Historically, the main function of an oscilloscope was to display waveforms. Today, the digital storage oscilloscope (DSO) is called upon to do far more than just that.
DSOs still display waveforms, but also can give you a processed display which may highlight or suppress aberrations or examine critical signal interactions. DSOs can perform highly customized measurements, process data in the frequency domain, look for and acquire highly definable faults, document, and archive and compress the data in a plethora of ways—all while providing portability, automation, and remote control. Given these capabilities, DSOs lend themselves well to portable data acquisition systems. A recent application involving DSOs as a part of a networked data acquisition system demonstrates these capabilities.
John Deere and Co. in Waterloo, IA, exercises many of the latest DSO capabilities at one time. This facility performs extensive testing of heavy-equipment diesel engines. Several dozen engine test cells are each fitted with modern DSOs that perform the tasks required to meet the company’s high-performance testing standards.
Critical to the testing regime is accuracy of data, test repeatability, and minimal errors. There are many technical requirements for this application:
Capturing the right data—dynamic engine parameters must be captured without fail.
Highly customized measurements—these measurements must be automatically scaled for amplitude (MegaPascals, MPa) and time (degrees) and logged over the duration of the test.
Timing not in time, but in angular displacement—analyzing engine performance requires signal analysis with respect to angular displacement, not time.
Display—the scope must present live data so the operator can, at a glance, verify the proper operation of the engine and sensors.
Documentation—waveforms must be documented as necessary.
Automation—user interaction should be minimal to reduce the possibility for error.
Accessibility—the network must have access to the instrument for setting up standardized tests and the data for review and archiving.
Capturing the Right Data
It is imperative to collect the right data. There are many challenges to accomplishing this requirement. Engine signals are very dynamic. The difficult task is to trigger on a specific point of a cycle in spite of the absence of a single, valid trigger signal.
In Figure 1, the red waveform is a series of timing pulses that represent fixed locations on the crankshaft. The signal is created from pins precisely inserted into the crankshaft and sensed with a magnetic pickup.
There are three timing pulses: -30°, top-dead center, and +60°. Each engine cycle encompasses two crank rotations—firing and exhaust—giving six timing pulses per cycle. The green waveform is from a proximity sensor that monitors the movement of the injector or needle. The blue and yellow waveforms are pressures (injection and cylinder) produced with transducers.
The ideal trigger point for display and acquisition is the top-dead timing pulse, but only when the cylinder is firing. To discriminate that one-in-six pulse, John Deere uses an advanced triggering function called gating. Gating allows a signal, in this case the needle-lift signal, which occurs just prior to the top-dead pulse to be used as an enabler for triggering.
Highly Customized Measurements
It is a valuable asset to have measurements automatically scaled in user-defined units (such as psi in the case of pressure) using an mx + b equation. If you were to ask for the amplitude of a pressure, the unscaled result might be 1.245 V. To interpret that measurement (i.e., how many psi equal 1 V) in a meaningful way, you would have to know the conversion factor of the transducer and any offset that may or may not be present.
A manual calculation then would ensue. A feature such as scaling would require entering the factor and offset one time, then every measurement made on that signal is automatically scaled every time.
In this application, a key timing measurement of interest is an interaction between two signals. It is imperative to know when the needle-lift signal rises with respect to top-dead center.
Again referring to Figure 1, a custom measurement can be defined to locate top-dead center (second zero crossing of the timing pulse). Another measurement can locate the rising of the needle lift, and a third measurement can determine the difference between the two. This difference value then is scaled in degrees.
It is vital to log these custom measurements throughout the test (which could take hours or even days) to the network. These measurements can be read at intervals predefined by the engineer. This data then is correlated with other pertinent data on the network.
The important point is that the signal data is not on the network, but the measurements are. The signal data could be tens of thousands of bytes of data where the measurements are short strings, less than 100 bytes. This saves the network from carrying large data files. The DSO basically is crunching the numbers in a highly customizable fashion.
