By Jennifer Haupt and Lynn Parker
What if defense contractors facing government budget cuts could replace $2 million of customized hardware and software with off-the-shelf equipment costing one-fifth the price? That is exactly what engineers from Northrop Grumman Electronic Sensors and Systems Division (formerly Westinghouse Electric) in Baltimore did when they built the radar systems for the U.S. Army`s New AH-64D Longbow Apache helicopter.
"If we had built a standard automatic testing bench, it would have cost in excess of $2 million, instead of one-fifth that for the new logic analyzer-based system. Plus, there was no reduction in the effectiveness of the testing with the less expensive set-up," says Hank Vitallo, a senior field engineer for Northrop Grumman. "You can`t argue with those results."
The Longbow fire control radar and the radar-guided Longbow missile will go on nearly a third of the upgraded fleet. The Northrop Grumman engineers put the radar system through a series of government qualification tests and electromagnetic interference (EMI) tests.
The Longbow fire control is a millimeter wave radar system to detect, classify, rank, and track ground and airborne targets. The antenna, transmitter, and receiver are located in a mushroom-shaped mast-mounted assembly perched above the Apache`s rotor blades for improved visibility and survivability. Located in the lower fuselage of the aircraft are the programmable signal processor and low-power radio frequency electronics
The Longbow system provides Army pilots, for the first time, with a "fire and forget" capability. This enables the pilots and their aircraft to remain safely masked behind hills or trees except for brief periods when pilots expose only the mast assembly for a single scan to acquire targets.
The old way
Typically, a system bench to measure the testing data requires several racks of equipment containing a workstation or mainframe computer for control and compiling of software; several VME-based single-board computers dedicated to specific tasks; and occasionally stand-alone items such as oscilloscopes and spectrum analyzers. These items connect to the radar to control it and collect critical performance data. A Digital VAX minicomputer post-processes the real-time data and stores it on magnetic tape.
These benches excel at automated production testing, but may cost as much as $2 million per copy, including the costs of development. "We had a huge problem in that funds were available for only five benches," Vitallo says. "We needed an additional four - with extensive modifications to support additional radars located in EMI testing, environmental qualification tests, software/hardware integration, and flight testing. But we needed a transportable instrumentation system that could be easily and cheaply interfaced to the radar. Plus, we had to have sophisticated, real-time triggering, and time-correlated data filtering that simply isn`t available in any standard bench."
Vitallo and his team looked for a less expensive and easier way to collect and analyze data using the Tektronix DAS 9200, a real-time data capture tool with high-end speed, channel count, memory depth and data analysis capabilities. It can capture and analyze large amounts of data.
The new way
The DAS and a 60 MHz Pentium PC were the center of a small, portable testing apparatus for the Longbow radar system project. Besides a savings of more than $1 million per unit, the new system provided advantages for data collection and analysis.
Because of the DAS high-impedance probes, the engineers could cut right into the cables to collect data with minimal data interruption. In addition, they were able to capture the data in real time at the test site, and plot it in readable form with easy-to-use Microsoft Excel templates, rather than complex and expensive VAX software.
Tektronix engineers developed a Windows application that takes the data out of the DAS, translates hex data into ASCII, and brings the data over to the PC in an ASCII file that looks like the DAS display. "If the files are binary, hex, or decimal in the DAS, we were able to get the same look in the ASCII files," Vitallo says. "That enabled us to dump the data directly into Excel."
The radar system has two array processors, so when it dumps out data-such as range versus frequency-it comes out scrambled and in a format difficult to visualize. The Northrop Grumman engineers were able to use an Excel interface to grab the data. From the array processors, Northrop Grumman engineers developed baseband digital video data for phase and amplitude. Visual Basic (built into Excel) created templates for reordering and plotting the data in a meaningful way.
"It takes only a few minutes to hook the DAS/PC transportable bench directly up to the aircraft itself and get a quick plot of critical fire control radar data in near real time," Vitallo says. "In the old days we had to process the data off-line on the VAX, which took hours."
Northrop Grumman experts use four Tektronix DAS systems altogether: one for testing the effects of susceptibility to interference; one DAS for each of the two benches for developing and testing the radar system`s software; and one "floater" DAS used for testing on the different aircraft wherever it`s needed.
Preserving data flow
The Tektronix DAS solved several potential problems regarding the data buses. First of all, the Northrop Grumman engineers needed a quick and cost-effective method to cut into the radar system`s cable bundles to view the data, without disturbing the data flow. They needed to consider the unique clocking schemes in the data buses. The engineers needed to record both array processors at the same time (in parallel). Tektronix experts developed a customized clocking program in the DAS to address the radar system`s several levels of clocking, and correlate the data from three serial/parallel buses.
"In the front end, the Tektronix DAS has a well-designed set of probes with extremely high impedance podlets, so there`s no intrusion on data," Vitallo says. "With these probes, we were able to cut directly into the middle of the cable bundles going into the radar system and tap them off without disturbing the data flow. We didn`t need to deal with any line receivers or build boards to re-drive the buses, which can complicate the process. It took only one week of the technicians` time to break into the cables and less than a week to write the first Excel templates to make this procedure work."
