AIAA 96-2874, Proceedings of 32nd Joint Propulsion Conference, Orlando FL, July, 1996.


Donald L. Jackson
Lockheed Martin Astronautics
Denver, CO 80201-0179

Barry T. Neyer, Member AIAA
EG&G Star City, Inc.
Miamisburg, OH 45343-0529

Michael K. Saemisch
Lockheed Martin Astronautics
Magna, UT 84044-0509

Current Address
Barry T. Neyer
Excelitas Technologies
1100 Vanguard Blvd
Miamisburg, OH 45342
(937) 865-5586
(937) 865-5170 (Fax)


This paper describes the development and qualification testing of the High Voltage Detonator and Adapter built by EG&G for Lockheed Martin Astronautics for use on the Titan IV space launch vehicle. Even though the program had an extremely tight schedule, a detailed development program was conducted to optimize the design where possible, to ensure the design was robust, and to provide insurance that the qualification testing would be successful on the first attempt. Details of the objectives of the development program and the results are provided. In addition, the required qualification testing is described and the results provided. The paper concludes that the resulting HVD is a robust reliable device. Further, it is shown that the development program proved to be worthwhile effort, resulting in changes that assured successful qualification.

Technical Papers of Dr. Barry T. Neyer


The Lockheed Martin Astronautics Titan IV launch vehicle program team found itself in the need of a fully qualified and delivered exploding bridgewire detonator, and they needed it quickly. They needed it in less than a year, in fact, after a contract was awarded to EG&G Star City. In order to meet this program objective, it was obvious that there would be barely enough time for one attempt at qualification; therefore complete success had to be assured. It was decided that a rapidly conducted but very thorough development program was needed to assure that, as much as possible, qualification would be a success.

Rather than create a new detonator design, an impossible task within the time allowed, the method chosen to meet this critical objective was to take an established design, the High Voltage Initiator (HVI) developed for a Navy application, and perform the minimum modifications required to turn it into a High Voltage Detonator (HVD). The HVI was a good choice for this approach since the HVI shared some of the heritage with the Titan program that is beyond the scope of this paper. This gave the HVI program advantages with certain design aspects. That is, the connector interface and firing interfaces were similar to the Titan interfaces. In addition, the manufacturability of the HVI allowed easy modification and building of testing devices in order to support a detailed development test program.

This paper describes the development and qualification of the Titan IV Solid Rocket Motor Upgrade (SRMU) High Voltage Detonator (HVD) and an HVD Adapter that successfully met this challenge. The robust design will ensure mission success for the required Titan IV functions. A companion paper, AIAA 96-2873 (Tibbitts et al. 1996), describes the design and manufacturing details of the HVD itself. This paper contains summaries of the development and qualification programs. Full details of the effort are contained in the EG&G Star City Reports RP1004 - HVD Development Test Report (Neyer 1996hd) and RP1005 - HVD and HVDA Qualification Test Report (Neyer 1996hq) .

HVD Development Challenge

The major challenge of the HVD development program was to produce a technically robust design and conduct a development program that would establish margin of critical design requirements and high confidence of passing qualification testing. The short time schedule required did not leave any time to recover from qualification test failure. Thus, a thorough development program was required. Because of the short time schedule, we had to begin production before we could complete the full development tests. Thus, critical design parameters were studied early in the development stage, while the less critical elements were studied in sufficient time to impact the final assembly.

The HVD was chosen as a replacement for another component late in the SRMU program, due to problems with the original component. Therefore, the form, fit, and function of the device were fixed before the project began. The input stimulus to the HVD was fixed by the existing design of the High Energy Firing Unit (HEFU) and High Voltage Cable (HVC). The output required of the HVD was determined by the requirements of the Reinforced Confined Detonating Cord (RCDC). Because some of the specifications for components connected to the HVD were extremely broad, the HVD had to be designed to accommodate a wide range of input stimuli.

Adapter Development Challenge

The HVD Adapter has several functions. It secures the HVD to the SRMU structure, provides mechanical alignment and standoff between the HVD and RCDC, and confines detonation products from the HVD and RCDC. EG&G Star City designed the HVD Adapter to overcome two major challenges.

First, the envelope, initially designed to accommodate a different EBW of smaller diameter than the HVD, was very limited and provided mounting locations not along the centerline of the adapter. The adapter center of gravity was cantilevered away from the mounting fasteners, which was expected to result in unacceptable resonance when exposed to a vibration environment. The HVD adapter was designed to move the fasteners closer to the center of gravity but the envelope constraints precluded complete mitigation of this concern.