Timing Not in Time, But in Angular Displacement
User-defined scaling can be a bit more complicated for the timing parameters. DSOs and data acquisition systems sample with respect to time. Timing measurements count samples and are related to time. To scale a timing measurement in a unit other than time, the sample rate must be controlled.
The sample rate can be synced with the engine’s angular displacement in one of three ways. The first method uses software where the signals are oversampled, rpm data is acquired concurrently, and samples are dropped, keeping only the ones that most closely coincide with the measure of angular displacement. Another method mounts an encoder (usually optical) on the shaft that creates a signal to sample × times per rotation. This method requires the mechanical sensor hardware to be maintained in precise order. The final, and often the best, method tracks the engine rotation with an external hardware decoder.
While setting up or running the test, it is vital to know that the engine and the transducers are working properly. The traditional application of an oscilloscope offers this basic functionality. All DSOs used at John Deere have color displays which allow overlap traces without sacrificing signal identification. In fact, the capability to discern important signal interactions is heightened by overlapping color traces, a capability lost on monochrome DSOs.
By syncing the DSO’s sampling with engine rotation, the display also is synced to the engine’s displacement. No time-base adjustments are needed when the engine speeds up or slows down.
Although the measurements are of keen importance, from time to time the waveforms also must be saved. Many DSOs have hard drives that can quickly store the waveforms upon command from the network or the local user.
Usually, access to the raw data is not so important. The bigger need is to simply view and archive the waveforms. A printer, integral or external, can accomplish this task, again without user intervention since the command may periodically come via the network as defined by the engineer setting up the test. Still, the local user can demand a hard copy any time it is appropriate. An integral printer offers the extended functionality of doubling as an oscillograph (real-time, continuous chart recording) for trending of the engine’s pressure signals over time.
The engineers that define what tests are to be performed select the proper DSO setup from a predefined array. This is done remotely—in fact, from another building.
The chosen setup is sent over the network to the instrument, and testing commences. The engineer may tailor certain variables in each instrument setup, like which measurements to perform, the measurement logging rate, and the hard-copy intervals.
The minimal interaction from the cell operator eliminates many human errors, communications errors, or data interpretation factors that may occur otherwise. With this arrangement, testing standardization and repeatability are optimized. Although the DSO is run over the network, the local cell operator may gain control of the instrument at any time.
The measurements and their logging are fully automated. There may be times when the raw data is useful for analysis beyond the visual scrutiny of a printout. The complete data sets of engine parameters can be acquired and shared over the network, either in an automated fashion or manually upon demand. Also, all the engineers have access to the data, live or archived.
The oscilloscope has matured from a tube that displays waveforms to a miniature data acquisition system capable of automatically processing large amounts of data quickly and in a highly customized fashion. Concurrently, there has been a surge of distributed, or networked, data acquisition. As the acquisition systems themselves become more powerful, the processing of data can become distributed as well, just as in this application.
About the Author
Pete Cipriani is the DSO marketing manager at Gould Instrument Systems. He is an electronic engineer with more than 13 years experience with Gould DSOs. Gould Instrument Systems, 8333 Rockside Rd., Valley View, OH 44125-6100, (216) 328-7284.
Hardware Decoders—In Sync With Engine Testing
Figure 2 is a block diagram of an external hardware decoder, the Gould Synchroscope® time-base module. During engine operation, a pick-off signal is acquired, conditioned, and divided if necessary. It then resynchronizes a generator that creates the proper sampling rate to match the engine’s rotation.
This sampling-rate signal is generated dynamically and continuously. It can be resynchronized once per revolution, or up to 1,000 times per revolution, to better track changes in engine velocity at 3,600 samples/revolution, or 0.1° per sample. Once the sampling is synchronized to the engine, timing measurements can easily be scaled to read in degrees. This solution best controls the cost, complexity, and maintenance of the system while offering a high standard of performance.
Copyright 1999 Nelson Publishing Inc.