By Kim Crane
Development
Engineer General Motors
Desert Proving Ground
Mesa, Ariz.

Joe Kopp
Vice President Engineering
Validyne
Engineering Corp. Inc.
Northridge,
Calif.

Edited by Sherri Koucky

Three-hundred channels of vehicle-sensor inputs are   configured into a portable, flexible, and easy-to-install unit. The thermal-test   system has configurable software to custom-tailor data displays.

Three-hundred channels of vehicle-sensor inputs are configured into a portable, flexible, and easy-to-install unit. The thermal-test system has configurable software to custom-tailor data displays.


Over 294 STI chassis and 2,000 cards are spread out   over GM tech centers worldwide.

Over 294 STI chassis and 2,000 cards are spread out over GM tech centers worldwide.


No thermal-test system on the market met the specifications required by General Motors, so a team of engineers designed a configurable one to meet almost any requirements. It is the only system of its type in existence and it all started from two basic problems: how to design a thermocouple input that would remain accurate over a wide range of changing temperatures, and how to consolidate the capabilities of almost a dozen independent test systems into one portable unit.

THE SPECS
The Desert Proving Ground in Mesa, Ariz., is a hub for conceptualization, design, development, and implementation of custom-measurement instrumentation for General Motors. Engineers there need equipment that can test the capability, durability, and safety of new vehicle systems. The tests are rigorous and run as long as necessary to prove the viability of all new vehicle features. What's more, they include systems that are mechanical, electronic, chemical, optical, and hydraulic.

Although the Proving Ground covers a full gamut of testing, the thermal-test system had to tolerate extreme high and low temperatures. To ensure expandability for future needs, the device should also be able to test other types of signals as well. It would incorporate features that are not only new to GM, but to the entire field of automotive test equipment.

Thermal validation of vehicle design is no small ticket because these tests go beyond the general performance of the vehicle system. They define limits, breaking points so to speak, where extreme temperatures and high-loading cause undesirable situations.

The tests vary. Some are as short as 10 or 15 min, while others span 8 hr. A few new-vehicle tests run several years. Typical tests include monitoring dozens of temperatures at points within the vehicle heating, ventilation, and air-conditioning (HVAC) systems. These can be under-the-dash and in-cab temperatures (influenced by the HVAC output), locations on the powertrain, and many points underbody and underhood such as batteries, alternators, catalytic converters, and exhaust pipes.

The test system must be able to endure extreme temperatures and rough use, and remain accurate. Accuracy and reliability are critical because new vehicle development can present unknowns or unexpected results to development engineers. Therefore, all data measured by the test system must not be opinionated but, rather highly reliable so quick and effective solutions can be implemented.

The first step in developing the system, designated as the "Standard Thermal Instrument" (STI), was to ask the question, "What is possible?" The STI had to be rugged, run on little power, and include many channels for inputs from sensors for accurate temperature, analog voltage, strain, and frequency. Another requirement was to create a way the system could interface directly with the internal vehicle network: the J1850 bus.

The temperature range of the system has to withstand ambients of –40 to 85°C, and thermocouple measurements, indeed, all sensor measurements, had to remain accurate across all gradient temperatures. A specification was created from the required features.

THE DELIVERED SYSTEM
It was apparent after first design iteration that the specified device and what can actually be designed are different animals. When theories are tested with real hardware, limitations appear that require modifications.

Design challenges that quickly arose included counteracting the thermal shock on the measurement system's electronics, as well as how to maintain accuracy while enduring radical temperature gradients. Most of this centers around keeping the thermocouple measurements accurate for the environment they work in. This aspect alone took nearly two years to completely develop and debug because of extreme cold and heat, as well as large temperature gradients.

Another requirement that created challenging design issues was 1,000-V galvanic isolation on all channels and inputs. This was necessary to address problems specific to mobile automotive tests such as high voltages generated by various vehicle components. For example, electric-car components can exceed 300 V. Other issues faced by mobile automotive testing include eliminating ground loops from exponentially large-interaction possibilities of a mixed array of 300 floating and grounded thermocouples, large electric dump-noise spikes, and magnetic/electrostatic coupling. These create problems because most electronic parts do not operate anywhere near that voltage level. Finally, the STI was also tasked with providing a full-test interface with the vehicle's system bus to synchronously sample network data with other system inputs.

Three-hundred channels of vehicle-sensor inputs had to be configured into a design that was portable, flexible, easy to install, and provided configurable software and computer interface to custom-tailor data displays.

Early on, it became apparent that the best approach was to take the initial concept and break it into different types of signal conditioning: frequency, thermocouple, voltage, and strain gage, then create a module to handle each type of sensor. Voltage and thermocouple inputs were combined into one card. This adds flexibility in channel assignments which maximize the number of thermocouple channels available and eliminate supporting separate voltage cards. Because of this, the STI can be configured based on a vehicle's specific needs. Different cards are mixed and matched for any test situation.

The 300-channel requirement was satisfied by building a single chassis that holds 15 cards or modules, each with 20 channels per slot. The thermocouple cards have subminiature thermocouple connectors. Other cards use 25-pin D or 37-pin D connectors. Each individual card has its own connector. Conversely, some data-acquisition connections use plug-in modules, but most have all rear connections. The STI system connections are all in front, including the system's power, substantially increasing efficiency for installation and diagnostics. This may not seem important, but when installing nearly 300 thermocouples, finding a bad one requires major effort.

To handle hostile environments, STI uses surface-mount technology. It is intrinsically durable, simply by virtue of the way parts are mounted and soldered. Even dimensional parameters, thickness, and construction technique of the circuit cards are factored in and selected for durability.

On the GUI side of the system, a computer creates configurations so engineers can select any piece of data the way it's needed to be sampled or displayed in real-time, or both. All data is in engineering units, avoiding post-processing conversion. The computer is also used to control and display the system's output during actual testing.

Mobile testing makes a computer screen difficult to read, especially in bright sunlight. Readability is crucial for effective testing and safety because operator burden is high in a car pulling a dynamometer and trying to maintain strict load conditions on a crowded test track. The last thing an operator wants is to search for critical information on a hard-to-read display. Most computers are designed for office environments, so it took over a year working with Panasonic Inc. to devise a tough and readable PC that meets these challenges.

The design team found clever solutions to meet the requirement of 1,000-V isolation between channels and input-to-output. Many data-acquisition systems use a multiplexed or scanned system with one A-to-D converter and many inputs. A multiplexed approach presents intractable problems given the system specifications. To avoid them, the STI uses a separate A-to-D and a processor for every channel. This approach allows effective galvanic isolation while meeting other system specifications such as sample data throughput. The trade-off is that the STI uses a bit more power than other designs.

To tap into the vehicle's system bus, one card has to contain the protocol to access vehicle nodes and parameters. This includes functional and physical messaging using all sample formats such as Slot, Flex, Ping, and Periodic. The benefits of tapping into data on the system bus are almost endless because all new vehicles have node processors controlling multiple functions, from rolling windows up and down to controlling fuel mixture and engine rpm. This capability provides a window into the state of vital vehicle parameters, ranging from development iterations to load-point assessments. A variety of different measurement systems were previously used, but each had inherent inefficiencies such as dual installation and file blending. Until now, no single system was able to integrate Class-2 data and external transducers, such as thermocouples, into a single time-synchronized data file.