Bandwidth of Filters Changes Over Time
Updated: Sep 27, 2022
Sample cross-examination of CFS scientist on aging IR filters. IR instruments use light filters at the end of the sample chamber. As these filters age, there are changes in the bandwidth of IR light passing through them. More polychromatic light passes through, The result is a degradation in good linearity.
Sometimes defence lawyers need to teach basic science to the Court through the evidence of experts.
Consider the images below.
How do we communicate these concepts to a Court?
How do we expose over-simplification by Crown experts?
The pencil is the sample in the sample chamber.
The paperclips are the instrument's electronic response to the transmittance of IR light through the sample chamber. An electrical signal is produced by the detector at the end of the sample chamber that has a relationship to transmittance.
The instrument has been calibrated at the factory such that the calibration curve recorded in the instrument's calibration software settings says 4 paperclips = 2 gms and 2 paperclips = 1 gm.In other words, 4 paperclips of electronic response in the detector is made to be equal to 1 gm in indication. That making of a relationship is the calibration of the instrument at the factory during a manual tweaking (5000C) or an auto calibration sequence (8000C or 5000EN).
The circuit that does that is called the linearizer.
That relationship established with a screwdriver on a 5000C or a software adjustment on the 8000C does not change until the next re-calibration at the factory or the Canadian authorized Service Centre.
But what if the electronic response changes "over time" because the filter changes colour or gets dirty? Perhaps the bandwidth changes. Perhaps the properties of the filter and the optical system change "over time". What happens if the IR light bulb dims?
The calibration curve becomes wrong. It needs changing. The instrument needs re-calibration. The adjustment require may be non-linear - it may be a different amount at each of 50, 100, 150, 200, and 250 mg/100mls. The new calibration curve relative to the old calibration curve may be a movement up, down, left, right, a stretch, or a rotation on any axis.
Q. But my suggestion to you is that it's not safe
for scientists from the Centre of Forensic Sciences to be
saying that every test stands on its own, without paying
close attention to the issues of linearity that
Mr. Kupferschmidt was talking about, because there is the
problem that as filters age, they may change over time, the
bandwidths may change. And if the Centre of Forensic
Sciences doesn't have any information about the bandwidth of
the filters, then it's not safe to assume that linearity is
going to be maintained over time.
A. The linearity of the instrument isn't going to
change. I don't know if I covered it the last time. It's
the response of that instrument that's going to change over
time, yes, because of changes over time with respect to the
light source or the filters themselves, correct. But, again,
the diagnostic checks and the calibration check that's
performed at the time of the test will let you know if
there's anything grossly wrong with the instrument that
requires it to go for service and possibly require either a
new light source and/or new filters for the detectors.
[how much is "grossly"? - does the nature or kind of the "approved" instrument change if the change is slight, moderate, or gross? - how much drift is acceptable? - what would your metrology expert say about this?]
Q. All right. Just to break that down, that
answer. I agree with you with respect to linearity between
the instrument response and the indication on the instrument,
assuming that the calibration curve is fixed, it hasn't
changed, because there's been no re-calibration. But --
A. So, wait, we're talking about the subject test
in relation to the calibration check response?
Q. I'm talking about, I'm talking about over time
there is always an issue of linearity of the concentration in
the sample chamber, compared to the indication, as the
bandwidth of that filter changes over time.
A. There could be, yes.
Q. Right. Especially when the Centre of Forensic
Sciences doesn't know the bandwidth of the filter. Is it
big? Is it small? We don't know that, right?
A. Correct. There's some information. I don't
know where that information came from. I don't have it at my
fingertips. I seem to recall that, again, there was
information about what the bandwidth was. Can that change?
Can the -- like a pair of sunglasses or anything that's
exposed to light for long, long periods of time, could
response to that change? Yes, absolutely.
[The reality is that bandwidth of IR filters may degrade over time. Some IR filters are better (and more expensive) than others. It is not safe to assume bandwidth is unchanged in an aging instrument. It is not safe to assume that the IR light used in the samle chamber is monochromatic. It is not safe to assume that the constant used in the Beer-Lambert equation (and assumed in the CFS Training Aid) is applicable anymore to an aging instrument. Get an expert.]
[The above cross-examination relates to a CMI instrument. These concepts will apply to any IR instrument used for a forensic purpose.]
[If the public and the Justice system wish to ensure reliabilty of approved instruments, the only way to do so is to legislate metrological supervision of police. Why don't we do that in Canada?]
Application to a Datamaster evidentiary breath instrument:
The following is an example showing a filter region of the IR spectrum used by Datamaster. Notice the shape of the peaks/valleys and the large bandwidth. And what if the bandwidth changes over time?