Model Specifications for Breath Alcohol Ignition Interlock Devices (BAIIDs), 8047-8059 [06-1423]

Agencies

• National Highway Traffic Safety Administration
• DEPARTMENT OF TRANSPORTATION

[Federal Register Volume 71, Number 31 (Wednesday, February 15, 2006)]
[Notices]
[Pages 8047-8059]
From the Federal Register Online via the Government Printing Office [www.gpo.gov]
[FR Doc No: 06-1423]


-----------------------------------------------------------------------

DEPARTMENT OF TRANSPORTATION

National Highway Traffic Safety Administration

[Docket No. NHTSA-2005-23470]


Model Specifications for Breath Alcohol Ignition Interlock 
Devices (BAIIDs)

AGENCY: National Highway Traffic Safety Administration (NHTSA), 
Department of Transportation (DOT).

ACTION: Request for comments.

-----------------------------------------------------------------------

SUMMARY: This notice seeks comments about what revisions are needed for 
the Model Specifications for Breath Alcohol Ignition Interlock Devices 
(Model Specifications) published by the National Highway Traffic Safety 
Administration (NHTSA) in the Federal Register on April 7, 1992 (57 FR 
11772). Model specifications are guidelines for the performance and 
testing of breath alcohol ignition interlock devices (BAIIDs). These 
devices are designed to prevent a driver from starting a motor vehicle 
when the driver's breath alcohol content (BrAC) is at or above a set 
alcohol level. Because changes may be necessary after more than 13 
years of experience with this technology, NHTSA is seeking comments 
regarding the need for revisions to the model specifications.

DATES: Written comments may be submitted to this agency and must be 
received by April 17, 2006.

ADDRESSES: Comments should refer to the docket number and be submitted 
(preferably in two copies) to: Docket Management, Room PL-401, 400 
Seventh St., SW., Washington, DC 20590. Alternatively, you may submit 
your comments electronically by logging onto the Docket Management 
System (DMS) Web site at https://dms.dot.gov. Click on ``Help & 
Information'' or ``Help/Info'' to view instructions for filing your 
comments electronically. Regardless of how you submit your comments, 
you should mention the Docket number of this document. You may call the 
docket at (202) 366-9324. Docket hours are 9:30 a.m. to 4 p.m., Monday 
through Friday.

FOR FURTHER INFORMATION CONTACT: Dr. James F. Frank, Office of Research 
& Technology (NTS-131), National Highway Traffic Safety Administration, 
400 Seventh St., SW., Washington, DC 20590. Telephone: (202) 366-5593.

SUPPLEMENTARY INFORMATION: On April 24, 1991 (56 FR 18857), NHTSA 
issued a notice and request for comments on proposed Model 
Specifications for Breath Alcohol Ignition Interlock Devices. BAIIDs 
are breath alcohol test instruments designed to allow a driver to start 
a motor vehicle when his/her BrAC is below a set alcohol level; 
conversely, the devices are designed to prevent a driver from starting 
a motor vehicle when his/her BrAC is at or above the set alcohol level.
    As explained in the April 1991 notice, a number of States passed 
laws authorizing the use of ``certified'' BAIIDs, giving those States 
the responsibility for developing certification standards and test 
procedures. Consequently, a number of States and manufacturers of these 
ignition interlock devices requested that the Federal government 
develop and issue certification standards for BAIIDs. After receiving 
and considering comments, NHTSA adopted and published model 
specifications for BAIIDs in the Federal Register on April 7, 1992 (57 
FR 11772).
    Since publication, many States have incorporated these model 
specifications, or some variation of them, into their State 
certification requirements, thereby serving the purpose for which they 
were originally intended. Forty-three States allow the use of BAIIDs, 
and they are currently being used in connection with sanctions for 
Driving While Intoxicated (DWI). Persons required to use BAIIDs are 
either under the supervision of a responsible state agency (e.g., a 
Motor Vehicle Administration) and/or under direct court supervision.
    The experience of the last 13 years has shown that the issuance of 
model specifications and test procedures for BAIIDs has served to 
encourage a degree of consistency among the States while at the same 
time providing sufficient flexibility for States to address their 
individual needs or legislative requirements. The model specifications 
and test procedures were drafted in such a way to enable States to 
adopt them with minimal effort. However, the ignition interlock 
industry has matured, the technology has changed, and the technical and 
social environments have changed in the past 13 years. Therefore, it is 
NHTSA's view that revisions to the model specifications are 
appropriate.
    NHTSA has not prepared a proposal for revised model specifications 
for BAIIDs at this time. Rather, NHTSA invites all interested parties 
to submit comments on what revisions are needed to update the model 
specifications. NHTSA is especially interested in obtaining comments 
from interested parties about the areas listed below. This notice also 
invites all interested parties to offer additional remarks, suggestions 
and commentary above and beyond the areas highlighted below:
    (1) Accuracy and precision requirements. Are the current 
specifications for 90% accuracy at 0.01% w/v above the set point in the 
unstressed testing conditions, and 90% accuracy at 0.02% w/v above the 
set point in the stressed testing condition appropriate? Should the new 
model specifications change the set point from 0.025% w/v?
    (2) Sensor technology. Should the model specifications limit sensor 
technology to alcohol-specific sensors? The model specifications 
currently include performance requirements but do not address what 
technology should be used to satisfy those performance requirements. In 
other words, the model specifications allow semi-conductor sensors, 
which were widely used during the early years after devices were first 
introduced into the marketplace. Alcohol-specific, fuel cell sensors 
appear to be more common today, but it is not clear whether the model 
specifications should limit devices to an alcohol-specific technology. 
NHTSA seeks comments regarding the advantages and disadvantages of 
limiting the model specifications to an alcohol-specific (fuel cell) 
technology, or other emerging technologies versus relying on 
performance requirements only.
    (3) Sample size requirements. The model specifications set the 
minimum breath sampling size at 1.5 liters. Informal comments received 
over the years have indicated that this requirement may be too high. 
NHTSA will consider lowering the breath

[[Page 8048]]

