Breathalyzer Software Revealed

Posted by Lawrence Taylor on September 4th, 2007

As I’ve indicated in previous posts, defense attorneys for years have been trying to discover the software source code used by manufacturers of various breathalyzer models.  (See "Secret Breathalyzer Software Still Secret")   The accuracy of these  machines, which essentially determine a suspect’s guilt or innocence, depends upon the accuracy of the software driving them; as the computer techs say, "Garbage in, garbage out".  But the manufacturers have refused to produce the information, relying upon a claim of "trade secrets" — that is, that the code of each model is a unique creation of the manufacturer.  And prosecutors, apparently more concerned with profits than with justice, have joined them in resisting disclosure.

Smart TV BOX

Recently, however, judges in Florida, Minnesota, New Jersey and a growing number of other states have begun ordering the manufacturers to reveal the inner workings of their machines to the defense.  (See "Judge: Divulge Breathalyzer Code…or Else".)  Not surprisingly, the manufacturers have refused to comply.  Until a few days ago….

New Jersey attorney Evan M. Levow was finally able to get an order from the Supreme Court of New Jersey forcing the manufacturer of the popular Draeger AlcoTest 7110 to reveal the source code.  Levow turned the code over to experts, Base One Technologies, to anaylze.

Initially, Base One found that, contrary to Draeger’s protestations that the code was proprietary, the code consisted mostly of general algorithms:  "That is, the code is not really unique or proprietary."  In other words, the "trade secrets" claim which manufacturers were hiding behind was completely without merit.

Some of the more interesting excerpts from the Base One report:

 1. The Alcotest Software Would Not Pass U.S. Industry Standards for Software Development and Testing: The program presented shows ample evidence of incomplete design, incomplete verification of design, and incomplete "white box" and "black box" testing. Therefore the software has to be considered unreliable and untested, and in several cases it does not meet stated requirements. The planning and documentation of the design is haphazard. Sections of the original code and modified code show evidence of using an experimental approach to coding, or use what is best described as the "trial and error" method. Several sections are marked as "temporary, for now". Other sections were added to existing modules or inserted in a code stream, leading to a patchwork design and coding style…

It is clear that, as submitted, the Alcotest software would not pass development standards and testing for the U.S. Government or Military. It would fail software standards for the Federal Aviation Administration (FAA) and Food and Drug Administration (FDA), as well as commercial standards used in devices for public safety…If the FAA imposed mandatory alcohol testing for all commercial pilots, the Alcotest would be rejected based upon the FAA safety and software standards…

4. Catastrophic Error Detection Is Disabled: An interrupt that detects that the microprocessor is trying to execute an illegal instruction is disabled, meaning that the Alcotest software could appear to run correctly while executing wild branches or invalid code for a period of time. Other interrupts ignored are the Computer Operating Property (a watchdog timer), and the Software Interrupt.

6. Diagnostics Adjust/Substitute Data Readings: The diagnostic routines for the Analog to Digital (A/D) Converters will substitute arbitrary, favorable readings for the measured device if the measurement is out of range, either too high or too low. The values will be forced to a high or low limit, respectively. This error condition is suppressed unless it occurs frequently enough…

7. Flow Measurements Adjusted/Substituted: The software takes an airflow measurement at power-up, and presumes this value is the "zero line" or baseline measurement for subsequent calculations. No quality check or reasonableness test is done on this measurement…

10. Error Detection Logic: The software design detects measurement errors, but ignores these errors unless they occur a consecutive total number of times. For example, in the airflow measuring logic, if a flow measurement is above the prescribed maximum value, it is called an error, but this error must occur 32 consecutive times for the error to be handled and displayed. This means that the error could occur 31 times, then appear within range once, then appear 31 times, etc., and never be reported…

Based upon a .08% reading from this machine, American citizens are accused of drunk driving and, in court, presumed by law to be guilty.



  • chuanfa

    So these devices have such a wide margin of error that they shouldn’t be used and every case in which the results of one of these devices were used as evidence could be overturned? This makes one wonder how many people that have been given a DUI are innocent and how many DUI’s got away due the these margins of error. I have never gotten a DUI and I find this a great injustice. The companies that produce these faulty devices should be held accountable. That police depts around the country have used these devices without testing or investigation should be punished as well. How many civilians have had to pay for this the hard way?

  • markus_b

    I don’t understand than police in the US can get away with a simple breathalyser test for a DUI conviction. These things are just not reliable and precise enough.

    Over here in Switzerland police are using breathalysers too, but only as a first test. If you fail the breathalyser they take a blood test and only the blood test is relevant for a DUI conviction.


  • azamor2

    I have been accused of an Extreme DUI in Phoenix, Arizona. I believe the information contained in this report would aid my in my defense. Dose anyone knoe where I can get a copy of this report?

  • jim

    Good luck in the hearings! I hope we prevail with all this!

    I know you are doing great work on this case!

    As well as all the defense attorneys involved.

