1. A p p e n d i x
B Summary of HALT
Results at a HALT Lab
This is referenced from Section 34.2 from the main portion of this book.
This study analyzed HALT data from many different companies from a
variety of industries and illustrates the following concepts:
1. HALT can be applied to a wide variety of electrical and electromechan-
ical products in almost any industry.
2. Products today are much more robust than in the past and therefore
these methods are necessary to improve product reliability.
3. Random vibration is much more effective than rapid temperature cy-
cling for accelerating defects, and the combined environment of ran-
dom vibration and rapid temperature cycling is even more effective.
4. The types of failures that HALT discovers are the same types of fail-
ures that are found in the field.
5. Don't overdesign your product—design to your specifications, then let
HALT show you were the weaknesses are.
6. Your product is likely more robust than you think. Typically, there are
just a few issues you need to concern yourself with, and HALT is an
excellent technique to find these issues.
The examples in this study were obtained between 1995 and 2010.
I subjected each product in the study to a standard HALT regime of cold
step stress, hot step stress, rapid temperature transitions, vibration step
stress, and combined rapid temperature transitions with vibration step
stress. For each product, we used either a Qualmark or a Chart HALT
chamber, similar to the ones shown in Figure B.1.
How Reliable is Your Product? 50 Ways to Improve Product Reliability 295

2. Figure B.1: Qualmark OVS2.5 HALT Chamber (left) and Chart Real36
(right) Used for Study
The study is comprised of data on 294 products from 157 companies across
thirty-eight different industries. The majority of products were electrical, but
several of the products had mechanical components as well. No specific data
about any one product is presented so that confidentiality is maintained
between the lab and its customers. The industries participating are shown in
Table B.1. This is an updated study from the one I performed in 1996. I
reviewed more data from the past fourteen years and added to the study as
shown in Tables B.1 and B.2. I reviewed the new results, compared them to the
old results, and found that they were very similar. Therefore, the takeaways are
the same. For this reason, I did not update with the new data for Figure B.2 and
for Tables B.3 through B.10.
296 Appendix B: Summary of HALT Results at a HALT Lab

3. Table B.1: Distribution of Companies by Industry
Industry Types Number of Companies Product Type
1 Networking Equipment 26 Electrical
2 Medical Devices 25 Electromechanical
3 Defense Electronics 11 Electrical
4 Microwave Equipment 6 Electrical
5 Computers 6 Electrical
6 Semiconductor Capital Equipment 6 Electromechanical
7 Video Processing Equipment 5 Electrical
8 GPS 4 Electromechanical
9 Robotics 4 Electromechanical
10 Smart Meter 4 Electrical
11 Commercial Aviation Electronics 3 Electrical
12 Display 3 Electrical
13 Oil and Gas 3 Electromechanical
14 Solar 3 Electrical
15 Storage System 3 Electromechanical
16 Homeland Security 2 Electromechanical
17 Automotive 2 Electromechanical
18 Computer Peripherals 2 Electromechanical
19 EDA Platforms 2 Electrical
20 Electric Car 2 Electromechanical
21 Fiber Optics 2 Electrical
22 Hand-held Equipment 2 Electromechanical
23 Industrial Applications 2 Electromechanical
24 LED/Lighting 2 Electrical
25 Optical 2 Electromechanical
26 Point of Sales Equipment 2 Electromechanical
27 Audio 2 Electrical
28 Recreation 2 Electromechanical
29 Government/Research 2 Electrical
30 Restaurant/Food Industry 2 Electromechanical
31 Space 2 Electrical
32 Audio 2 Electromechanical
33 Supercomputers 2 Electrical
34 Teleconferencing Equipment 2 Electromechanical
35 Test Equipment 2 Electrical
36 Wireless 2 Electrical
37 Mobile Phone 2 Electromechanical
38 Nuclear 1 Electromechanical
TOTAL 157
How Reliable is Your Product? 50 Ways to Improve Product Reliability 297