sampling requirement, and/or including a requirement for both a minimum 
sample size and minimum back pressure at the input (mouthpiece) of the 
device. NHTSA requests comments regarding such a change.
    (4) Temperature extreme testing. The model specifications call for 
testing at -40 [deg]C, -20 [deg]C, +70 [deg]C and +85 [deg]C, but allow 
for the removability of alcohol sensing unit so it may be kept warm 
(cool) when the vehicle is expected to be subject to extremely cold 
(hot) temperatures. NHTSA seeks comments about whether this approach to 
temperature extreme testing is sufficient, or whether more stringent 
demands should be made on equipment.
    (5) Radio Frequency Interference (RFI) or Electromagnetic 
Interference (EMI) testing. The RFI testing protocol in the model 
specifications, however incomplete, uses power sources that are no 
longer commonly in use. New power sources (e.g., cell phones) that have 
output power commensurate with equipment in use today need to be 
identified. NHTSA welcomes comments suggesting appropriate levels of 
power for use in this RFI testing.
    (6) Circumvention testing. The model specifications offer a number 
of procedures for evaluating whether existing devices can be easily 
circumvented. NHTSA seeks comments about whether these test procedures 
have proven adequate, or whether new or modified tests should be 
incorporated into the model specifications.
    (7) The Vehicle-Interlock Interface. Anecdotal reports from 
ignition interlock manufacturers have suggested that it is sometimes 
difficult to install existing interlock systems in some of the newer 
electronic ignition systems. NHTSA seeks comments from all interested 
parties about whether NHTSA should establish any guidelines regarding 
the vehicle-interlock interface. More specifically, NHTSA invites 
comments regarding the feasibility and likelihood of incorporating 
generic hardware into vehicles to which commercially-available ignition 
interlocks could be connected.
    (8) Calibration stability. NHTSA invites comments regarding whether 
the calibration stability testing is sufficient in length and/or 
whether ignition interlocks should be required to hold their 
calibration for longer periods of time, thereby requiring less frequent 
calibration checks.
    (9) Ready-to-use Times. NHTSA seeks comments about whether it 
should establish a ``ready-to-use'' time period for extreme cold 
temperatures, such that devices must operate within a given period of 
time under extreme cold conditions.
    (10) NHTSA testing. NHTSA seeks comments about whether it should 
undertake the responsibility for testing of ignition interlocks against 
its model specifications and subsequently publish a Conforming Products 
List (CPL) of devices meeting those NHTSA guidelines.
    (11) International Harmonization. NHTSA seeks comments about the 
importance of the harmonization of the ignition interlock model 
specifications with standards in other parts of the world, such as the 
European Union, Canada, and Australia.
    (12) Specifications for Ignition Interlock Programs. NHTSA seeks 
comments about whether the current ignition interlock community (users, 
manufacturers, states, etc.) favors NHTSA developing model 
specifications for ignition interlock programs, in addition to model 
specifications for devices.
    (13) Acceptance Testing. NHTSA understands that its current model 
specifications involve ``type-testing'' of various models of BAIIDs. 
NHTSA seeks comments about establishing standardized acceptance-testing 
procedures, in addition to the current type-testing guidelines. It is 
not clear what testing might be included in such model specifications, 
or who would conduct the testing.
    (14) NHTSA seeks comments from interested parties on any additional 
areas they believe will enhance the revision of the model 
specifications. This request for comments need not be limited to the 13 
areas identified above.
    In order to assist readers in preparing comments, the current model 
specifications are reprinted as an Appendix to this document.

    Issued on: February 10, 2006.
Marilena Amoni,
Associate Administrator for Program Development and Delivery.

Appendix--Reprint From 57 FR 11774-11787 (April 7, 1992)

Model Specifications for Breath Alcohol Ignition Interlock Devices

Purpose and Scope

    The purpose of these specifications is to establish performance 
criteria and methods of testing for breath alcohol ignition 
interlock devices (BAIID). BAIIDs are breath alcohol sensing 
instruments designed to be mounted in an automobile and connected to 
the ignition key switching system in a way that prevents the vehicle 
from starting unless the driver first provides a breath sample. 
These devices contain an instrument to measure the alcohol content 
of a deep lung breath sample. If the measured breath alcohol 
concentration (BrAC) is at or above a set level the ignition is 
locked and the vehicle will not start. These devices are currently 
being used as a court sanction. Drivers convicted of Driving While 
Intoxicated (DWI) may be required to use these devices on their car 
under court supervision. These specifications are intended for use 
in certification testing of BAIID's used under court supervision.

Definitions

D1 Alcohol

    Ethanol; ethyl alcohol: (C2H5OH).

D2 BrAC

    Breath Alcohol Concentration (BrAC) is expressed in percent 
weight by volume (% w/v) based upon grams of alcohol per 210 liters 
of breath in accordance with the Traffic Laws Annotated, Section 11-
902.1(a) (Supp. 1983). A BrAC of 0.10% w/v means 0.10 grams of 
alcohol per 210 liters of breath (similarly, the Blood Alcohol 
Concentration or BAC associated with a BrAC of .10% w/v means .10 
grams of alcohol per 100 milliliters of blood; except for the 
difference in the referenced volume measure--210 liters of breath 
vs. 100 ml of blood--the referenced grams of ethanol are identical). 
Alcohol concentrations in either breath or in air mixtures can be 
expressed in milligrams of alcohol per liter of air (mg/l); to 
convert mg/l to units of percent weight by volume, multiply by 0.21.

D3 BAIID (Breath Alcohol Ignition Interlock Devices)

    These interlock devices are designed to allow a vehicle ignition 
switch to start the engine when the BrAC test result is below the 
alcohol setpoint, while locking the ignition when the breath test 
result is at or above the alcohol setpoint.

D4 Alcohol Setpoint

    The Alcohol Setpoint is the Breath Alcohol Concentration at 
which the BAIID is set to lock the ignition. It should be noted that 
the alcohol setpoint is the nominal lockpoint at which the BAIID is 
set at the time of calibration.
    Ideally, there should be no occasions when a person with zero 
BAC is blocked from starting a vehicle engine due to the interlock. 
Therefore, to help protect against the response of the alcohol 
sensor to vapors other than ethyl alcohol, such as tobacco smoke or 
mouthwash, and the natural production of gases by human subjects, 
some leeway is necessary at the low end. At the other extreme, a BAC 
of 0.05% w/v has been shown to produce evidence of behavioral 
impairment in some individuals, and in some parts of the country 
(e.g., Colorado and the District of Columbia) 0.05% w/v can be 
presumptive evidence of impairment and grounds for legal action. The 
setpoint must be between the limits of .00% and .05%.
    With some known exceptions, use of a 0.025% w/v alcohol setpoint 
should minimize the possibility that users who have not recently 
ingested alcohol will have trouble starting their engines. A 
discussion of the rationale for selecting 0.025% can be found in 
section 4.1. State interlock program developers requiring use of 
these BAIIDs

[[Page 8049]]

should be aware that even at BrACs which are lower than many states' 
mandated ``legal limit,'' some drivers will already have their 
driving ability impaired.

D5 Breath Sample

    The breath sample is normal expired human breath containing 
primarily alveolar air from the deep lung. See section 4.2 for a 
more detailed discussion.

D6 Fail-Safe

    When the BAIID device cannot operate properly due to some 
condition (e.g., improper voltage, temperature exceeding operating 
range, dead sensor etc.) the BAIID will not permit the vehicle to be 
started.

D7 Tampering and Circumvention

D7.1 Tampering

    An overt, conscious attempt to physically disable or otherwise 
disconnect the BAIID from its power source and thereby allow a 
person with a BrAC above the setpoint to start the engine.

D7.2 Circumvention

    An overt, conscious attempt to bypass the BAIID whether by 
providing samples other than the natural unfiltered breath of the 
driver, or by starting the car without using the ignition switch, or 
any other act intended to start the vehicle without first taking and 
passing a breath test, and thus permitting a driver with a BrAC in 
excess of the alcohol setpoint to start the vehicle.

D8 Safety and Utility

D8.1 Safety Feature

    Any specification related to insuring that the BAIID will 
prevent a driver with a BrAC above the alcohol setpoint from 
driving.

D8.2 Utility Feature

    Any specification related to insuring that the BAIID will 
function reliably and not interfere with driving by operators whose 
BrAC's are below the alcohol setpoint.

D8.3 Optional Feature

    Any specification that is not specifically recommended at this 
time but may be necessary to include at some future issuance of 
certification specifications. Non-inclusion at this time is due to 
lack of evidence that failure to include constitutes a significant 
problem. Also the optional feature may, if implemented, cause the 
cost and complexity of the interlock device to rise substantially.

