in a controlled manner; if it returns too fast, it will overshoot, sending the steering system into oscillation.

Pushing the rack at a constant speed during the entire rack stroke was a major challenge for the hydraulic technology. The challenge was met by replacing the hydraulic cylinder with an electromechanical assembly press (EMAP) built by Promess (Brighton, MI).

The EMAP is an electrically driven CNC press consisting of an encoder-equipped servomotor driving a ballscrew ram. A variety of sensors can be attached to the ram to monitor its movement. In this application, the sensors provide precise control of the rack motion and highly accurate feedback on system performance. A relatively simple mechanical device, the EMAP is robust and able to survive in the production environment with minimal maintenance. It also includes PC-based control hardware with a PLC control interface, MicroSoft Windows-based motion control and monitoring software, and an integrated load cell and preamplifier.

The fact that the EMAP is essentially a packaged system was very important in this application, because Delphi allowed only 18 weeks from the time the order was placed to the start of production. Fortuitously, the press was almost a dimensional drop-in replacement for the cylinder. It was only necessary to relocate the drive motor on the press into an available space on the test cell, and modify several minor components.

The custom software to control the operation was a bit more of a challenge, because the press is normally used in assembly applications, and the output requirements of this test system are different. Promess met the challenge and a retrofit kit for the test cell was shipped to Australia in April 2002. Technicians completed the retrofit on the original test cell and had it back in production in less than 4 days.

At the time, another assembly and test system, destined for a Delphi plant in Mexico, was almost ready for run-off. Based on its experience in Australia, Delphi knew the standard approach was not going to work well enough for the installation in Mexico, and asked for the EMAP to be incorporated into this system as well. This was easily accomplished by building another retrofit kit like the one for the Australian system, which was installed before the system was run off and shipped to the Mexican plant.

The Heart of the System

The EMAP is normally equipped with a load cell to provide force feedback to the control system. In the assembly and test cell for Delphi, a second load cell is installed in the coupling that connects the EMAP to the test apparatus. The second load cell has a much lower measuring range, so built-in overload protection is required to protect the cell from the maximum test forces.

The standard load cell is used during the high-force phases of the test, and the auxiliary load cell provides feedback on low-force tests. This arrangement provides the required test precision at both ends of the force continuum with minimal additional cost.

At the heart of the system is a digital electronic control device called EMAC. Because EMAC is a digital device, rather than analog, the signal processing capabilities are significantly more robust than could be achieved with a PLC. The EMAC is essentially a fully integrated, programmable multiaxis motion controller and monitoring system that can accept both digital and analog signal inputs and provide both dedicated and user-definable outputs.

Feedback Torque Control

Once the rack-and-pinion assembly and test cells were shipped to Delphi plants in Australia and Mexico, a subsequent system became the test bed for the next stage of evolution. A separate electric torque control system was replaced with a pair of motorized torque control (MTC) units, also from Promess.

The MTC consists of a servomotor, a torque transducer and an angular position encoder that provide real-time information on both torque and angle through the entire torque cycle. Mechanical overload stops prevent damage to the torque transducer. The MTC is controlled directly by the EMAC controller, and feeds torque and angle data back, establishing a completely closed-loop system.

The relatively small size of the EMAC frees up considerable real estate in the electrical panels and saves floor space for the customer. But the real long-range benefit is going to be in the flexibility and performance that can be built into the system. This will offer the opportunity to include extensive diagnostics so the test cell can identify the causes of production failures, and even field failures.

In the future, it should also be possible to control the wear-in process more precisely by monitoring the behavior of the rack-and-pinion unit in real time. Instead of just cycling the assembly a specified number of times, it should be possible to monitor the torque trend, and then adjust the rack bearing adjuster plug to achieve the specified operating torque value 100 percent of the time.