Schaffer Online Library of Drug Policy Sign the Resolution
Contents | Feedback | Search
DRCNet Home
| Join DRCNet
DRCNet Library | Schaffer Library | Drugs and Driving 


Development of a System for Real-Time Breath Alcohol Analysis

C.M. Bell and H.J. Flack

Victoria Forensic Science Centre, Forensic Drive, Macleod, Victoria, 3085, Australia


A system was developed for real-time measurement of 4 variables relevant to breath alcohol analysis, using a Dräger Alcotest 7110 Mk IV analyser as the base instrument for the system. The 4 variables measured were blood alcohol concentration (BAC), volume, temperature and pressure of the sample delivered. The accuracy of the measurement of each variable was ±0.004% BAC, 10% of the reading, 0.2°C and 0.5 cm H2O respectively, within the measurement range of interest. Each variable was sampled at a rate of 4 times per second. Some limitations with the system were observed. Real-time data collection was performed by personal computer and data associated with a sample delivered linked automatically. The system provides a usable research tool for study of the variables associated with breath alcohol analysis.


Modern breath alcohol analysers only provide a result of analysis of the final portion of breath sample, however, some have the capability of providing readings in real-time (i.e. while the sample is being delivered). Instruments based upon the measurement principle of absorption of infra-red (IR) radiation already utilise real-time monitoring to detect mouth alcohol (ALCOMAT® Operating Instructions). Some instruments can be specially modified to provide the readings from real-time analysis and have been used by other workers (Gullberg, 1990). Other variables relevant to breath alcohol analysis, such as the volume, temperature and pressure of sample delivery, can also be measured in a similar fashion. This paper is a description of a system developed for real-time measurement of all the 4 abovementioned variables, as well as discussion of its performance and the items required to create a workable system.


Two Dräger Alcotest 7110 Mk IV breath alcohol analysers (based upon IR absorption at 9.5 µm) were fitted with customised software to provide real-time measurement of BAC and volume at a rate of 4 times per second. The software (version FSL 1.00) was purchased from and fitted by Dräger Australia. Real-time measurement of temperature and pressure was achieved using two specially made assemblies purchased from Elpro Designs, Moorabbin, Victoria. They included an attachment that fitted into the Alcotest breath inlet hose (between the mouthpiece and the hose) fitted with a thermistor for temperature measurement (Bowthorpe glazed bead type, of low thermal mass and rapid response time, bead diameter 0.5 mm) and a fine bore flexible polyethylene connecting tube running to a pressure transducer (Honeywell diaphragm / piezo resistor type, gauge pressure 0-5 psi), situated in a module containing the electronics associated with this assembly.

The data output from each Alcotest instrument was routed through an Elpro assembly. The data from the linked Alcotest and Elpro devices was collected in one operation into a personal computer using Telix v3.11 as a communications link. Other data relevant to the sample being delivered were also entered onto Telix and saved with the real-time download using a customised script program written for this purpose. The data was downloaded and saved as an ASCII file and could readily be imported into a spreadsheet package (eg. Lotus 1-2-3) as required.

The 4 measurements performed by each system were calibrated according to the methods outlined in Table 1.

Table 1
Methods of Calibration

BAC Wet simulation (at 34.0°C) with 0 and 0.15%BAC, using the standard Alcotest IR CAL routine.
Volume Delivery of a known flowrate of 14L/min of saturated air at 34.0°C, using the standard Alcotest FLOW CAL routine.
Pressure By generation of a static pressure of 10 cm H2O in a blind system with reference to a water manometer.
Temperature At a static temperature of 34.0°C, in an oven, with reference to a calibrated and certified thermometer.


An example of data download collected by the computer is given in Table 2. The first time increment is usually smaller in that sample delivery is more likely to begin part way through a cycle of the IR signal. The time increment thereafter is either 0.25 or 0.26 seconds. At the end of sample delivery a final BAC and volume result are recorded. These results indicate what an Alcotest fitted with standard operating software would report and may include messages, such as "Alcohol in Mouth" or "Insufficient Sample". The time taken to report the final BAC and volume results is approximately 10 seconds from the end of the real-time data download.

Table 2
Raw Data Download

Time (seconds) BAC (%) Volume (L) Temperature (°C) Pressure (cm H2O)
0.00 0.000 0.003 30.4 3.0
0.06 0.001 0.007 30.8 3.5
0.31 0.002 0.030 30.8 3.0
0.57 0.450 0.061 31.1 4.0
etc. ... ... ... ...
14.90 0.151 1.554 31.5 0.0
15.16 0.152 1.577 24.7 -0.5
24.88 0.151 1.577    

The performance of each measurement carried out by each system, excepting volume, was determined over a range of readings in a similar fashion to that described in Table 1. For volume, saturated air at 34.0°C was delivered using a piston-cylinder type air delivery apparatus specially custom-built and calibrated for this purpose, delivering known volumes up to 5 L.

