Application Note 101

   Testing Air Bag Inflators
at
Atlantic Research Corporation

The Task: Recording Transient Data

Recording multi-channel time-related data during transient events can be a difficult data acquisition problem, but transient testing is common in some industries. Two obvious examples are crash testing and testing of explosives. These applications require recording once-occurring high-speed transient data so that the amplitude-time relationship of several data channels can be accurately compared.

Today, the preferred solution to most testing requirements uses digital data acquisition systems because they make data easier and faster to process, analyze, and store. Yet, when you use digital systems for transient testing, you must employ special care to design a recording system that faithfully captures the time-amplitude characteristics of the data without adding distortion caused by the digital process. That requires careful attention to several important system considerations:

· The system requires very high channel sampling rates to capture an accurate representation of the waveshape.

· Crosstalk between channels must be eliminated or minimized to prevent amplitude distortion.

· Preservation of inter-channel relationships requires that all channels be sampled simultaneously and controlled by a common precision clock.

· The latency of the host computer must not add time-related errors to the data.

Traditionally, digital systems employ multiplexed signals in which multiple channels share one or more high-speed analog-to-digital converters. The disadvantages of the multiplexed approach include:

· Channel-to-channel interference caused by insufficient amplifier bandwidth and parasitic capacitance.

· Channel sampling rate dependent upon the number of channels.

· The need for complex and unique high-speed communications interfaces.

· These systems usually require external signal conditioning and amplifiers for low-level sensor inputs.

To overcome these problems, Neff's System 495 uses a proven, wide range, guarded differential amplifier-per-channel technique with a dedicated high-speed analog-to-digital converter for each channel. Features include:

• A true differential amplifier-per-channel with ranges of ±5mV FS to ±10.24V FS in 12 programmable Steps
• A tri-filar wound transformer on every channel to protect against high frequency common mode noise
• Bridge conditioning with programmable excitation and auto-balance built in
• Six-pole filter with four programmable cutoff frequencies
• Dedicated 1 MHz, 12-Bit ADC or 250 KHz, 14-Bit ADC per channel
• Dedicated plug-in memory with up to 64 MSamples/Channel

 

System 495 Data Acquisition & Recording System

Features

• Differential Amplifier and ADC Per Channel
• 12 Programmable Full Scale Gain Steps
• 4 Programmable Filter Steps
• Simultaneous Sample & Hold
• 1 MHz Channel Sample Rate at 12-bits
• 256 kHz Channel Sample Rate at 14-bits
• 1 MHz Aggregate Real-Time Throughtput
• Dual Speed Clock
• Programmable Bridge Excitation Voltage
• Program-initiated Bridge Auto-Balance
• Bridge Excitation Read Back
• Shunt-R Calibration
• Voltage Substitution Calibration

This design provides the features that are important to high-speed transient recording systems:

• Very high channel sampling rates to 1 MHz: With the dedicated ADC per channel transient recorder design, only the ADC limits sample rates.
• No crosstalk: Each channel is independent with no multiplexing or shared circuits.
• Negligible skew: All channels are sampled simultaneously and controlled by a common precision clock.
• Independent of computer latency: System clocks control sampling rate, triggers initiate recording and data is stored on-board. This eliminates errors caused by the inability of the host computer to pace the high-speed recording process.

 

Inflator Testing at Atlantic Research Corporation

Atlantic Research Corporation, a division of Sequa Corporation, designs and builds inflators for automobile airbags. An inflator is basically an explosive device with somewhat predictable characteristics. ARC has facilities in Knoxville, Tennessee and Camden, Arkansas. Their corporate offices are in Gainesville, Virginia.

At each of the inflator facilities, there are bays designed to test all the different styles of inflators built by ARC. Multiple pressure measurements along with firing circuit characteristics are collected beginning at the instant of firing and continuing for several milliseconds thereafter. Neff high-speed data acquisition systems are critical to the operation of these inflator test bays.

The Knoxville, Tennessee manufacturing plant uses the Neff acquisition system to do Lot Acceptance Testing. There, the company produces several styles of inflators and a production run of one style of inflator is classified as a lot. Samples from each lot are tested to verify performance against an established standard. Lot acceptance or rejection is based on the pressure performance of each sample.

ARC Knoxville is expanding and upgrading their Neff System 490 equipment to the System 495. The new systems provide them with data transfer rates that are faster by a factor of 20.

Also located at Knoxville, the engineering staff is responsible for the design and support of inflators built by ARC, and for the Engineering Prototype Lab where new designs are built and tested. This lab also uses the Neff data acquisition system for all of its pressure data collections.

ARC, Camden Arkansas, is responsible for design and manufacture of the inflator energetics. Lot inspection energetics, and new design energetics are installed in take-apart inflator hardware and deployed in a pressure tank. The Neff system is, again, used for data acquisition.

ARC chose the System 495 by Neff for the upgrade of these lab facilities because it allows easy conversion and expansion, faster transfer rates and preserves the existing test capability. One other very important issue, the proven reliability demonstrated by the System 490 encouraged ARC to look to Neff when they needed the new equipment.

The Test Process

Inflator testing is done in special test bays. To set up a test, inflators are installed in fixtures that are attached to, or located inside a pressure vessel. The physical size of the inflator basically determines the physical size of the pressure vessel used; The standard sizes of pressure vessels are 1 cubic foot, 60 liter, 100 liter and 146 liter. Pressure transducers are installed in the pressure vessel, and sometimes on the inflator body.

The system includes the trans-ducers, the System 495 and a personal computer with custom software. Each pressure transducer and the ignition current signal is assigned a separate channel. The system is armed and the ignitor is fired with the high-energy current pulse. As pressure builds inside the tank the System 495 records pressure at the various points within the tank along with the characteristic of the current-pulse.

