Automotive Testing Expo took place October 20-22 in Novi, MI. The show combined the auto expo and engine test expo—in halls A and B, but joined so attendees saw both halls all as one big area. Some of the engine test exhibitors had fascinating stuff—micro-finished crankshafts and systems that applied a special anti-wear coating to the insides of the cylinders. Some people had special casting capabilities. Great products, but not closely related to test.
On Tuesday morning, both Yokogawa and Newton’s 4th caught my eye because of the power-measuring instruments they had on display. A power-measuring instrument is the main product of Newton’s 4th, but Yokogawa also was showing its line of ScopeCorders, although in a less prominent position.
Also demonstrating power analysis was Keysight Technologies, which featured its IntegraVision power analyzers, which are suitable for automotive engineers who want to quickly and interactively measure power consumption, power conversion efficiency, and operational response to stimulus. The power analyzers provide an intuitive combination of accurate power measurements and touch-driven oscilloscope visualization.
Newton’s 4th will be introducing a new machine soon, but it wasn’t quite ready for the show. According to the company’s representative, it will have an attractive combination of price and performance. One aspect he stressed was the capability to lock onto a signal within half a cycle—apparently something that the competition can’t do. Without the signal-locking capability, your readings would remain inaccurate if the source continues changing speed and the instrument never completely locks to it. Noisy PWM signals can be hard to measure for this reason. The Newton’s 4th architecture is designed to overcome this problem.
Later, I met a rep at ZES Zimmer, a German power-analyzer company that claims to have the edge on all other companies. Another problem with power analyzers is aliasing. Like any digital instrument, they digitize the input, and that means aliasing can occur. To avoid aliasing, other manufacturers use an input filter that guarantees the sampling rate is sufficient for the bandwidth. Great, but the filter destroys any high-frequency content, which is a large problem with fast switching PWM signals that do have lots of HF content.
ZES Zimmer runs two parallel acquisition systems, one filtered and the other not. So, after you have acquired your signal, you can compare the two outputs to see if they are similar—no aliasing—or sufficiently different that aliasing obviously has occurred. The objective is to avoid chopping off bandwidth on the suspicion that HF power might exist. If it doesn’t, you may have overly reduced the B/W such that the result accuracy suffers.
Similarly, the Data Physics 900 Series Dynamic Signal Analyzers with SignalCalc 900 Series software allows you to simultaneously measure data from the same channel at different sample rates and frequency resolutions.
Structured data sets
Mike Albright at Signal.X described the software he has developed for big structured data sets. Structured is the key word. This means that the data is of the same kind and can have simple tests applied such as to remove outliers. Also, the software helps develop metrics that allow you to filter the data to create a new data set having only relevant elements. There’s not a lot new in this, other than working with such large data sets; however, what is new is the way companies are using data.
Rather than preparing a PowerPoint presentation of the results of some data-mining exercise, people today are doing that as only an example of what can be done. Ideally, they want to share the large data sets among other researchers who can establish their own metrics and extract other details from the sets. It’s not just a bandwidth and storage problem, but also one of having the flexibility to access the data in many ways not necessarily originally intended when it was acquired.
The latest technology from Dewetron is an iPad-like user interface built into a Dewetron display. You can stretch/shrink waveforms to suit and save test setups so the results are reproducible.
Jon Milbrandt and Gary Schneider at Hi Techniques emphasized rugged small DAQ systems specifically developed for automotive/aero applications. Schneider said a recent application involved strapping one of the waterproof systems to a jet ski for testing.
Spectral Dynamics was sharing its stand with MSI-DFA (Maryland Sound International-Direct Field Acoustic Test), a company that specializes in high-power acoustic test. An engineer from the company described work performed on satellite acoustic testing. Apparently, the complex sound developed during blastoff can excite more resonant modes in the satellite than get tested by conventional vibration test. So, they use stacks of amplifiers and speakers to develop a 1-MW sound field and measure the mechanical responses.
Astro-Med has changed its name to Astro-Nova—representing a rebranding after the company sold its medical business. Astro-Nova has not been bought but is just dropping the medical reference.
Wineman Technology, a frequent company frequently at NIWeek, was displaying some of its custom test systems.
Rohde & Schwarz has developed a radar target simulator box. When all cars are autonomous, everyone will need radar to sense what is happening around it. Contrary to a simplistic view of a pulse and a return, there can be several simultaneous targets that each returns a slightly different signal.
The instrument R&S has developed receives the outgoing pulse from the radar, down-converts it, manipulates it by delaying it, adding copies each with their own delays and losses, and so on, making up a compound IF signal representing possible target situations. The combined signal is up-converted and returned to the radar antenna. This approach is limited at least in part by the finite time taken to process the data—about 220 ns in the present version, although reductions are planned for future versions. This limits how close the simulated target can be to the radar.
Nevertheless, R&S is involved with Asian, European, and American customers to try to determine how best to evolve the instrument. At present, it can solve lots of problems, but probably raises many to the extent that there are an almost infinite number of possible target scenarios. In fact, generating lists of standard scenarios is one direction the project may go—similar to some wireless standards.
I attended two of the 20-minute technical sessions on the conference floor—one from Newton’s 4th about the factors that contribute to a power analyzer’s accuracy or lack of it, and one by Mike Hoyer from HBM about what the company calls “eDrive” data acquisition and power analysis. The basic idea is to acquire all the signals at the same time. By synchronously acquiring the input voltage and current, the frequency inverter output I and V, and the motor speed and torque, you have the raw data that you can slice and dice as required for subsequent analysis. And, you can correlate one action with another—when the brake was applied, the speed slowed down, and the motor current decreased. The effects are correctly related in time.
As the Newton’s 4th presentation stressed, accurately measuring dynamic power can be difficult. Factors include the necessity to use a discrete Fourier transform, not the FFT, because the DFT output includes phase. This is important because you need to account for the phase difference between V and I signals that are meant to be contributing power. And, you need a robust calibration procedure to ensure a flat frequency response across the entire analysis bandwidth.
Electrons aren’t the only things that need tested. People also make physical things that require environmental testing of all sorts—temperature, shake and bake, squeak and rattle, humidity, and so on. As might be expected at an automotive testing show, many companies were there that provide these kinds of test capabilities. A talk by a Volvo car engineer about squeak and rattle testing included a matrix showing the likely interaction when dissimilar materials were in contact.
And, on the shaker side, Vibration Research has developed a so-called instant degrees of freedom (iDOF) process for fast fixture mapping. So, before you dial up your latest custom-Kurtosis test, you can quickly run the iDOF process to determine the effects being added by the fixture that are not coming from your DUT.
Finally, with the continued emphasis on simulation, which is extensively used by all car companies, you might wonder if test expos have become totally passé. Can’t you just find somebody with a Cray installation, run some programs overnight, and build the prototypes the next morning? Why do companies still engage in extensive testing?
An editorial in the show issue of Automotive Testing Technology International by industry veteran Gene Lukianov summed up the situation. He said, “Mathematical analysis and simulations, while extremely powerful, are inherently limited by the assumptions made and the capability of the analysis to deliver narrowly focused results that need to be interpreted. Simulations don’t tell you what you did not ask: testing does.”