The accuracy of the BAC measurement was as expected for a Dräger Alcotest 7110 device - typically ±0.004 %BAC for readings up to 0.150 %BAC. These devices display lower readings than expected at higher BAC. The BAC readings for the real-time download, towards the end of sample delivery, were comparable to the final result normally produced by the Alcotest when fitted with standard software. The differences between the real-time BAC values and the final result produced by the Alcotest is that the final result is a result obtained from averaging over 8 seconds of analysis time (i.e. equivalent to 32 real-time readings) and that the zero setting (established upon the initial purge with room air) is subtracted. Any contamination of room air with alcohol therefore will cause a final result to be lower than the real-time BAC values obtained towards the end of a sample delivery. One other factor that may cause a difference between these two values is that the sample expands slightly upon the cessation of delivery with the reduction of gauge pressure from typically 15 to 0 cm H2O.

One consideration for BAC measurement accuracy when using the system for testing of human subjects is the effect of carbon dioxide upon the reading. The selected IR transmission window for the filters fitted in the Alcotest 7110 is centred at 9.5 µm. Carbon dioxide shows an absorption at this wavelength (Banwell, 1966). Other Alcotest 7110 instruments have shown a linear response to CO2. Physiological CO2 levels are typically 3-5% in breath (Best and Taylor, 1943), therefore a contribution to a reading would be approximately 0.002% BAC. Dräger Australia have indicated that a CO2 correction is also applied as part of obtaining the final result reported (i.e after the 8 seconds of analysis time).

The accuracy of the volume measurement was better than ±10% and linearity was displayed over the range tested (0-5 litres). Some variation in the results obtained was attributable to the possibility of air leakage from the air delivery apparatus. Proper preparation of the air delivery apparatus was important to obtain reproducible results. The temperature and pressure measurements displayed accuracy of ±0.2°C and 0.5 cm H2O respectively. Greater temperature variations were observed at extremes of the measurement range (i.e. 15 and 40°C) but were of no consequence for practical use. Pressure measurement was linear over the range tested (0-30 cm H2O).

The speed of response of each measurement was difficult to determine. Some attempts were made to provide simulated BAC's that changed in a stepped fashion, however no useful results were obtained. Several factors do, however, indicate that the speed of response of some measurements in the system is fast. The thermistor used was of very low thermal mass and would therefore react quickly to changes in temperature. The volume reported in real-time was comparable to the volume delivered when part-way through sample delivery. The time for a simulated BAC to attain the expected reading did not greatly exceed the calculated delays caused by dead volume of the breath hose and sample chamber of the Alcotest. Finally, plots of BAC and temperature versus time for simulated samples displayed a rapid attainment of plateau values.

The temperature and pressure measuring attachment fitted to the breath inlet tube was unheated. When in use, it was partially heated by the internally-heated breath inlet tube of the Alcotest. The temperature of the attachment just prior to initiating a test was generally 28-30 °C. Condensation in the attachment was monitored, yet did not occur, nor cause any problems with testing. Schoknecht (1993) describes a similar attachment for breath temperature measurement that was heated, and also heated the mouthpiece, to avoid condensation. Such an inclusion in the attachment is worthwhile as then the attachment may be either heated or unheated, depending upon the requirement for the particular experiment.


The system developed and reported here provides a usable research tool that enables the collection of the BAC, volume, temperature and pressure of a breath sample over the delivery. It may be used to study the variables associated with breath alcohol analysis to an extent that has not previously been possible. It is semi-portable and has the convenience of semi-automation of the data collection and storage. Relatively large quantites of data are collected and stored, and may be processed easily using standard software for personal computers to produce profiles of breath test variables over the duration of the delivery of a sample.


The development of the real-time measurement system was undertaken with a grant from the National Drug Crime Prevention Fund. The basic Alcotest 7110 instruments were kindly supplied by the Victoria Police Traffic Alcohol Section from a recent purchase of such instruments by the Transport Accident Commission. The assistance of Mr. Nick Farrell in writing of the script in Telix and processing some of the data generated was greatly appreciated, as was the production of the air delivery apparatus by Mr. Mario Flack.


ALCOMAT® Operating Instructions, Siemens Ltd., p16.

Banwell, C., "Fundamentals of Molecular Spectroscopy", (1966), Publishers McGraw-Hill, p83.

Best and Taylor, "Physiological Basis of Medical Practice", 3rd Edition, (1943), Publishers Williams and Wilkins, Baltimore, p523.

Gullberg, R.G., "The mathematical analysis of breath alcohol profiles generated during breath exhalation", J. Anal. Toxicol. (Nov/Dec, 1990), v14, p358.

Schoknecht, G., "The influence of temperature on breath-alcohol analysis", Alcohol, Drugs and Traffic Safety - T92, (1993), Publishers Verlag TUV Rheinland, Cologne, p392.