The Northrop Grumman engineers were worried about picking up single-ended data and not the differential in the severe EMI environment, so they added a double over-braided shield to the cable (#24 twisted shielded pair wires for each data bit). When they collected the data, they were surprised that it never skipped a beat and there were no false alarms, even at the end of a 30-foot instrumentation cable.
"We had to watch the cable impedances and terminations very carefully when we break into the cables - the key is properly terminated data buses and using high impedance probes," Vitallo says. "Often, designers at the board level assume line impedance values of 90 to 130 ohms, and for many it is a shock to see those high-frequency clocks trashed by simple cable mismatch between two boxes separated by only 10 feet on the aircraft." Typically, measured differential impedance is only 45 to 70 ohms for #24 twisted shielded pair on the aircraft.
The engineers knew that to make the cable interfaces work, lead lengths teeing off the bus must be kept short. Technician Donald Harris devised a means of embedding the DAS flying lead sets into breakout boxes so that any engineer could simply insert or remove the DAS interface cables without disturbing the radar. In this manner, they could quickly and safely move the DAS to different tasks - without even turning off radar system power.
The Northrop Grumman engineers also helped Tektronix build a bus probe module for the Mil-Std 1553 bus. The module, which is a 2-by-5-by-6-inch box, connects to the bus via a coax cable and converts the data to a format that can be fed into the DAS or any logic analyzer. The module automatically switches to the redundant bus whenever the aircraft does, so the DAS can`t miss any data.
These boxes are applicable to electromagnetic interference testing and qualification of the Longbow Radar. Thirty-foot cables carrying radar instrumentation feed data outside the sealed, copper-lined screen test room. The engineers triple-shielded them and calculated that the DAS should accurately receive the single rail signals even under intense electromagnetic field bombardment.
"We have had absolutely no problems with dropped data bits or glitched clocks," Vitallo says. "Still, it was a relief to see theory and practice shake hands after cranking up that first field exceeding 350 volts/meter."
According to Vitallo, there are two general methods of interfacing the DAS probes to the radar system cable. The simplest interface is when the user positions the DAS at the end of a dedicated cable with no other receivers. In this case, the engineer simply selects a terminating resistor matching the characteristic line impedance and positions it as close as possible to the DAS problet. If the lines are dual rail, it`s generally best to terminate both rails to a common resistor matching the differential characteristic impedance. If there`s a concern about radiated emissions or unbalanced lines, unused DAS probes should be attached to the group of rails not being monitored in order to closely maintain the twisted pair balance and thus cancellation of the radiated field from the wire pair. Because the DAS probes are single rail, the Northrop Grumman engineers used the shield of each twisted pair as a signal return.
In the second case, when it is necessary to hack into or near the mid-point of a cable length, the same rules apply except that the bus simply passes through the box - no terminations are used at this point. Now, maximizing the impedance of the probe intersection is most important. It may require the use of additional damping resistors included with the DAS`s flying probe sets.
The key is keeping the total probe/lead set impedance at least a few times greater than the characteristic impedance of the cable being intersected at the highest frequency clock being picked off of the cable bundle. Finally, to minimize crosstalk and maintain the highest possible impedance, the lead lengths should be kept short. The propagation time from the intersection point to the DAS problet should be less than one-third the rise time of the data signal being monitored.
Excel templates
The Northrop Grumman engineers needed to quickly plot and analyze hundreds of thousands of data points whenever the DAS triggered on a system anomaly, so that they would only interrupt the sweep generators driving the standard 200 volts/meter fields when absolutely necessary. This effectively saves one week of test time for each polarization and antenna position of the radiating test antenna. The problem was how to get that data into a PC for analysis, and in a form that any engineer or technician can use.
Using Excel spreadsheets, Northrop Grumman developed Visual Basic macros to sort the data, identify valuable data, and arrange it in ways that were graphically meaningful. In real time, the DAS takes a snapshot of the data and triggers on fire control radar anomalies to capture and feed data to the Excel templates for post-processing. The engineers developed Excel templates for sorting baseband/video, fast Fourier transforms, calibration data, and servo responses.
According to Vitallo there are several advantages with using Excel. "It`s cheap, capable, and well-known - anybody can make charts with Excel," Vitallo says. "I was amazed that within five or six days the programmers had created several Visual Basic macros to convert and sort the radar`s raw data. The best thing was that whenever a new problem came up, any engineer responsible for a specific software/hardware design could quickly walk up to our bench and make his own plots to analyze-there was zero learning curve once the basic templates were in place."
Designing an off-the-shelf testing apparatus proved to be a time-saver for the Northrop Grumman engineers, as well as an accurate and cost-effective investment for their customer. Not only was Northrop Grumman able to directly save the government millions of dollars, but there is no long term cost associated with storage and maintenance of equipment. The DAS, PC, and Excel will move on to new tasks, or be returned to inventory.
Jennifer Haupt and Lynn Parker are technical writers based in Seattle, Wash. Their telephone number is 206-285-5280, and fax is 206-285-5286. Haupt`s e-mail address is [email protected], and Parker`s e-mail is [email protected].