The System 495 amplifies each data channel using a wide bandwidth differential amplifier. A 6th order low-pass Bessel filter with four program selectable cutoff frequencies filters the amplified data. This anti-alias filter removes unwanted high frequency noise and sensor ringing. A sampling A/D converter converts the data into digital information.

Data samples are temporarily stored in DRAM memory, with a separate plug-in memory module provided for each channel. ARC uses the 256Ksample module. After the test, the host computer retrieves each channel's data over a standard SCSI interface.

ARC uses the System 495 “Block Record” operating mode. In this mode, blocks of data are recorded, initiated by an external trigger signal. When the trigger signal is received a programmable number of samples are recorded.

The System 495 samples and records the high-speed analog transient data at sampling rates up to 250KHz. Memory and sample rate management circuitry is provided to control all channels. This circuitry includes a programmable sample rate clock, 24-bit memory address register and 24-bit sample counter. Sampling can be clocked by the internal 5 MHz clock or by an external TTL (or 5V CMOS) level clock. ARC takes advantage of the internal sample rate clock to acquire time-multiplexed data. Others sometimes use the external clock feature to synchronize sampling with external events. A control register bit selects the clock to be used. The internal clock is programmable from 0.298 Hz to 2.5 MHz in 200 nSec steps.

Other Important System 495 Features and Specifications

System Channel Capacity: Multiple-enclosure 495 systems can be interconnected in a master/slave array with the master providing a common clock to all enclosures. Each 7-inch high 19-inch rack case can house up to 16 data channels. Sixteen cases can be interconnected for a system with 256 data channels.

Bridge Transducer Conditioning: A plug-on Bridge Conditioner Mode Board is available for bridge-transducer applications. It provides facilities for mounting bridge completion, shunt R cal and balance limit resistors. Single, two and four arm bridge circuits can be used with either local or remote excitation sensing. The excitation supply is programmable from 0 to 10 volts with a 12-bit DAC. Maximum operating output current is 100 mA, and is short-circuit protected. Auto Balance is programmable.

Operating Modes: The System 495 can operate in two different modes; the Pre/Post event trigger mode and the Block Record mode. When operated in the Pre/Post mode, A/D data is written continuously to memory prior to the receipt of a trigger signal; new data overwrites old data in a circular manner. When the trigger is received, sampling continues until a programmed number of samples are recorded. The memory then contains N samples taken after receipt of the trigger signal and (Available Memory - N) samples taken before the trigger.

The host program polls the System 495 status register to determine when recording has stopped. Next, the host reads the current memory address pointer location and subtracts the sample count value to determine when the trigger signal occurred. The host then reads the channel's memory data into a file. The various parameters used for recording can be appended to the front of the file for use in future data reduction and display.

The block record mode is used to record separate blocks of data, each started by receipt of a trigger signal. The host computer sets the System 495 memory address pointer and specifies a sample count value. When a trigger signal is received, a programmed number of samples are recorded starting at the specified memory address location. The host polls the System 495 to determine when recording has stopped then reloads the new address pointer and sample count values.

The next trigger signal initiates the next record block providing the system has been rearmed. After the required number of record blocks has been recorded, the host reads the channel memory data into a file with numbered record blocks.

Recording Trigger: A trigger signal is used to start recording acquired data into channel memory. Three methods of generating the trigger signal are provided: an external logic level, an analog level threshold detection on one channel, and program control. Control register bits are provided to enable external TTL and analog level detection trigger modes.

The external TTL logic signals may be either pulses or level transitions of either polarity selectable by jumper. The analog level detector initiates a trigger signal when a channel's output signal level exceeds a preset threshold level. The threshold level is defined by an 8-bit value (7 bits plus sign) ranging from -10.24 to +10.24 volts. Resolution is 80 millivolts. Polarity is programmable so that a trigger can occur when the signal goes either more positive or more negative than the threshold level. Recording of the programmed number of samples continues even if the signal level falls back below the trip point.

Program control is primarily used for calibration and checkout before a test is performed. The host computer can position the memory address pointer, define a sample count value and start recording. With a known input signal, system accuracy tests can be performed. The address and sample count registers can also be checked.

Real Time Monitoring: In addition to storing data locally in memory, a subset of each channel's data can selectively be transmitted to the host at an aggregate rate of 1MHz for monitoring or recording. The host computer can read the current A/D data values of all channels in the system. This is a "real time" monitor function accomplished by executing an input scan list. Each channel will return the current A/D register contents at the time it is read. The host computer can use this data to drive an external display, and to test setup and system calibration.

Other Specifications:
Full-Scale Input: 12 programmable steps provide bipolar input ranges from ±5 mV to ±10.24 V
Full-Scale Accuracy: ±0.1% of FS at zero frequency
Full-Scale Stability: ±0.02% of FS at constant temperature; ±0.005%/°C
Input Impedance: 10 Megohms shunted by 500 pF
Common Mode Voltage: ±10 Vdc peak AC operation, ±30 V without damage
Common Mode Rejection: 66 dB + gain in dB to 120 dB max, dc to 60 Hz with 350-ohm imbalance
Filter: 6-pole Bessel filter with four programmable cutoff frequencies. A plug-in resistor network determines each frequency over the range of 100 Hz to 200 KHz. Default frequencies are 10 KHz, 20 KHz, 50 KHz and 100 KHz.
Phase Coherency: Phase shift matched channel-to-channel within 5 degrees with amplifiers on same gain and filter step, dc to cutoff.
Weight: Each case (without amplifier/ADC cards) weighs 30 pounds.

Acknowledgements:
Charlie Woods at ARC Camden, Arkansas
Rick Robbins at ARC Knoxville, Tennessee
Special thanks to William Stout at ARC Knoxville, Tennessee for providing details.