Figure 1: HVD Adapter Cross Section

Second, relative orientation of the connectors for the HVD and mating HVC was random and necessitated an adapter design feature that would allow the HVD to be rotated to mate with the HVC. The resulting design consisted of three pieces, a base, a cylinder, and a top as shown in Figure 1. The base is attached to the structure. After the HVD and RCRD are torqued into the cylinder, the cylinder is placed into the base. The top is attached to the base with two fasteners capturing the cylinder between the two.

The challenge for the adapter development program therefore became establishment of margin with regard to detonation transfer and confinement of detonation products with no structural failure. In addition, the adapter had to be designed to mitigate the effects of the environments on the HVD / adapter / RCDC subsystem during dynamic testing.

HVD and Adapter Qualification Challenge

Once development testing was concluded, the results were reviewed, and final design selected. Complete units were then built to be subjected to a qualification program. The program encompassed operational environments to ensure mission success, but did not introduce unknowns or over-test conditions that could cause a failure unrelated to actual application conditions. The qualification requirements, based on a tailored version of DOD-E-83578A, had already been thoroughly scrubbed by the Titan IV ordnance community. Any failure would be a setback that would cause the original objectives to not be met.

In addition, even though not detailed in the requirements, it was decided that the HVD should not be completely qualified. The subsystem containing the HVD should also be qualified to ensure all interfaces have been properly characterized and to simulate actual "test as we fly" conditions as much as possible. Past experience had shown that component interactions must be understood and can potentially create environments in excess of those predicted for a particular component.

Description of the HVD Device

The HVD is a hermetically sealed, glass-ceramic filled, Exploding Bridge Wire (EBW) detonator with an internally sealed spark gap. Illustrated in Figure 2, it is built according to EG&G Star City drawing 324001. The HVD design is based on two other components that were designed and fabricated at EG&G Mound, the predecessor to EG&G Star City. One is the High Voltage Initiator (HVI) that was made for the Navy's Trident II gas generator launch system. The other is the MC4217, an EBW that was designed by Sandia National Laboratories for the Department of Energy. The HVD is essentially identical to the HVI except it has the hermetically sealed EBW powder column and bridgewire of the MC4217. The powder column uses an initial pressing of 2-(5-cyanotetrazolato)-penta-amine cobalt (III) perchlorate (CP) and an output pellet of octahydro-1,3,5,7-tetranitro-1,3,5,7-tetrazocine (HMX).

Figure 2: Cross Section of an HVD

HVD Test Device

The HVD test device was designed to mimic the output half of the full HVD assembly. Shown schematically in Figure 3, the test devices were manufactured according to modified HVD procedures. Like the HVD, it consists of an Inconel 718 shell filled with S-glass ceramic. The pins extend out the back of the device for electrical input. The powder cavity was machined into the S-glass, just as in the actual HVD output half. The cavity dimensions were changed according to the development test plan. For many of the development tests, this device was sufficient and cost much less than a full HVD. Some of the tests, such as the threshold tests, required a device such as this, because it is difficult to determine the threshold of the bridgewire-powder interface, with the spark gap in the unit. Full HVD units were used in the Design Verification Lot and for certain HVD to CDC adapter margin tests. Some of the development tests used HVI shells and fired HVIs because the HVI's form and electrical performance are essentially the same as the HVD's

Figure 3: Cross Section of an HVD Test Device

HVD Development Program


The objectives of the development program were:

Interfaces and Systems Identified for Margin Assessment

HEFU/HVC to HVD Interface

The firing input interface of the HVD was to an already designed High Energy Firing Unit (HEFU) and a High Voltage Cable (HVC). Both the physical and firing input aspects of this interface were defined in the interface specifications for the HVD and were not changeable.

The High Voltage Cable (HVC) attaches to the HEFU and connects to the HVD. It was designed to interface with a component that had insulators at the base of the pin, something the High Voltage Initiator design did not possess. The purpose of these insulators was to increase the distance of the electrical ground path from pin to case to ensure high voltages did not short to ground when firing. The HVC has a rubber insulator with a cut-out to match these insulators providing a nice sealing interface as well. Modification of the HVC was not programatically possible.

An early design decision to be made in the development program was to determine the need for these insulators on the HVD and design and test them if needed. Because the insulators are made from a thin material bonded to the conducting pins, they could crack if the pins underwent even a small amount of bending.