D9 Certification Tests

    Tests performed to check the compliance of a product with these 
specifications.

D10 Stress Tests

    Any testing protocol which imposes on the BAIID an environmental 
or use-related challenge, such as extreme temperatures, voltages, 
vibrations, or frequent usage.

D11 Filtered Air Samples

    Any human breath sample that has intentionally been altered so 
as to remove alcohol from it.

D12 Device

    A breath alcohol ignition interlock device (BAIID).

D13 False Negative

    A breath alcohol concentration determination that incorrectly 
permits a vehicle to be started when the driver's BrAC is at or 
above the setpoint.

D14 False Positive

    A breath alcohol concentration determination that incorrectly 
prevents the vehicle from being started when the driver's BrAC is 
below the setpoint.

Model Specifications and Test Requirements

1.0.S/T Safety Specifications (S) and Safety Tests (T)

1.1.S Dual Accuracy and Precision Limits (High End)

    The accuracy and precision shall be determined as described in 
paragraphs 1.1.1.S to 1.1.4.S when tested in accordance with section 
1.1.T.
    The accuracy specifications for the BAIID will be different 
depending on the test interventions. Two conditions are recognized: 
unstressed and stressed.

1.1.1.S Baseline Accuracy in the Unstressed Condition

    Following a calibration, and when tested at neutral ambient air 
temperature (10-30 [deg]C), all BAIIDs shall lock the vehicle 
ignition 90% of the time when the true alcohol content of the breath 
sample is 0.01% w/v BrAC (0.01g ETOH/210 liters air) or more above 
the alcohol setpoint.

1.1.2.S Accuracy After One or More Stress Tests

    Following any one or more Stress Tests in which the BAIID is 
subjected to conditions as specified in Definition D10, the BAIIDs 
shall lock the vehicle ignition 90% of the time when the true 
alcohol content of the breath sample is 0.02% w/v BrAC (0.02g ETOH/
210 liters air) or more above the alcohol setpoint.

1.1.3.S Standard Deviation (Precision)

    The accuracy requirement as specified in 1.1.1.S is equivalent 
to distributions of test results with a mean equal to the alcohol 
setpoint (e.g., 0.025% w/v), and a standard deviation equal to 
0.0078% w/v BrAC. The accuracy requirement specified in 1.1.2.S is 
equivalent to a distribution of test results with a mean equal to 
the alcohol setpoint (e.g., 0.025% w/v) and a standard deviation 
equal to 0.0156%.
    Accordingly, under 1.1.1.S, 0.01% w/v BrAC above the alcohol 
setpoint (90% criterion) is equal to approx. +1.28 standard 
deviations. Similarly, under 1.1.2.S 0.02% w/v BrAc above the 
alcohol setpoint (90%) criterion is equal to approx. +1.28 standard 
deviations. This value of standard deviation, derived from a table 
of cumulative normal probabilities can be regarded as equivalent to 
a one-tailed test of significance, and represent the maximum 
allowable imprecision under conditions of perfect accuracy. When 
there is analytic inaccuracy in addition to imprecision, the 
allowable standard deviation will be lower.
    The stable criterion for all test purposes is set as 90% correct 
test outcomes at .01% w/v above the setpoint for Section 1.1.1.S and 
90% correct outcomes for .02% w/v above the setpoint for Section 
1.1.2.S.

1.1.4.S Proportions

    The safety requirement must specify the proportion of tests at 
BrACs of .01% w/v or .02% w/v above the alcohol setpoint at which 
the ignition must be locked. The table below shows the 90% criterion 
for unstressed and post-stress testing.

 Table 1.--Test BrAC Level at Which the Ignition Must Be Locked at Least
   90% of the Time Depending on Whether Test Is Unstressed or Stressed
------------------------------------------------------------------------
                                              Test BrAC level (% w/v)
            Alcohol setpoint             -------------------------------
                                            Unstressed       Stressed
------------------------------------------------------------------------
 0.025% w/v*............................           0.035           0.045
------------------------------------------------------------------------
* Recommended.

    Because the values referenced for allowable error (e.g., 90% 
criterion) are derived from a standard table of probabilities, 
values could also be specified for any point along the hypothetical 
normal distribution with mean equal to the setpoint. For example, 
testing a 99.5% lock criterion (2.57 standard deviations) for the 
unstressed and stressed tests (by using 0.045% and 0.055% w/v 
solutions respectively) would have no practical value because a real 
test of the criterion would require at least 200 repetitions in 
order to reliably detect 1 failure. Therefore all testing as 
specified in 1.1.T is referenced to a 90% lock certainty, requiring, 
as will be noted below, 20 test repetitions for which there may be 
no more than 2 failures.
    A matrix of safety test requirements as specified in Appendix A 
shall be required for full certification of an interlock device. 
Accuracy of thermometers used to monitor simulator temperature and 
the purity of alcohol used shall be traceable to the National 
Institute of Standards and Technology (formerly National Bureau of 
Standards). All test reports must clearly

[[Page 8050]]

specify the equipment used, the manufacturer, model number and 
calibration dates.
    A qualified testing laboratory, chosen by a state to conduct 
these certification tests, shall be capable of establishing their 
own procedures. For reference, however, Appendix B contains the list 
of equipment, setup procedures for testing, and a protocol for 
mixing alcohol test solutions.

1.1.T Accuracy/Precision Tests (High End)

    Two sets of criteria apply to the test outcome, depending on 
whether the BAIID had recently been subjected to a stress test. 
Paragraph 1.1.1.T specifies the criteria for the unstressed tests, 
paragraph 1.1.2.T specifies the criteria for the stress tests.
    All tests shall be conducted on two different BAIIDs. These will 
be referred to subsequently as Device A and Device B.
    The testing shall be repeated 20 times on device A, and 20 times 
on device B. Two types of results shall be recorded: pass/fail, and 
a digital readout. The pass/fail information can be read from the 
user display on the front of the interlock unit. A three decimal 
place digital readout of the vapor alcohol concentration sensed can 
be read from the BAIID display, if available, otherwise it shall be 
taken from an externally connected laboratory test instrument that 
monitors the BAIID's evaluation of the alcohol concentration of the 
introduced sample.

1.1.1.T Unstressed Accuracy/Precision Test Specifications (High End)

    The baseline accuracy testing is conducted as a measure of the 
BAIID's ability to hold to or exceed a 90% accuracy criterion when a 
test solution is .01% w/v above the alcohol setpoint. Accuracy 
testing with this criterion shall be conducted at room temperature 
and initially precede all others to ensure that the fundamental 
operation of the BAIID is initially adequate under no-stress 
conditions.
    If either BAIID fails to lock on more than two occasions in 
those twenty trials with an alcohol concentration of 0.01% w/v above 
the setpoint specification, then it has failed the no-stress 
accuracy test criterion of 90%.

1.1.2.T Stress Accuracy/Precision Test Specifications (High End)

    This accuracy testing is conducted in conjunction with all 
subsequent Stress Tests to be specified in following paragraphs. 
This test protocol is a measure of the BAIID's ability to hold to or 
exceed a 90% accuracy criterion when a test solution is .02% w/v 
above the alcohol setpoint. This test shall be conducted at whatever 
temperature is called for by the test protocol utilizing the test 
criterion.
    If either BAIID fails to lock on more than two occasions in 
those twenty trials with an alcohol concentration of 0.02% w/v above 
the setpoint specification, then it has failed the post-stress 
accuracy test criterion of 90%.