    You know what I want too, and this is in the hole, hidden away. A ace in the hole that will eventually come to light, someday, especially in New Jersey.

    It is OK that it is hidden, because it gives everyone a 2nd shot to crack Drager!

    I want an analytical instrument such as what Perkin Elmer, Algilent, Thermo Orion, Hitachi, or Waters, would make. An analytical instrument that can stand up to FDA scrutiny for hardware and softeware and FDA requirements for data handling.

    It is not something out of the norm I am looking for.

    Analytical instruments in companies such as Merck, Johnson and Johnson, Schering Plough, Bristol Myer Squib, and 1000’s of other Pharma companies and Over the Counter manufacturers in New Jersey , use everyday, millions of individual pieces of equipment, that meet the requirements I am talking about to produce a “legal ” result, as far as the FDA is concerned!

    I have a lot of respect for the FDA , I having been in the analytical business for 19 years!

    The FDA are real sticklers, if you will, when it come to data!

  • jim

    The following is what the real argument should have been against the Alcotest.

    A very important aspect of the Alcost breath testing instrument is the area under the curve, the peak area of the peak produced by alcohol. I am not sure if they are measuring the “oh” group peak or the “ethyl peak”, but for this argument it doesn’t matter.

    As an analytical chemist with 20 years experience, this visual picture/evidence/document, is just about the most important document there is.

    The picture of the test.

    Picture a straight line across a page of paper 8 inches in width. picture about 6 inches down the line a tall thin triangle rising about 6 inches from the line (baseline), symmetrical in shape, and about 3 millimeters in width at the baseline.

    This is what I call a peak a classic desired peak.

    The amount of area inside the shape of the peak is calculated by a computer program and converted into a number called “counts”.

    What the program does is this, it through its programing, determines when the peak starts and stops. draws a baseline under the peak, then calculates the area in counts , inside the triangle.

    This count of a breath sample is compared to the count of a known standard solution, a mathematical calculation is performed, and a final BAC is determined.

    This is fundamental , most basic, analytical chemistry, for peak based instruments, like an IR in the Alcotest.

    In my career I have revived , looked at, millions and millions of peaks generated from such instruments to determine if the start and stop of the peak was determined correctly, if the baseline was determined correctly, if the peak shape was ok, if the right peak was being used in the calculation, and if there were any interferences.

    Without this picture, it is impossible for Drager, me, you, the state, the FDA, or anyone else in the world, to say the result is accurate.

    In the millions of peaks I have revived in 20 years, I have found a lot of problems along the way.

    The start was determined wrong, the end was determined wrong, the baseline was drawn incorrectly, the wrong peak was used to calculate.

    An instrument can be running fine for 100 samples in a row, have one off, and 100 more good samples. It is just the nature of analytical instruments for this type of analysis, and every peak needs to be looked at and signed off on.

    With the Alcotest there is no picture. We have absolutely no way of knowing if anything looked ok in a breath test.

    Do you understand what I am saying?

    It is not like I am looking for the world here, a simple IR machine that produces a picture is not only common, it is expected, required, taken for granted.

    The cost of the Alcotest is ridiculously high for an instrument that can not even produce a picture of the test.

    To take this one step further, the “counts” for each standard and sample would be listed on the screen and printout, and an example calculation would be typically performed for the first sample by hand,vto ensure the program was calculating correctly, and available to perform for every sample tested.

    All of this is a no brainier.

    Do you understand?

    The process that calculates the area under the curve is called integration.

    Unless their is a visible hardcopy picture of the peak showing the start, stop, and baseline of the integration…. showing the peak, at the proper , in the case, wavelength, but in HPLC retention time, and it is reviewed to determine the peak shape was ok, the correct peak was used, there are no interferences, and the integration had the right start-stop-baseline, there is absolutely no way to say beyond a reasonable doubt any breath test result is correct.

    A key aspect in the Alcotest case are the manuals supplied by Drager to the Chun defense teaM, which one of the attorneys, not my former one, was kind enough to supply me with last summer for review….which I did review….

    In these manuals are example pictures of the peaks produced by an IR scan for alcohol.

    The methodology clearly shows a baseline and peak for the analysis , as I described earlier.

    My question to Drager would have been, how do you know your instrument, the Alcotest, produces these peaks, integrates them correctly, every time, 100% without flaw, drawing a perfect baseline to the peak every time, with a perfect start and stop tic marks, and thus perfect integration.

    The answer is they really do not know. They may have some hook up at the factory but nothing was presented at the hearings. The most important aspect though is, we do not know for each and every breath test the integration and result are flawless.

    Even a slight movement of the baseline drawing on the slightest wrong incline, or a start or stop of the peak that is even the slightest up on the peak rise or fall, will create the wrong result. Maybe the start or stop was too soon or too late and occurred in valley beneath the baseline, thus causing a bad result.

    There is no way to ever now how much off the mark it is without seeing the picture.

    Most instruments have software where you can change the integration parameters to force the instrument to properly integrate certain peaks, however, in doing this, the standards, and bracketing standards, must be under the same parameters, and this reprocess of results must be documented with explanation as to why.