The gap between the base of the pins and the connector in the HVI design served as the preferred discharge path of electrostatic discharges pin to case. Increasing this distance would make this path less preferred, possibly forcing discharge through a deleterious path.

To determine the need for the insulators, the approach was to first determine the potential for arcing the firing pulse from one of the pins to ground without any insulator being present in the HVD or HVC. In addition, the effects of reduced pressure, which allows arcing at a lower voltage, must be considered. One of the HVD functions is performed at high altitudes when the solid rocket motors are staged from the core vehicle. The tests were conducted at pressures between atmospheric pressure and 0.5 Torr, with both DC voltages and pulse stimuli from the HEFU simulator. The simulator was designed to have similar electrical characteristics as a "worst case" HEFU, but had a variable voltage. ("Worst case" for these experiments was a broader pulse, to increase the likelihood for electrical breakdown.)

A compromise insulator design concept was also tested: a round 0.063" radius node at the base which provided a 0.060" high insulator was thought to be easily manufactured, did not possess the weaknesses of tall thin insulators, and only increased the ESD path slightly.

The rubber insulator present in the HVC was removed for these tests. These vacuum tests, a deliberate over-test, show margin and account for uncertainty in the degree of sealing from this insert.

Figure 4: Steady State DC Breakdown vs. Pressure

The results of the DC tests are shown in Figure 4. The results show the expected decrease in breakdown voltage as the pressure decreases to 2-3 Torr as predicted by Paschen's law. Also, the effect of the raised node insulator is to increase breakdown voltage by increasing arc length. The extra distance has its greatest effect at atmospheric pressure.

The pulse tests used the HEFU simulator charged to 3000V (the highest HEFU specification value). The simulator delivered a typical fire pulse: about 2s wide and 1500 amps peak current. The results showed initially that this was insufficient voltage to arc at any pressure within the design gaps of the HVD. Electrical breakdown could be induced only by creating a point gap of width 0.015," and reducing the pressure to 1 to 5 Torr. Because the pin-to-case and pin-to-pin gaps are both approximately 0.090" in the HVD, there is no possibility for external electrical breakdown to occur over an operational range much larger than the design requirements.

It was concluded from these sets of tests that the design without raised nodes had a large amount of margin over shorting a high voltage pulse to ground. Therefore the current design was acceptable without these nodes. However, the nodes proved to be easy to manufacture and were incorporated into the HVD. The benefits that they provided in terms of increasing margin and also ensuring a pressed contact between the end of the rubber boot in the HVC and the HVD. Performance during ESD was analyzed and determined not to be a concern; this was born out during the design verification testing to be described later.

Other aspects of the HVC interface were also studied but are not detailed here. An increased pulse width and effects of salt water contamination on the interface were tested and successfully showed margin of the selected design in these areas.

Bridgewire to CP Interface

One of the lot qualification requirements is a threshold test (Group II in Table 2). The threshold for firing the HVD is determined by two conditions requirements: (1) sufficient voltage to break the spark gap, and (2) sufficient energy to ignite the explosive powder. A reliable design requires that the spark gap threshold be sufficiently higher that the bridgewire / explosive powder interface. In such a design, all HEFU pulses that cause breakdown of the internal spark gap have sufficient energy to reliably detonate the device. Thus, it is highly unlikely that a stray electrical pulse could dud the device.

We studied both the spark gap breakdown and bridgewire / explosive powder thresholds independently during development to ensure that we understood each separately. The spark gap threshold test is described in a later section; this section describes the bridgewire / explosive powder interface studies.

We performed threshold tests on seven combinations of CP powder particle size, density, bridgewire, and environmental conditions. Units for these tests incorporated nominal HMX pellets and CP column length. These tests were designed to serve several purposes. The main purpose was to verify that the proposed design had adequate margin to meet the qualification requirements. A second goal was to ensure that the design was robust; that there were no combination of design parameters (introduced by manufacturing variability, for example) that would cause a failure. A third goal was to test different combinations of designs in order to validate the predictions of the modeling that we used to design the HVD.

All of the threshold tests were performed using the test device shown in Figure 3. Using this device allowed us to determine the threshold of the bridgewire / explosive powder interface because there was no spark gap in the circuit. The threshold tests were conducted using the program Optimal (Neyer 1994o), based on the Neyer D-Optimal Test (Neyer 1994), a more efficient replacement for Bruceton and Langlie tests. The analysis was performed using MuSig (Neyer 1994m), which uses the likelihood ratio test (Neyer 1991) to compute realistic confidence intervals.