1.2.S Breath Sampling Requirement

    All BAIIDs must require that a minimum of 1.5 liters of breath 
be introduced through the mouthpiece and run through the instrument 
before the alcohol content is measured. Compliance with this 
requirement can be determined by testing in accordance with 
paragraph 1.2.T.

1.2.T Breath Sampling Requirement Tests

    The specification stipulates at least 1.5 liters of air be 
introduced before sampling the alcohol concentration. To determine 
that the interlock device is sampling alveolar air, spirometric 
measurement shall be made on both devices A and B at both the 
minimum acceptable and maximum acceptable delivery pressures as 
specified by the manufacturer.
    If the sampling head of the interlock device is incapable of 
being fitted with a spirometer at the outlet to collect and measure 
all of the vented sample, then this test may be conducted in an air 
tight laboratory box with a transparent viewing window. In such a 
case, place the interlock in the box (fitted with a power outlet as 
needed), connect the output of the simulator to the inlet of the 
interlock via an air-tight feed line, and install a fitting on the 
vent port in the wall of the box. Connect the spirometer to the vent 
port. Measure the volume of air escaping from the vent port as an 
index of the volume of air introduced into the interlock. Record the 
volume of air when the sample is accepted by the interlock device.
    Alternatively, a plastic bag suitably outfitted may be used in 
place of the box. The suitability of this alternative shall be 
verified by using a large (one to three liter) calibration syringe 
to demonstrate that collected volume equals input volume.

Begin Stress Testing Protocols

1.3.S Calibration Stability

    All BAIIDs must meet the accuracy requirements set in paragraph 
1.1.2.S when tested in accordance with paragraph 1.1.2.T after 
having been operated according to paragraph 1.3.T for 7 days longer 
than the period of time specified by the manufacturer in their 
application for certification. Thus, if the manufacturer intends to 
require their BAIID be brought in for maintenance and calibration 
every 30 days, 45 days, or 60 days, this period of time plus 7 more 
days (or 37, 52, or 67 days respectively), would be used to 
determine whether the BAIID met the calibration stability 
requirement.

1.3.1.S Lockout After 7 Days Beyond Service Interval

    A BAIID must prevent engine ignition if it has not been 
recalibrated for a period in excess of 7 days beyond the 
manufacturer's recommended service interval. A warning must precede 
lockout when the manufacturer's recommended interval has passed.

1.3.T Calibration Stability Test

    After completing all other tests required under section 1, the 
BAIIDs shall be recalibrated and remain in a fixed location in the 
testing laboratory for the period of time specified by the 
manufacturer for regular maintenance and calibration, plus 7 days. 
The calibration stability testing should proceed under two 
conditions: alcohol-free and with alcohol present. For nine out of 
ten test days, the BAIIDs shall be run through 10 test cycles per 
day using a human breath sample known to contain no alcohol. On the 
tenth test day, ten tests shall be performed with a known 
concentration of 0.10% w/v ethanol delivered from a simulator.
    The calibration stability regimen shall be repeated five days a 
week during this interval. For example, if a manufacturer's 
recommended calibration interval is 60 days, this will require 
approximately 10 weeks (60+7=67 days) of testing, a total of 500 
calibration stability tests. At least 50 of those tests then would 
be conducted with alcohol. Practically this would involve testing 
with alcohol once every two weeks.
    Before continuing to the next phase of stability testing, the 
protocol described in Section 1.3.1.T should be evaluated.
    Following the calibration stability regimen, the BAIIDs shall be 
retested according to the high end accuracy criteria as set forth in 
1.1.2.S and the test procedures as set forth in 1.1.2.T. In 
addition, however, if the BAIIDs pass the accuracy/precision tests 
according to the criterion of 1.1.2.S (90% accuracy with a test 
solution .02% w/v above the setpoint), then the devices must then be 
recalibrated and be able to pass according to the criterion of 
1.1.1.S (90% accuracy with a test solution .01% w/v above the 
setpoint).

1.3.1.T Evaluation of Lockout for Expiration of Service Interval

    In the course of conducting the calibration stability regimen, 
the BAIID must be shown to prevent ignition if it has not been 
serviced. Determine that the warning signal alerts the user when the 
service interval expires. Determine that lockout ensues in 7 days.
    Return to l.3.T to continue with the recalibration phase of 
testing.

1.4.S Power

    If the BAIID device is designed to be operated from a 12 Volt DC 
vehicle battery, then it shall meet the accuracy requirements 
specified in paragraphs 1.1.1.S to 1.1.4.S when operated within the 
normal range of automotive voltages of 11 to 16 Volts DC, when 
tested in accordance with paragraph 1.4.T.

1.4.T Power Test

    If the submitted BAIID draws its power from the vehicle battery, 
then the device shall be subjected to accuracy testing at both the 
high and low voltages according to the following protocol.
    Devices A and B shall be selected and supplied with 11 Volts DC 
power and then subjected to the test protocol as set forth in 
section 1.1.2.T for accuracy testing.
    Devices A and B shall be selected and supplied with 16 Volts DC 
power and then subjected to the test protocol as set forth in 
section 1.1.2.T for accuracy testing.

1.5.S Temperature

1.5.1.S Operating Range

    All BAIIDs shall meet the accuracy specifications in paragraphs 
1.1.1.S to 1.1.4.S when operated within a temperature range of +85 
[deg]C to -40 [deg]C (+185 [deg]F to -40 [deg]F) and when tested in 
accordance with paragraph 1.5.T for their ability to operate 
properly at low and at high temperatures.

[[Page 8051]]

1.5.2.S Note on Extreme Operating Range

    The BAIID manufacturer may chose to meet the specifications for 
temperature extremes (-40 [deg]C and +85 [deg]C) by having the 
alcohol sensing unit be removable (e.g., so that it may be kept warm 
(cool) when the vehicle is expected to be subject to extremely cold 
(hot) temperatures).
    If the removable alcohol test unit is not removed, and as a 
result is exposed to temperatures outside the manufacturer's 
recommended operating range, then the BAIID shall fail-safe or the 
ignition be rendered inoperable.

1.5.T Temperature Tests

    The following tests cover both the challenging and extremely 
challenging operating ranges. See section 2.3.T for warm-up utility 
tests that can be conducted in tandem with these temperature stress 
tests.

1.5.1.1.T -40 [deg]C

    Devices A and B shall be temperature stabilized for a period of 
1 hr. in an environmental chamber set at -40 [deg]C. After the 
period of temperature stability elapses, the BAIIDs shall be 
subjected to an accuracy regimen as specified in section 1.1.2 T.

1.5.1.2.T -20 [deg]C

    Devices A and B shall be temperature stabilized for a period of 
1 hr. in an environmental chamber set at -20 [deg]C. After the 
period of temperature stability elapses, the BAIIDs shall be 
subjected to an accuracy regimen as specified in section 1.1.2 T.

1.5.1.3.T +70 [deg]C

    Devices A and B shall be temperature stabilized for a period of 
1 hr. in an environmental chamber set at +70 [deg]C. After the 
period of temperature stability elapses, the BAIIDs shall be 
subjected to an accuracy regimen as specified in section 1.1.2 T.