4. The majority of companies included in this data have products that are intended
for the office environment. The product's end-use environments are shown in
Table B.2.
Table B.2: Distribution of Products by Environment Type
Environment Percent Thermal
Vibration Environment (1)
Type of Total Environment (1)
Office 38% 0 to 40°C Little or no vibration
Vibration only from user of
Office with User 17% 0 to 40°C
equipment
1–2 Grms vibration, 0–200
Vehicle 17% -40 to +75°C
Hz frequency
Field 15% -40 to +60°C Little or no vibration
Vibration only from user of
Field with User 9% -40 to +60°C
equipment
1–2 Grms vibration, 0–500
Airplane 4% -40 to +75°C
Hz frequency
Note 1 All values are approximates based on a combination of data from individual customers and
from Telcordia and military specifications.
B.1 Results of Study
I combined and summarized the results for all of the products tested. The HALT
methodology I used was similar for each of the products represented—the
average time to complete a HALT was three to four days. The majority of
customers completed the HALT in one visit, correcting the problems at their
facility after the end of the HALT. Some, however, decided to divide the HALT
into two or more visits, verifying corrective actions before implementing them
on production versions of the product.
The failure percentage by stress type is shown in Figure B.2. The order we
applied the stresses for each product was
1. Cold step stress
2. Hot step stress
3. Rapid temperature transitions
4. Vibration step stress and
5. Combined environment (consisting of vibration step stress combined with
rapid temperature transitions)
298 Appendix B: Summary of HALT Results at a HALT Lab

5. Because we subjected all of the products to the stressing in this order, it
becomes apparent when reviewing Figure B.2 why the failure distribution
occurred as it did. For instance, because we always performed cold and hot
step stress prior to rapid temperature transitions, a much higher percentage of
failures occurred during the step stress. Our clients corrected these issues or
created workarounds for these issues prior to performing rapid temperature
transitions. Likewise, because we always performed rapid temperature transi-
tions and vibration step stress prior to performing combined environment, a
much higher percentage of failures occurred during the rapid temperature tran-
sitions and vibration step stress. Our clients corrected these issues (or created
workarounds for these issues) prior to performing combined environment. Nev-
ertheless, note that it is important to perform combined environment because
had we skipped this stress, we would have missed twenty percent of the
potential failures.
Combination of Cold Step
Vibration and Stress = 14%
Rapid
Temperature
Transitions =
20%
Hot Step
Stress = 17%
Rapid Temperature
Vibration Step Transitions = 4%
Stress = 45%
Figure B.2: Failure Percentage by Stress Type
In Tables B.3, B.4, and B.5, I summarize the HALT limits and present them in
three different ways. For Table B.3, I grouped all of the products together, and
I show the HALT Limits for the entire set of products. For each limit, I give the
average, most robust, least robust, and median values. For Table B.4, I
How Reliable is Your Product? 50 Ways to Improve Product Reliability 299

6. grouped the products by environment using the same environment categories
as in Table B.2. For Table B.5, I grouped the products by product application
using the three product applications of military, commercial, and field.
For all products, I used a HALT chamber with input vibration consisting of
broadband energy from 0 to 10 kHz. This was a key factor in the types of failure
modes uncovered because the products had a wide range of component sizes
and technologies, and each required different frequency bands for excitation,
i.e., large mass parts require low frequency (10 Hz to 1 kHz) to excite them
while small mass parts require high frequency (> 1 kHz). In addition, the HALT
chamber had a temperature rate of change of 60°C/minute which we used for
rapid temperature transitions and combined environment stressing. This was
also important because we know some of the failures we uncovered were due
to the temperature rate of change.
For the temperature data, I recorded the thermal data in °C and used the
average product temperature at the point of failure for each product tested as
the measurement point. For the vibration data, I recorded the data in Grms in
a bandwidth from 0 Hz to 3 kHz (even though the system was providing energy
up to 10 kHz). I took the measurement up to 3 kHz because this is traditionally
the bandwidth that HALT practitioners have used and is a measure of the
maximum product response level at the point of failure for each product tested.
For each individual reading in each of the averages, the data point I used was
the worst case data point if I tested more than one product and was the limit of
the product before our clients implemented any corrective actions. For all
vibration readings, I used product response values and not table inputs.
Table B.3: HALT Limits by Limit Attribute
Thermal Data, oC Vibration Data, Grms
Attribute LOL LDL UOL UDL VOL VDL
Average -55 -73 +93 +107 61 65
Most Robust -100 -100 +200 +200 215 215
Least Robust 15 -20 +40 +40 5 20
Median -55 -80 +90 +110 50 52
The results shown in Table B.3 are significant because they indicate that even
though the majority of the products were commercial products using commer-
cial grade components for the office environments, the average limits achieved
300 Appendix B: Summary of HALT Results at a HALT Lab