Figure 5 shows the results of the threshold test for the 1.0 density fine particle CP with 0.035 inch long bridgewire. This set of parameters is the same as the final design of the HVD. A plot showing the confidence regions of this data set is shown in Figure 6.

Figure 5: 1.0 Density CP Analyzed with PlotSen

Because these tests were all conducted using a limited sample size, the All-Fire level (99.9% at 95% confidence) can not be compared with the corresponding All-Fire level estimated from the larger sample size. Nevertheless, the tests showed that the bridgewire / explosive powder interface would reliably initiate the explosive at CDU voltages much smaller than the 1600 volt requirement.

Figure 6: 1.0 Density CP Analyzed with MuSig

Table 1 lists the results of the analysis on all devices tested. The large variation in standard deviations shown in Table 1 is due to the large variation in the bridgewire lengths in the test devices. These devices were manufactured by hand using an open setup. This manufacturing arrangement results in greater variability in the bridgewire lengths, which directly translates into performance variation. The average standard deviation reported for the seven tests is approximately 60 volts. Because the manufacturing tolerance for the HVD is much tighter than for the test devices, we felt confident that the standard deviation would be less than 60 volts for the bridgewire / explosive interface on the HVD.

Table 1: Bridgewire Threshold Tests

Test Group

Bridge Length


Stand Dev.

1.0 Density, course CP




0.9 Density, fine CP




1.0 Density, fine CP*




1.1 Density, fine CP




1.2 Density, fine CP




1.0 Density, fine CP, After Environments




1.0 Density, fine CP, After Environments, Fired Cold




*HVD design based on testing and Sandia modeling.

Figure 7 shows the threshold of the HVD test devices as a function of density. The data shows that the threshold decreases as density increases. As can be seen from the results, the HVD threshold will meet program requirements at and density between 0.9 and 1.2 g/cc.

Figure 7: HVD Threshold as a Function of CP Density

The 1.0 density CP with a 0.035 inch length bridgewire was chosen for the HVD design based upon modeling performed by Sandia National Laboratories. The design was required to work with all combinations of design margins for the High Energy Firing Unit (HEFU). The HVD parameters were chosen based on the modeling to ensure that the HVD would function with any HEFU in the allowed parameter range. The model showed that the design chosen would have adequate margin for any allowed HEFU. The 1.0 density design was chosen because it had the best combination of reliability and safety over the entire system design range.

Spark Gap Interface

The design of the HVD is such that the Qualification Test requirement of a 45 shot threshold test is primarily a test of the enclosed spark gap, and not of the bridgewire / CP combination. The HVDs are required to have an all-fire voltage of less than 1600 volts. To ensure that the HVDs pass this requirement, it is required that both the spark gaps and the bridgewire / CP must have an all-fire voltage of less than 1600. This test was designed to measure the all-fire of the spark gaps. (The All-Fire requirements for the HVD were established during Qualification Testing, reported in a later section.)

Spark gaps similar to the HVD spark gaps, but with a higher breakdown voltage, were placed one at a time into an unloaded HVD with a heavy bridgewire. The resistance of the heavy bridgewire remained constant during the test and from test to test. The criterion for success for this test was that there be current flow through the device. If no current flows, then obviously the real HVD could not function. As was shown in the previous section, the all-fire level for the bridgewire / CP interface is much lower than that for the spark gap. Thus, if the simulator charge voltage is large enough to cause current flow through the spark gap, a real HVD would fire.

Figure 8: PlotSen Analysis of Spark Gaps

The threshold tests were conducted using the Neyer D-Optimal Test (Neyer 1994). Figure 8 shows the analysis of the spark gap test, while Figure 9 shows the MuSig confidence region analysis. Analysis of the data shows an all-fire level (99.9% at 95% confidence) of these test gaps of 1634 volts. This test suggests that the HVD spark gaps would have an all-fire voltage of 200 volts less, or approximately 1430 volts. Because the all-fire voltage of the HVD gaps is well below the requirement of 1600 volts, this component of the HVD meets the all-fire requirement.

Figure 9: MuSig analysis of Spark Gaps

The Effects of Explosive Column Design on Output

The nominal explosive column of the HVD consists of 42.5 mg of CP pressed against the bridgewire to a density of 1.0 g/cc (0.136" diameter, 0.180 length) and a 40 mg HMX pellet (1.7 g/cc; 0.100" length; 0.133" diameter) loaded against the CP. In order to establish margin of the nominal design, two series of tests were performed in which parameters of the explosive train were modified and the performance of the HVD monitored.