1.5.1.4.T +85 [deg]C

    Devices A and B shall be temperature stabilized for a period of 
1 hr. in an environmental chamber set at +85 [deg]C. After the 
period of temperature stability elapses, the BAIIDs shall be 
subjected to an accuracy regimen as specified in section 1.1.2 T.

1.5.2.T Extreme Conditions Beyond Manufacturers Claimed Accuracy

    If the BAIID manufacturer has chosen to meet the specifications 
for temperature extremes (-40 [deg]C and +85 [deg]C) by having the 
alcohol sensing unit be removable (e.g., so that it may be kept warm 
(cool) when the vehicle is expected to be subject to extremely cold 
(hot) temperatures), then the fixed or permanently installed portion 
of the BAIID only shall be exposed to the extreme temperature 
specification. Then, when the sampling head is reconnected to the 
device, the BAIID must meet the accuracy requirements as specified 
in paragraphs 1.1.1.S to 1.1.4.S when tested in accordance with 
paragraph 1.5.T. This testing shall be conducted promptly following 
reconnect so as not to allow the sensor to become equilibrated to 
the chamber temperature. Warming of the sensor is acceptable between 
trials if necessary to meet the specification.
    If the sampling head is not removable and the temperature range 
within which the BAIID is claimed to operate properly is narrower 
than that provided for in paragraph 1.5.1.S, then at the extreme 
temperatures outside the range specified by the manufacturer, the 
BAIID shall fail-safe.

1.6.S Vibration

    All BAIIDs shall meet the accuracy requirements specified in 
paragraphs 1.1.1.S to 1.1.4.S after they have been subjected to the 
vibration tests in accordance with paragraph 1.6.T.

1.6.T Vibration Stability Test

    These tests are performed to determine BAIID fitness for the 
automotive environment. If the BAIID consists of more than one 
module, it will be necessary to shake each module separately. Before 
testing, inspect housing thoroughly for cracks.

1.6.1.T Test 1

    Subject device A to simple harmonic motion having an amplitude 
of .38 mm (0.015 in.) [total excursion of 0.76 mm (0.030 in.)] 
applied initially at a frequency of 10 Hz and increased at a uniform 
rate to 30 Hz in 2.5 minutes, then decreased at a uniform rate to 10 
Hz in 2.5 minutes.

1.6.2.T Test 2

    Subject device B to simple harmonic motion having an amplitude 
of 0.19 mm (0.0075 in.) [total excursion of 0.38 mm (0.015 in)] 
applied initially at a frequency of 30 Hz and increased at a uniform 
rate to 60 Hz in 2.5 minutes, then decreased at a uniform rate to 30 
Hz in 2.5 minutes.

1.6.3.T Variations

    Perform the vibration tests as described in paragraphs 1.6.1.T 
and 1.6.2.T in each of three directions, namely in the directions 
parallel to both axes of the base and perpendicular to the plane of 
the base.

1.6.4.T

    Repeat the test protocol for accuracy as specified in 1.1.2.T 
for both BAIIDs. The BAIID shall meet the accuracy requirements as 
specified in section 1.1.2.S.

1.6.5.T

    After the vibration regimen, inspect both BAIIDs to identify any 
cracks in the exterior casing and failures in the tamper-proof 
points of interface with the automotive environment. If cracks or 
failures are identified, then the test unit fails. The manufacturer 
shall be allowed to submit subsequent devices for this test phase, 
but no more than 1 of 6 shall be allowed to fail this phase.

1.7.S Radio Frequency (Electromagnetic) Interference (RFI)

    Radio frequencies generated inside the vehicle have the 
potential to interrupt signal processing, or sample evaluation at 
the BAIID.
    The BAIID shall be accurate according to the specifications set 
forth in Section 1.1.2.S. and tested according to Section 1.1.2.T 
when exposed to radio frequencies generated by common in-vehicle 
appliances, such as CB radios or cellular telephones.
    It should be noted that full characterization of RFI 
susceptibility of BAIID is beyond the scope of this effort. The 
following protocol shall be implemented as a limited test for 
whether intentionally generated RFI interferes with BAIID 
performance.

1.7.T RFI Testing Protocol

    In an actual vehicle in which a BAIID is installed, the sampling 
head of the BAIID shall be connected to the alcohol-air delivery 
tube in preparation for testing according to the specifications as 
set forth in Section 1.1.2.T. The sampling head of the BAIID shall 
be positioned so that it is adjacent to (within 2 cm), but not 
touching, any BAIID electronics processing unit which is mounted 
inside the vehicle on or under the dashboard.
    The antenna of a transportable cellular telephone with an output 
power of not less than 3 watts shall be placed within 5 cm of the 
sampling head/box of the BAIID. A telephone number shall have been 
keyed into the cellular telephone. The alcohol sample shall be 
introduced into the BAIID concurrent with the issuance of a ``send'' 
signal to the telephone.
    During each cycle while the BAIID is evaluating the alcohol 
sample, and while the telephone continues to transmit, the antenna 
of the telephone shall be positioned in one of three orthogonal 
(i.e. 90[deg]) orientations in relation to the BAIID. All three 
orthogonal orientations shall be tested.
    In order to ensure the safety of the individual conducting the 
tests, these tests shall not be run more than six (6) minutes in any 
given one hour period (see American National Standard Safety Levels 
with Respect to Human Exposure to Radio Frequency Electromagnetic 
Fields, 300 kHz to 100 GHz, approved by the American National 
Standards Institute on July 30, 1982). Additionally, it is an 
appropriate rule of thumb for the test lab personnel to make sure 
their eyes (as well as the rest of their bodies) are kept at a 
distance of at least 30 cms. from the transmitting antenna during 
the tests.
    The performance of the BAIID shall be evaluated according to the 
criteria of 1.1.2.T. The performance of the data recorder shall be 
determined to accurately reflect the test results found on the user 
display of the BAIID.

1.8.S Tampering and Circumvention

    The BAIID must provide a method to detect two classes of misuse, 
tampering and circumvention.

1.8.1.S Tampering

    The BAIID must provide a secure method to detect and store the 
time and date of tampering attempts made by the following means:

1.8.1.1.S--interrupting the power source of the interlock device 
causing it to fail, or to fail to record ignition activity,
1.8.1.2.S--vehicle engine starts not preceded by a passed interlock 
test, except during the free restart interval as provided for in 
1.9.S.

    Information about unauthorized starts that are stored internally 
shall not be lost when the interlock device is disconnected from the 
vehicle battery.

[[Page 8052]]

1.8.2.S Circumvention

    The BAIID must be able to detect, or protect against, 
illegitimate air samples introduced to the sampling head. 
Illegitimate samples may be delivered from the following sources:

1.8.2.1.S--non-human delivery sources of air samples such as 
balloons or compressed air containers,
1.8.2.2.S--human sources of air samples that are altered through 
filtration or other means after leaving the mouth,
1.8.2.3.S--human sources of air samples provided by anyone other 
than the driver of the vehicle. This specification does not imply 
the BAIID be able to detect a unique breath signature, but to 
preclude curbside assistance to an impaired driver, the BAIID shall 
require that a second breath test be required once a vehicle has 
been underway for at least 5 minutes but not more than 30 minutes.

    The BAIID must detect or minimize these types of circumvention 
in accordance with the criteria as specified in paragraph 1.8.T.