7. were far beyond the limits specified by the component manufacturers. In fact,
the majority of the components used in commercial products are rated to
operate from 0°C to +70°C with little or no vibration being specified.
Table B.4: HALT Limits by Environment
Thermal Data, oC Vibration Data, Grms
Environment LOL LDL UOL UDL VOL VDL
Office -62 -80 92 118 46 52
Office with User -21 -50 67 76 32 36
Vehicle -69 -78 116 123 121 124
Field -66 -81 106 124 66 69
Field with User -49 -68 81 106 62 62
Airplane -60 -90 110 110 18 29
The results shown in Table B.4 follow closely with the expected order of limits
(although the absolute levels are probably beyond expectations).
Table B.5: HALT Limits by Product Application
Product Thermal Data, oC Vibration Data, Grms
Application LOL LDL UOL UDL VOL VDL
Military -69 -78 116 123 121 124
Field -57 -74 94 115 64 66
Commercial -48 -73 90 95 32 39
The results shown in Table B.5 again follow the expected order of limits closely,
with the military products being the most robust. However, notice that the tem-
perature limits achieved for commercial products was close to the temperature
limits achieved for military products. This is another strong argument for the
use of commercial-off-the-shelf (COTS) components for military products
because the commercial components are almost as robust, usually providing
ample margins even in the military application, and commercial components
are almost always significantly less expensive than military components.
How Reliable is Your Product? 50 Ways to Improve Product Reliability 301

8. B.2 Failure Summaries by Stress Type
Tables B.6 through B.10 are failure summaries for each type of stress applied.
The significance of the data in these tables is that the majority of failures shown
are common field failures for commercial equipment. The HALT process
merely accelerates what will most likely take place over a much longer period
of time under field usage.
One way to determine if HALT will be effective in your application is to compare
Tables B.6 through B.10 with your field failures. To the extent they match up is
likely to be how effective HALT will be for you.
Table B.6: Cold Step Stress Failures
Failure Mode Qty
Failed component 9
Circuit design issue 3
Two samples had much different limits 3
Intermittent component 1
Table B.7: Hot Step Stress Failures
Failure Mode Qty
Failed component 11
Circuit design issue 4
Degraded component 2
Warped cover 1
Table B.8: Rapid Temperature Transition Failures
Failure Mode Qty
Cracked component 1
Intermittent component 1
Failed component, cause unknown 1
Connector separated from board 1
302 Appendix B: Summary of HALT Results at a HALT Lab

9. Table B.9: Vibration Step Stress Failures
Failure Mode Qty
Broken lead 43
Screws backed out 9
Socket interplay 5
Connector backed out 5
Component fell off (non-soldered) 5
Tolerance issue 4
Card backed out 4
Shorted component 2
Broken component 2
Sheared screws 1
RTV applied incorrectly 1
Potentiometer turned 1
Plastic cracked at stress point 1
Lifted pin 1
Intermittent component 1
Failed component 1
Connectors wearing 1
Connector making intermittent contact 1
Connector broke from board 1
Broken trace 1
How Reliable is Your Product? 50 Ways to Improve Product Reliability 303

10. Table B.10: Combined Environment Failures
Failure Mode Qty
Broken lead 10
Component fell off (non-soldered) 4
Failed component, cause unknown 3
Broken component 1
Component shorted out 1
Cracked potting material 1
Detached wire 1
Circuit design issue 1
Socket interplay 1
304 Appendix B: Summary of HALT Results at a HALT Lab