Figure 10: HMX Length Dent Depth Measurements

In the first series, a total of 12 test units were tested with the CP load maintained at nominal parameters and the HMX pellet length reduced to 0.075" and 0.050" while maintaining a nominal density of 1.7 g/cc. For these tests the HVD cavity length was modified to accept the difference in HMX pellet length. Half of the units were monitored for output by a dent block and the remaining half by VISAR. A previous paper (Neyer 1995v) described the advantages of using the VISAR to determine the output of detonators. Because the qualification specification required the use of dent test results, those results are presented here. The results of this test are compared with output from the nominal HVD and presented in Figure 10. The output requirement for the HVD is 0.010" dent and as can be seen in Figure 10, the nominal design length of 0.100 provides more than adequate margin to meet the requirement.

Testing the output with VISAR showed flying plate velocities of 2.66 0.2 km/s for the 0.075" pellet length and 2.44 0.06 km/s for the 0.050" pellet length compared with 2.80 2.85 km/s for nominal pellet length. Since additional work performed during development indicated that a flyer velocity of 1 km/s was sufficient to initiate the interfacing RCDC, the 0.050" HMX pellet length was considered sufficient to reliably initiate the RCDC. Because the minimum dent required in Lockheed Martin engineering was 0.010," however, the HMX pellet length remained at 0.100".

The second series of tests was performed to evaluate the effect of CP column length on output of the device. Units for these tests were modified to incorporate a CP column with nominal density of 1.0 g/cc but with modified lengths of 0.150", 0.120", 0.090", and 0.060" (0.180" nominal). A nominal HMX pellet was used for these tests and output was measured with dent blocks. Results of these tests as well as from nominal HVD design are presented in Figure 11.

The data show that a growth to detonation in a column of 1.0 density CP will not occur within the first 0.120," but will by 0.150." The 0.180" length of CP was chosen for the HVD design to ensure that there was adequate margin.

Figure 11: CP Length Dent Depth Measurements

The conclusion from these series of tests is that the HVD design is robust at the nominal design parameters for explosive column and the performance is stable within the established design tolerances.

Structural Robustness of the HVD

Another significant difference in the HVI from the HVD was in the dynamic environments the HVD was required to survive prior to functioning. The HVD qualification environment includes a random vibration spectrum of 31.42 grms. Several specific areas of concern were identified as the HVI was being adapted to the HVD and each was studied in the development phase to determine the acceptability of the design.

Spark Gap / Bellows

An obvious concern was the effects of the random vibration effect on functioning of the unit. The bellows act as both an electrical conductor of the firing pulse between the spark gap and the input pin and as a spring to force itself and the spark gap into physical contact with the pin contacts and each other. If the dynamic forces induced into the bellows were large enough to overcome the spring force, an electrical discontinuity would be created. Even though the requirements allowed testing of the device in dynamic environments without any measurement of continuity and THEN firing, it was decided that this did not adequately ensure mission success since some of the units are fired during flight. Tests were therefore established to verify continuity during exposure to the predicted environments.

A unit was assembled with a nominal bellows. In order to detect discontinuity, the spark gap was replaced by a conductor to allow continuous current flow through the HVD. An aluminum slug of equal mass, size and shape was inserted in the place of the spark gap. A 100 mA constant current source, connected to a 100VDC power supply, was utilized to supply loop current. Any momentary interruption in the circuit causes a voltage increase at the point from about 0.2 volts to 10 volts. A Fluke 87 multi-meter, set to peak-hold mode, and a Hewlett Packard 54111D digitizer constantly monitored voltage across the HVD. The tests showed the capability to withstand up to 73.6 grms without a loss in continuity. Because this was such a significant over-test, even a "worst case" bellows has ample margin over the qualification environment of 31.42 grms.

HVD Margin to Environmental Conditions

The environmental requirements for the HVD were also greater than the HVI in terms of thermal cycling, random vibration and shock. In order to ensure adequacy of the design and show robustness, 30 flight-like units were built and subjected to all of the qualification environments, but with increased levels. These tests established both the confidence for passing qualification and the robustness of the design. This testing is described in the following sections.

Design Verification Test Program

RF Testing

RF testing was performed during the Design Verification. HVDs were randomly selected and tested according to MIL-STD-1576, Method 2204 (tailored to EBW testing) to measure the RF Impedance and RF Sensitivity of the HVDs.