1.8.T Tampering and Circumvention Tests

1.8.1.T Tampering

1.8.1.1.T Power Loss

    The BAIID shall be able to register any external (non-sealed) 
loss of power. Any attempt to disconnect the BAIID from the vehicle 
in which it is installed shall be recorded electronically. To 
conduct this test disconnect external 12 Volt DC power source to the 
Device A or B and determine that there is a record of power loss 
noted by the interlock device. This may be noted on a memory chip, 
or by another indicator which can be detected by the service 
technician.

1.8.1.2.T Circuit Tampering

    The BAIID shall be able to register any engine start (whether or 
not the ignition switch is turned ON) which occurs without passing 
the BrAC test. This test will require use of an installed BAIID. To 
conduct this test, it will be necessary to ``hotwire'' the engine. 
The procedure for doing this will vary with the type of engine. One 
example is to attach one end of a wire to the primary side of the 
ignition coil (coming from the distributor) and the other end to the 
vehicle battery's positive pole. Then short the appropriate 
terminals on the starter relay or starter motor to determine if the 
vehicle is able to be started. If the vehicle starts, shut it off 
and then repeat this test 3 times on either Device A or B.
    An interlock device ought to be capable of either preventing a 
vehicle from being successfully hotwired, or be capable of 
registering all such successfully completed bypasses of the 
interlock device. If the installed device fails to achieve either of 
these criteria and permits circuit tampering, then it fails this 
test phase.

1.8.2.T Circumvention

1.8.2.1.T Non-Human Samples

    The BAIID shall be capable of detecting or failing 80% of the 
non-human breath samples introduced through one of the following:
     Mylar balloon
     Rubber (toy) balloon
     Compressed air (aerosol can or other source)
    The balloons must be large enough to deliver the minimum volume 
requirement, 1.5 liters. The non-human circumvention test battery 
shall be conducted in accordance with section 1.1.T, except the 
sample introduced shall be alcohol-free air introduced through the 
three air sources identified above. These sources are exemplary and 
not necessarily the best or only sources suitable for this class of 
circumvention.
    The devices A and B shall each be subjected to this 
circumvention testing. The criterion of failure in this case is more 
than two passed tests out of a series of 10. This is not a test of 
accuracy of alcohol detection, but a test of how well the BAIID can 
detect air samples that deviate from a normal breath sample.

1.8.2.2.T Filtered Samples

    BAIIDs shall be capable of detecting or failing 80% of the 
filtered samples when filtered by either dry or wet filtering 
systems such as the following:
     Commercial cat litter, silica gel
     Heated water
     Approx. 4 ft. or 1.5 meter long Tygon tube (\3/8\'' 
i.d.)
    The filtered sample circumvention test battery shall be 
conducted on both devices A and B in accordance with section 
1.1.2.T. In this case all elements of the testing procedure as 
specified in 1.1.2.T shall be identical except that the sample shall 
be filtered by interposing two different filtering systems, in 
separate tests, between the sample simulator and the interlock 
device. The dry filter can be composed of any tube packed with a 
suitable absorbent material, such as those identified above, but in 
doing so, the technician must keep in mind the constraints of 
absorbent capacity and the relationship between packing and 
blowability. For example, a 2\1/2\ inch piece of cardboard tubing 
(\3/4\ inch diameter) might be used. It might be packed with 12 
ounces of commercial cat litter, each end of the tube being stopped 
with cotton wadding. The wet filter shall ideally consist of water 
heated to 34 [deg]C in a capped cup fitted with inlet and outlet 
hoses. The filter device shall be made of common materials that are 
widely available. For example, a 6 oz. styrofoam coffee cup might be 
used with \1/4\ inch rubber or Tygon tubing used for inlet and 
outlet hoses. In the case of use of the 4 ft. long Tygon tubing as a 
filter, the tube shall be chilled to 0 [deg]C and attached securely 
to the BAIID mouthpiece before attempting to provide a sample.

1.8.2.3.T Rolling Retest To Thwart Curbside Assistance

    After passing the test allowing the engine to start, the BAIID 
shall require a second test within a randomly variable interval 
ranging from 5 to 30 minutes. During the rolling retest, the retest 
setpoint shall be .02% w/v higher than the startup setpoint to 
preclude a false positive test result.
    In order to alert the driver that a retest is to be required, a 
3 minute warning light and/or tone shall come on. The driver would 
then have 3 minutes to retest. If the engine is intentionally or 
accidentally shutdown after the 3 min. warning but before retesting, 
the retest clock shall not be reset. Retesting takes priority over 
free restarts (see Sect. 1.9). Test that the free restart is not 
operative when the BAIID is awaiting a rolling retest sample.
    The consequences of a failure to take the retest, shall be 
threefold. First, the refusal to perform a rolling retest shall be 
flagged and recorded on the data recorder. Second, the BAIID shall 
warn the driver by a unique auditory or visual cue that the vehicle 
ignition will enter a lockout condition within a period of 5 days, 
and that the assignee shall report to the BAIID program monitor 
promptly. Third, the lockout shall proceed within 5 days.
    A retest that is taken as required and subsequently failed shall 
result in an alert condition that is flagged on the data recorder. 
The BAIID assignee shall be signalled that the BAIID program monitor 
must be notified promptly of the violation, the automatic lockout 
shall proceed.
    The test protocol shall determine that both devices A and B are 
capable of performing according to this specification.

1.9.S Sample-Free Restart

    After a stall, a sample-free restart shall be possible for 2 
minutes. This free restart does not apply, however, if the BAIID was 
awaiting a rolling retest that was not delivered.

1.9.T Sample Free Restart Test

    The BAIID shall permit a free restart (no breath sample 
required) for 2  .25 min. Conduct six tests with an 
alcohol-free sample from either a human or non-human source. Three 
tests at 1.5 min, three at 2.5 min. Use devices A and B. The BAIIDs 
shall allow a start without requiring a sample for all of the first 
three tests, and fail to start without a sample on the subsequent 
three tests.

1.10.S Data Recording

    An active monitoring program will require vehicle use 
information. A BAIID shall have the capability to record the nature 
of such use and the test outcomes during the stipulated period. The 
following kinds of information shall be recorded by the BAIID:
     Efforts to disable the unit
     Date of vehicle use
     Time of vehicle use
     Pass/fail records
     BrAC levels
     Starting and stopping of vehicle engine
     Service reminders issued (date)
     Date service performed

1.10.T Data Recording Test

    Perform test according to manufacturer's instructions. Determine 
whether readout is satisfactory and understandable. Test to be 
certain that the BAIID memory remains intact for multiple printouts 
if desired, or until the service technician chooses to reset/erase 
the memory.

2.0.S/T Utility Specifications (S) and Utility Tests (T)

2.1.S Dual Accuracy and Precision Limits (Low End)

    The accuracy and precision for the utility specification shall 
be determined in a

[[Page 8053]]

manner parallel to that described in paragraphs 1.1.1.S to 1.1.4.S 
except for the test solution of alcohol to be used in the simulator. 
In the case of the utility specification, as with the safety 
specification, there is a dual criterion depending on the existence 
of stress test protocols. No stress test protocols are specifically 
provided for here in conjunction with utility specifications, since 
these are not strictly highway safety question. Certifying 
authorities wishing to conduct stress-involved protocols for the 
utility specification could conduct them in a parallel fashion to 
those provided for and beginning in Section 1.3. Nonetheless, a 
parallel dual set of specifications is proposed here for States 
wishing to conduct such testing.