    I am telling you, with 100 percent certainty, that the picture of analysis by analytical instruments of the nature of the Alcotest is the most fundamental document to support the result!

    yet is doesn’t exist!

    What really gets my goat, if you will, is that this really shouldn’t be up to me to point out such fundamental basic aspects of analytical chemistry.

    This is something that Drager should know, probably does know, the State should know, probably turns a blind eye to, defense attorneys have no idea about, … and only analytical chemists that were tested on it and charged with a dui care about.

    One can not even begin to validate an instrument like this in all seriousness, one that can take away a persons livelihood, freedom, driving privileges, respect, without having the picture.

    This is not like the measuring of the ph in your pool water.

    This is exact analytical chemistry science. Exact also made by statute, in virtue of the State differentiating between 0.07, 0..08, 0.09. 0.10% BAC >>

    To not be able to even begin to validate an instrument like this, where the result is so important, and basic IR instrumentation allows for it, because Drager didn’t engineer it to produce a picture with the counts, is really unbelievable, and wrong, and not just.

    What can I do to make people understand?

    I would like to be able to get in the above argument in my particular dui case.

    I will have to somehow figure out with my attorney how exactly we can get this argument on record, in order to have an appeal with this basis…. keeping in mind any appeal I  have would be based on many things including speedy trial and lack of due process, but also for this analytical argument to also come on record so it can also be appeal able.

    I don’t know what the case law says on this, but I certainly hope to be my own expert witness as to analytical results of peak based analysis by software driven analytical instruments.

    This on appeal, would be the new “Chun”.

    This fundamental aspect of this type of analysis, as I have described above, was not the main attack of the Chun defense team, and should have been.

    In my opinion they could have won on this argument alone.

    It is not a question of what Alabama is doing, as the Chun case seemed to rely on, or what other States are doing, or what other manufacturers of breath testing equipment are doing.

    It is a question, and case, of such fundamental basic analytical chemistry standards, as they exist today in NJ.

    It is not nuclear science, it is not rocket science, it is so basic, and so fundamental, and so low level, that analytical chemists like me, with just a basic college education, and low salary, take for granted.

    Producing a picture of an IR scan and its integration, is so basic and fundamental, it is equivalent to a auto mechanic checking the air in someone’s tire ( as far as it is being done),… that is to say NOT THAT IT IS EASY TO DO, BUT IT IS JUST SOMETRHING SO BASIC, SO FUNDAMENTAL, IT IS TAKEN FOR GRANTED IN THE ANALYTICAL WORLD.

    The science is exact, the analysis is exact, the statute is exact,  so what are we missing?

    We are missing the most fundamental piece to the puzzle, the actual picture!

    Before they make any political compromises as to the Chun case, understand one thing, my arguments are valid, not argued in Chun, and will be put up on appeal all the way, as the new Chun, if my case is not dismissed or acquitted after trial.
    There is no way this Alcotest, an analytical instrument, capable of taking away someone’s freedom, should be used without supplying the picture of the IR scan and the counts from the integration.

    One lawyer said to me, “why, because you say so”,

    Well, don’t take me word for it, I am only one analytical chemist in NJ….. talk to the others who work in analytical laboratories, talk to directors, managers, supervisors, and chemists in these labs. Interview 1000 of these people, … I guarantee not a one would disagree with me….

    It is not based on opinion or judgement, it is basic science.

    Believe me, the time and money and resources these companies use to analyze drugs and OTC products, in terms of even just amount of paper produced, is great, and they wouldn’t do it if they didn’t have to.

    My basic argument is that the taking away of a person’s freedom is so important, that the a picture of the IR scan of a breath analysis is paramount.

    There is simply no way to sum up what I am saying in a few lines or paragraphs, …. there is no simple way to write a long argument to describe it either…

    One must go see for themselves.

    This is something the Chun defense team refused to do, the State refused to do, and Drager refused to talk about.

    The level of importance on the breath test BAC result is high, and thus needs to be supported by the picture of IR scan and integration count result.

    Don’t take my word for it, go to major labs in NJ, major analytical labs at J&J, Shering Plough, Merck, Brystol Myers Squibb, OTC factory labs, contract labs, —go to these places and ask what they do, look at what they do, see for yourself.

    It is not a question of preferance, or policy from lab to lab, it is basic and fundamental, and required in code of federal register.

    Just take one of the most basic otc drugs we have in NJ,.. acetominophen.

    It is not ok to test and releaee acetominophen, by injecting the sample into an hplc and only getting a final result out, without a picture of the test or the integration results.

    Why is this?

    It is because peak based analysis, where quantification with integration is used, needs to be supported by the picture and integration result….. for the reason i described in several posts above.

  • John k Anderson

    Are these machines really accurate if not why r states allowing to be used when there are other ways to test people this is wrong and these companies should be held reliable

  • Okay you are correct, YouTube is best video distribution site, since YouTube is a lightly no much streaming time rather than other web sites.