All HVDs passed the RF Impedance and RF Sensitivity tests.

After passing the RF Impedance and RF Sensitivity tests the HVDs were fired at the specified all-fire voltage of 1600 volts using the fireset simulator and cable that have the same circuit characteristics as the HEFU. The HVD output was determined by a dent test according to MIL-STD-331, Test 301.1 (tailored to take four baseline readings instead of two).

All HVDs fired. The minimum dent was 15.1 mils, with an average of 16.2 mils and a standard deviation of 0.4 mils. The function time, measured from start of current rise to explosive output was 3.0 microseconds for all HVDs. All HVDs remained intact with no housing failure after functioning.

Environmental Over-testing

To ensure that the HVDs would pass the Lot 1 Qualification Tests, some of the Development and Design Verification HVDs were subject to the same environments required for lot qualification, but with more extreme test ranges. The threshold tests on environmented devices discussed in a previous section had a larger thermal shock range than required (-40 F to + 170 F versus a requirement of -29 F to +161 F), longer and more sever vibrational testing (50 grms for 8 minutes versus a requirement of 31.42 grms for 3 minutes) and a more severe shock profile due to the excessive ringing in the test fixture. In addition, the test devices that were fired cold were fired at a temperature of -40 F versus a requirement of -20 F.

Throughout the development program, the goal was to over-test wherever possible to detect any design flaws. Because the HVDs and test devices passed all of the over-tests, we were confident that there would be no problems during Lot Qualification.

Adapter Development Program

In order to attain confidence for passing qualification testing, the following objectives were established to determine content of the adapter development test and analysis programs:

Assessment of adapter design concluded that the design parameters for which to establish confidence were detonation transfer, structural integrity to contain products of detonation without structural failure, stability when subjected to dynamic environments, and environmental seal of the adapter when HVD and RCDC were installed.

Detonation transfer was initially assessed by VISAR testing of the HVD that concluded that after the flyer had traveled 1.5 mm its velocity was approximately 2.9 km/s. The initiation threshold of the RCDC was attained by flyer velocity of approximately 1.0 km/s. Based on VISAR plots the standoff between HVD and RCDC was established to be 1.1 to 1.9 mm (0.044" to 0.076"). VISAR data indicated adequate flyer velocity at min standoff to produce reliable detonation transfer. To confirm the predictions of the analysis, development testing incorporated detonation transfer margin testing as follows:

All tests were successfully conducted at ambient temperature thereby providing confidence that a detonation transfer failure would not occur during qualification.

An early design of the adapter had a cutter that helped to shear the closure disk of the HVD. Flash X-ray work at Sandia revealed that the flying plate tended to fracture and flew at an angle when the cutter was present, but remained intact and flew straight on when there was no cutter. The flash X-ray data also showed that the flyer was robust far past the impact point in the adapter design. The detonation transfer margin testing discussed above was conducted on both cutter and non-cutter designs. Both designs demonstrated acceptable detonation transfer margin.

In order to establish that the adapter would properly contain the pressures due to detonation of the HVD and RCDC and would not produce a blowout failure of the RCDC, six modified adapters were prepared for test. These adapters had the internal volume reduced to 80% of the minimum design volume, the external dimensions reduced to maintain the same wall thickness as the minimum design value, and the thread engagement of the HVD and RCDC reduced. Three adapters with HVDs and RCDCs installed were fired at -20F and three at +161F. All tests were successful with no structural failure of HVD, adapter, or RCDC. The program therefore concluded that a blowout or structural failure would not occur during qualification.

Significant concern existed over dynamic stability of the adapter when subjected to shock and vibration. Initial vibration testing and monitoring of adapter response showed significant cylinder resonance in the applicable frequency range. An extensive program was undertaken utilizing damping materials, various tie down torques, and design modifications in order to achieve dynamic stability. The final design incorporated three point mounting feet between base and SRMU structure, and reduction of central outer diameter of cylinder to ensure the ends of the cylinder were captured by clamp halves. Final testing of the HVC / HVD / adapter / RCDC subsystem showed cylinder response to closely follow vibration input to the adapter.

The final aspect of adapter development involved incorporation and validation of an environmental seal for the adapter when the HVD and RCDC were installed to mitigate the concern over moisture in the detonation transfer interface. The RCDC already incorporated a seal on its end tip. The sealing properties were validated by a flow-through helium leak test. An O-ring was incorporated at the interface between HVD hex and adapter cylinder. The integrity of this seal was verified by a flow through helium leak test.