2.1.1.S Baseline Accuracy in the Unstressed Condition

    All BAIIDs shall allow the ignition to remain locked no more 
than 10% of the time when the true alcohol content of the breath 
sample is 0.01% or more below the alcohol setpoint and testing is 
being conducted under ambient temperatures in the range of 10-30 
[deg]C in a newly recalibrated BAIID.

2.1.2.S Accuracy Under Stress Conditions

    Under conditions of stress testing, the BAIIDs shall allow the 
ignition to remain locked no more than 10% of the time when the true 
alcohol content of the breath sample is 0.02% w/v or more below the 
alcohol setpoint.

2.1.3.S Standard Deviation (Precision)

    Precision guidelines shall be parallel to those described in 
Section 1.1.3.S.

2.1.4.S Proportions

    This is to specify the proportion of tests at BrACs of .01% w/v 
and .02% w/v below the alcohol setpoint at which the ignition must 
be unlocked. The table below shows the 90% criteria of accuracy for 
unstressed and post-stress testing.

   Table 2.--Test BrAC Level at Which the Ignition Must Be UnLocked at
    Least 90% of the Time Depending on Whether Test Is Unstressed or
                                Stressed
------------------------------------------------------------------------
                                              Test BrAC level (% w/v)
            Alcohol setpoint             -------------------------------
                                            Unstressed       Stressed
------------------------------------------------------------------------
0.025% w/v*.............................           0.015           0.005
------------------------------------------------------------------------
* Recommended.

2.1.T Testing of Utility Specification (Dual Criteria)

    All utility tests shall be conducted on the two BAIIDs, devices 
A and B. Two sets of specifications can apply, but only one of these 
specifications, the baseline or unstressed protocol (2.1.1.T) is 
specifically utilized.

2.1.1.T Utility Accuracy Testing of Unstressed BAIID

    The accuracy testing is conducted as a measure of the BAIID's 
ability to hold to or exceed a 90% accuracy criterion when a test 
solution is .01% w/v below the alcohol setpoint. This test shall be 
conducted at room temperature and precede all other utility tests to 
ensure that the fundamental operation of the BAIID is adequate under 
no-stress conditions after recent recalibration.
    The test shall be repeated 20 times on device A, and 20 times of 
device B. Two types of results shall be recorded, pass/fail, and a 
digital readout representing the BAIID's evaluation of the alcohol 
concentration of the introduced sample.
    If either BAIID locks more than twice in those twenty trials 
then it has failed the no-stress accuracy utility test criterion of 
90%.
    A failure to meet the accuracy criterion shall disqualify the 
BAIID.

2.1.2.T Utility Accuracy Testing of Stressed BAIIDs

    If the certifying authority chooses to conduct tests of the 
utility specification for stressed BAIIDs, it is recommended that a 
protocol be followed that parallels those proposed for Stressed 
BAIIDs beginning in Section 1.3, and that the criteria for 
evaluation be .02% w/v below the setpoint for 90% unlocked accuracy.

2.2.S Clearance Rates

    The BAIID shall permit a test within 3 minutes of a previous 
test at a BrAC < .05% w/v.

2.2.T Clearance Rate Test

    The BAIID shall reset to zero and be ready for a retest within 3 
minutes of a previous test at BrAC = .05% w/v.
    Test adherence to this criterion by introducing a .05% w/v 
sample into devices A and B, activate a timer upon receipt of the 
test result, record the test result. Record the elapsed time before 
the BAIID indicates a ``ready'' condition. Repeat this three times 
for each BAIID.

2.3.S Warm Up

    The BAIID shall be ready for operation within 5 minutes of being 
turned on at -20 [deg]C (-4 [deg]F).

2.3.T Warm Up Test

    The warm up period during which the BAIID heats the sensing head 
shall require no more than 5 min at -20 [deg]C (-4 [deg]F).
    This test can be conducted as part of the environmental chamber 
tests specified in section 1.5. After stabilization in the 
environmental chamber at -20 [deg]C for 4 hr. activate timer 
concurrent with activation of the BAIID. Record the time required 
before receiving a ``ready'' condition.

2.4.S User's Display

    The BAIID shall provide certain types of informational feedback 
to the driver. These messages include: BAIID readiness for sample, 
test outcome, and warning messages.

2.4.T User Display Tests

2.4.1.T Operational Modes

    Indicators must be plainly visible or clearly audible to the 
user denoting the following:
     Unit is ON
     Unit is READY FOR TEST
     Unit has RECEIVED ACCEPTABLE SAMPLE

2.4.2.T Outcome

    Unit must plainly indicate the test results with a minimum 
message of:
     PASS or FAIL

2.4.3.T Warnings

     UNIT must be SERVICED and CALIBRATED SOON

2.5.S Temperature Package

    To reach conformance with temperatures below -20 [deg]C or above 
+70 [deg]C, the manufacturer may make available a mechanism or 
procedure that can achieve the warm-up (cool-down) needs. This can 
be accomplished via removal of the sampling head from the vehicle 
for bringing inside the home, or via provision of a heating jacket, 
or other procedures.

2.5.T Low Temperature Package Tests

    Evaluate manufacturers' proposed procedure for temperatures as 
low as -40 [deg]C.

2.6.S Altitude

    The manufacturer shall place a notice in the BAIID manual and on 
the device noting that the alcohol sensing unit is more sensitive to 
ethanol at higher altitudes, and that attempts to start at altitudes 
higher than that for which the BAIID is calibrated could result in a 
lockout even when the BrAC is lower than the alcohol setpoint.

2.6.T Altitude Test

    The BAIID must provide some written notice to the user of the 
possibility of a lockout at higher altitudes if it is unable to 
maintain accuracy at ground elevations up to 2.5 km.

3.0.S/T Optional Features Specifications (S) and Optional Features 
Tests (T)

3.1.S Optional BrAC Display

    Knowledge of the relation between drinking and BrAC can be a 
useful educational tool for motivated users. Therefore it is 
suggested that states give consideration to whether a BAIID give a 
BrAC readout to the user--in addition to a mere pass/fail 
indication--after a test.

[[Page 8054]]

3.1.T Optional BrAC display

    Evaluate the adequacy of the display indicator which informs the 
user of the BrAC test result.

3.2.S Optional Sample Acceptability Criteria at Inlet

    To improve circumvention protection, sample evaluation criteria 
as specified in 3.2.1.S and/or 3.2.2.S may be required. These 
criteria are noted as optional at this time, but may be necessary in 
order to eliminate the most commonly identified methods of 
circumvention. Further discussion can be found in Sec. 6.2.

3.2.1.S Optional Temperature Window of Sample

    Imposing a criterion requiring the sample to fall in a range 
between 32-48 [deg]C will improve rejection of bogus samples at 
neutral ambient temperatures. Other criteria may need to apply, 
however, when air temperatures fall outside the neutral range.

3.2.2.S Optional Minimal Pressure of Sample

    Filtered samples may suffer pressure losses. A minimal pressure 
requirement of 12 inches of water will help screen out filtered 
samples.