In conclusion, the development program was adequately defined to identify all potential failure modes, and resulted in design modifications and establishment of confidence for the adapter qualification program.

Qualification Testing

HVD DOD-E-83578A Testing

Qualification testing requirements were provided by Lockheed Martin but had their genesis primarily in the government specification DOD-E-83578A. These requirements had previously been thoroughly scrubbed and were key criteria in designing and developing the HVD. Organizations involved in the tailoring of the requirements, besides Lockheed Martin, were the Air Force Titan IV Program Office, Aerospace Corporation, TRW, Alliant TechSystems (the SRMU subcontractor to Lockheed Martin) and the Air Force Range Safety Organizations. Justifying the finalized requirements for the HVD are beyond the scope of this paper, but were thought by all parties to ensure both mission success and safety of the final product.

Prior to being accepted for random selection as either a test unit or deliverable unit, each HVD was subjected to acceptance tests. These are the same acceptance tests that were performed on all deliverable units. They cover visual, helium leak, electrical and radiographic inspections.

Table 2: HVD Qualification Tests





















Individual Unit Acceptance Tests









ESD Testing









RF Impedance









RF Sensitivity









All-Fire Test









Pin Pull









High Temp Exposure









Thermal Cycling



























Drop Test (2 Meter)









Individual Unit Acceptance Tests









No-Fire 500 VDC









No-Fire 250 VAC









Ambient Firings (1600 Volts)









Temperature Firings









*The HVDs for RF impedance and RF sensitivity were tested during development.

The requirements for qualification were based on a 191 unit test program. The units were assigned to a group, each of which had a specific set of test requirements. The breakdown of the overall program is shown in Table 2. In addition to the groups specified by the tailored requirements specification, a group was created for the "Subsystem Tests" that were not required but thought were needed to test as close to flight-like conditions as possible. These subsystem tests are shown in the right hand column of the table. The table is interpreted as follows. The number at the top of each column is the number of the group. The next number indicates the total number of units assigned to that group. As you read down each column, the number indicates the number of units of this particular group that were subjected to the particular test on that line.

The key parameters of the qualification were to thermal cycle the parts from -29F to +161F, shock to 1300g and random vibration test to 31.42 grms. To ensure proper testing and the absence of over-test, the shock and vibration test fixtures underwent certification from Lockheed Martin with dummy units prior to the actual qualification tests.

The units must pass a set of tests similar to the individual unit acceptance tests after exposure to environments. This ensures that there is not significant degradation during exposure of the environments and is a primary concern of the requirements specification.

The test methods were based on the methods required by MIL-STD-1576 and were detailed in a program plan approved by the government prior to testing. This prior approval ensured community concurrence with the methods and procedure details prior to initiating the tests.

The RF sensitivity and impedance tests were performed by the Franklin Institute. The remainder of the tests were conducted at EG&G Star City and EG&G Mound, Miamisburg, Ohio.

Threshold Tests

Figure 12: PlotSen Analysis of Threshold Data

Figure 12 shows the results of the analysis of the sensitivity test performed during Lot Qualification. The threshold test showed that the HVDs have a mean of 1023 Volts, and a standard deviation of 6.25 volts. The small standard deviation shows that the HVDs are extremely consistent in their firing characteristics.

Figure 13: MuSig Analysis of Threshold Data

The MuSig graphical analysis is shown in Figure 13. The confidence regions are much smaller than for the threshold test on the HVD test device and the spark gaps. These devices were built using certified tooling by experienced assembly technicians, while the components testing during were constructed "by hand."

The ProbPlot (Neyer 1994p) analysis of the data is shown in Figure 14. This plot shows probability levels as a function of the charge voltage on the HEFU capacitor. The analysis shows that the 95% confidence 99.9% level is 1055 volts, well below the 1600 volt requirement. The HVDs pass the all fire requirement with a margin of more than 60 standard deviations. Moreover, because the HVDs are extremely consistent in their behavior Figure 14 also shows that the HVD would pass the much harder requirement of 99.9999% at 99.5% confidence.

Figure 14: ProbPlot Analysis of the Threshold Data

Function Tests

The results of the qualification were 100% successful. There were no anomalies or failures associated with any of the tests. No waivers to any of the required tests were required. In all cases, the units showed that they not only passed the required tests but that there was no significant degradation or near-miss test results. The results were detailed in the qualification test report that provided the actual procedures as-run and the tests data. This report was a formal submittal to the government for approval.

All HVDs fired at the combinations of temperature and charge voltage required. All dents were far above the 10 mil specification. Figure 15 shows the dent depths measured as a function of fireset voltage for all the qualification tests conducted at ambient temperature (23 Threshold tests from 1000 to 1400 volts, 50 tests from groups 1, 4, 5, 6, and 7 at 1600 volts, 12 tests from group 7 at 2500 volts, and 5 tests from group 7 at 3000 volts.). Also shown are the average depth and the standard deviation for four voltages. (The average and standard deviations have been offset by 100 volts in the graph for clarity.) The data indicate that the dent depth is independent of charge voltage.

Figure 15: HVD Dent Depths at Ambient Temperature

The tests conducted at hot or cold temperatures are not shown on the graph. The dent blocks for these tests were conditioned with the HVDs to ensure that the HVD temperature was at the specified value when the HVDs were fired. The dent depths were lower for the cold shots (13.5 mils to 14.8 mils) and higher for the hot shots (15.8 mils to 17.8 mils). The change in dent depth with temperature is due to changes in the dent block material properties with temperature.

The average for all the dent depths for all HVDs tested at ambient temperature is 15.67 mils, with a standard deviation of 0.56 mils. Most of the variation is probably due to measurement error. Nevertheless, the data indicate that the dent depths exceed the 10 mil requirement by ten standard deviations.

Adapter Subsystem Testing

The final 9 HVDs from the qualification program were devoted to subsystem testing in which the HVC, HVD, adapter, and RCDC were assembled in flight-like configuration and subjected to qualification thermal cycling, vibration, shock, individual acceptance tests, no fire tests, and firing tests at -20F, ambient, and +161F. During vibration and shock tests the HVC and RCDC were routed in a manner to simulate flight installation. At the completion of environmental tests the subsystems were disassembled for visual, leak, and radiographic inspections and no fire tests. No indications of degradation due to environments were noted. Radiographic inspection of the RCDC revealed no breaks or neck downs in the MDF. The subsystems were then reassembled into their original configuration, temperature conditioned, and test fired. All firings were successful with proper swell cap indications from the RCDC output ends.

Conclusion / Closing Remarks

The High Voltage Detonator (HVD) met the extreme challenge of short schedule and a "Must succeed the first time" requirement. This was due to several factors:

The resultant design is a robust, reliable unit exceeding all of the Titan IV requirements.


The authors wish to thank John Adams, Andy Demana, Jim Edwards, David Haas, Dan Knick, Dick Massey, Al Munger, Connie Schaeffer, Mark Stoltz, Julie Thomes, and Al and Carol Tibbitts of EG&G Star City, Miamisburg, OH and Jere Harlan and John Merson of Sandia National Laboratories, Albuquerque, NM.


DOD-E-83578A, "Military Specification, Explosive Ordnance for Space Vehicles (Metric), General Specifications for," USAF, 15 October 1987.

MIL-STD-1576, "Electro-explosive Subsystem Safety Requirements and Test Methods for Space Systems," USAF, 31 July 1984., 1994.

Elton E. Tibbitts, Julie A. Thomes, T. Andrew Demana (1996), "Design, Description, and Fabrication of the High Voltage Detonator," proceeding AIAA 96-2873 of the 32nd AIAA/ASME/ASEE Joint Propulsion Conference, Lake Buena Vista, FL.

Barry T. Neyer (1991) "Sensitivity Testing and Analysis," Proceedings of the 16th International Pyrotechnics Seminar, June 1991, Jnkping, Sweden.

Barry T. Neyer (1994) "A D-Optimality-Based Sensitivity Test," Technometrics, 36, 61-70.

Barry T. Neyer (1995v), "Use of VISAR Instead of Dent Tests", Use of VISAR Test to Replace Detonator Dent Tests," Proceeding AIAA 95-2702 of the 31st AIAA/ASME/ASEE Joint Propulsion Conference, San Diego, CA, 1995.

Barry T. Neyer (1996hd), "RP1004 -- HVD Development Test Report," EG&G Star City Inc.

Barry T. Neyer (1996hq), "RP1005 -- HVD and HVDA Qualification Test Report," EG&G Star City Inc.

Neyer Software (1994m), MuSig Program, Neyer Software, Cincinnati, Ohio.

Neyer Software (1994o), Optimal Program, Neyer Software, Cincinnati, Ohio.

Neyer Software (1994p), ProbPlot Program, Neyer Software, Cincinnati, Ohio.