3.2.T Optional Sample Acceptability Criteria Test

    These optional features, if adopted, will have been tested in 
tandem with the circumvention test protocols in paragraphs 1.8.2.T. 
If the acceptability criteria are incorporated into the design of 
the BAIID, it is expected that fewer bogus air samples will have 
resulted in a pass condition.

3.3.S Optional Smoke Protection

    Tobacco smoke is known to produce false positive results on 
semiconductor type interlock devices. Smoke from burning fields, a 
common seasonal event in some rural areas, may similarly be a source 
of error. Protection of the sampling head from ambient smoke 
conditions may be necessary under some conditions.

3.3.T Optional Smoke Protection Test

    To evaluate the potential of air borne smoke to interfere with 
the accurate sensing of alcohol, perform testing according to 
paragraph 1.1.T and/or 2.1.T (depending on the testing authority's 
interest in safety or utility concerns), in a chamber filled with 
smoke from burning vegetal substances or similar conditions.

3.4.S Optional Dust Protection

    Fine dust can cause problems with electronic equipment by 
forming conductive bridges. However, of even greater concern with 
the interlock device is the ability of fine dust to absorb vapors. 
This is a specification that may be of concern in arid regions, or 
where there will be BAIIDs installed in construction vehicles. 
States subject to dust conditions may want to require some kind of a 
housing that protects the BAIID sampling head from exposure to 
powdery dust. Dust protection is incorporated in the Australian 
Standard for BAIIDs.

3.4.T Optional Dust Protection Test

    If a test for dust protection is required by a state, the 
certification authority may want to follow the clearly specified 
test procedure in the Society of Automotive Engineers Recommended 
Environmental Practices For Electronic Equipment Design--J1211, page 
20.122, Sect. 4.5.

3.5.S Optional CB Radio Alert Condition

    Under conditions of a failure to take the required rolling 
retest, or a failure to pass a rolling retest (as provided for in 
paragraph 1.8.2.3.T), a signal could be transmitted over a 
restricted CB channel that can be monitored by the police which 
alerts nearby cruisers that an impaired driver is operating a motor 
vehicle. This optional feature can be regarded as support for the 
anti-circumvention feature as described in paragraphs 1.8.2.3.S and 
1.8.2.3.T.

3.5.T Optional Alert Conditions Test

    No test protocol is proposed.

4.0 Commentary on Safety Specifications

    These specifications have been divided into safety and utility 
specifications. This distinction has been made in the Definitions 
Section D8. Safety issues are by far the more important and the 
majority of the testing is devoted to insuring that BAIIDs perform 
as expected under conditions of normal field use. It is expected 
that normal field use will involve a wide range of driving and 
outdoor conditions, as well as having a minimum of 5% of users 
trying to circumvent or tamper with the BAIID in order to drive 
while impaired.
    The ethanol sensing technology that has been adapted to the 
automotive environment for BAIID devices is mostly based on the 
Tagucci semiconductor device. The semiconductor devices are not as 
specific or stable as evidential field use breath testers. However, 
the purpose of the BAIID is not to accurately measure in mg/ml the 
BAC of a driver, but to prevent the person with a high BAC from 
operating a motor vehicle. For this reason, the specification has 
allowed greater leeway in the accuracy test criteria, but has also 
included a protocol for circumvention protection. In the associated 
technical report strong recommendations are made for a central 
authority within each State to maintain authoritative programmatic 
control of the BAIID option.

4.1 Accuracy

    With respect to accuracy, these specifications establish a range 
of acceptable performance, especially under so-called ``stress'' 
conditions such as temperature extremes, vibration, power 
variability, etc. For this reason a ``double standard'' is proposed 
which is conditional on the recent stress exposure of a test unit. 
The reasoning for this is as follows.
    First, a newly recalibrated BAIID that is not subjected to 
stress tests ought to be held to a higher standard than one which 
has been so subjected. Field experience with the installed units 
using semiconductor technology has shown that there is considerable 
average error (in the range up to 0.015% w/v) following 60 days of 
routine field use of a BAIID.
    These specifications do not provide for accuracy testing under 
compound stresses, such as low temperature with low power at high 
altitude. Rather than proposing tests for compound stresses to 
accuracy here, the requirement for such tests should rest with the 
certifying authorities of the States who can best determine their 
unique situation evaluation requirements. Clearly, northern Rocky 
Mountain States would be more interested in combined high altitude 
and low temperature tests than would States in the southeast. 
Similarly, many questions have not been researched which may prove 
significant. For example, would a BAIID calibrated for use at high 
elevation be able to meet the accuracy specification when tested at 
the coldest temperatures at sea level? These questions are too 
specific for inclusion in national guidelines, but may be important 
regionally.
    When measuring accuracy and precision of any instrument it 
should be understood that all measuring devices have a certain 
natural amount of dispersion of scores around a mean (average) true 
value. Because of this fluctuation, the setpoint of the interlock 
device needs to be clearly specified in a way that accommodates this 
natural variability. In this specification, the worst acceptable 
deviation under conditions of perfect accuracy have been identified. 
This allows for inaccuracy and imprecision to trade-off as long as 
the overall probability of error is lower than the constant 
specified.
    The proposed specifications for interlock devices ostensibly 
acknowledge three lock points:
     The alcohol setpoint (the nominal lock),
     The virtual lock (90% certainty),
     The near absolute lock (99.5% certainty).
    The alcohol setpoint is defined as the interlock device-measured 
BrAC value at which the ignition will lock.\1\ That is, the alcohol 
setpoint is the BrAC value at which the interlock is set. Due to the 
inherent variability in these measuring devices, this nominal 
lockpoint will be the mean of a distribution of true blood or breath 
alcohol concentration values as determined by evidentiary BrAC 
equipment. Interlock imprecision is the deviation from that value. 
The higher the precision of the interlock, the smaller will be the 
dispersion of true BrAC values around the stipulated alcohol 
setpoint.
---------------------------------------------------------------------------

    \1\ This standard recommends that .025% w/v be chosen as the 
setpoint.
---------------------------------------------------------------------------

    The virtual lock point will be the actual, or true BrAC above 
which the vehicle must fail to start 90% of the time. The difference 
between the setpoint and virtual lock values will be a gray area 
which reflects both imprecision and inaccuracy. The guideline 
specifies that there should be a maximum permissible standard 
deviation from the setpoint equal to 0.0078% w/v BrAC under 
conditions of no-stress. Following stress protocols, the maximum 
permissible standard deviation under conditions of perfect accuracy 
is equal to .0156% w/v.
    The third type of lockpoint is the near absolute lock point and 
is of theoretical interest only because many hundreds of repetitions 
would be needed to test it. The

[[Page 8055]]

near absolute lockpoint is equivalent to +2.57 standard deviations 
in a normally distributed sample of trials where 99.5%, practically 
all, start attempts must fail. In the unstressed condition, this 
would be .02% w/v above the setpoint and .04% w/v above the setpoint 
in the stressed conditions. The implication of this is that for 
devices which are tested against the specification (even with its 
most lax accuracy standard), a person with a BAC equal to .065% w/
v--still well below the legal limit of most States--would almost 
certainly be locked out.
    Since the condition of virtual lock is defined operationally as 
1.28 standard deviations above the alcohol setpoint, and the 
absolute lockpoint is 2.57 standard deviations above the setpoint, a 
brief explanation of standard deviation (sd) is relevant.
    Standard Deviation--The sta