Electrochemical sensors of most types usually have a limited recommended humidity range for operation, for example 30% to 95%. At or near 100% they are liable to absorb moisture and flood.
At very low humidity the electrolyte volume will shrink enough to produce low results. This first became apparent many years ago when screeners put into Arizona were found to have a very short sensor life. There is however another more subtle situation. When the temperature falls below zero, the water content of the atmosphere drops sharply. The worst situation is where the outside temperature is low and the sensors/units are stored unprotected in a heated building, where the relative humidity is close to zero. Please note that prolonged exposure to such low humidity invalidates the warranty.
In some instances, sensors appear to recover when exposed to high humidity again, but data are as yet incomplete and inconsistent.
Meanwhile, as for fixes where continuous low humidity might be a problem, either store sensors/units in airtight containers or in a medium/high humidity environment; or design units so that the sensor is sealed off when not in use. It seems that it is storage which is critical, if they are being used frequently with wet gas/breath the problem does not seem so critical.
Sensors are heated in the following applications.
- Operation in a cold climate (police hand-helds).
- Interlocks, where they must operate from a start at very low temperatures.
- Fixed installations (e.g. coin-ops) where there is no power penalty and the system can benefit from greater speed of operation, needs no temperature compensation, and also not be potentially subject to condensation from breath.
For battery powered units, heating is unusual owing to the power drain.
For -20°C operation it is usual to heat the sensor. It is very slow at that temperature and gives about 10% of the ambient value. Also there are likely to be problems with condensation from breath, which leads to unreliable results. So it is good to get some heat into the sensor and also into the breath pathway to prevent condensation. Worth mentioning at this point, in your design, keep the sample pathway from mouthpiece to sensor as short as possible, or condensation can occur there leading to low results.
As indicated earlier, the calibration will depend on variables in your system – electronics, software, and sampling system.
First, the speed of response increases with temperature.
Secondly, there is a temperature coefficient to the response. The measured output (peak or integrated) rises with increase in temperature until somewhere in the region 20C to 40C, then gently falls. We are often asked to provide temperature coefficient data for sensors, but they depend on the details of the installation and anyway vary a little from sensor to sensor, so for most accurate set-up (for police use at least) you should calibrate each unit individually at least at three temperatures – 5°C, 20°C, 40°C are commonly chosen – generating a compensation algorithm.
Finally, it may shorten sensor life to some extent, particularly if accompanied by very low humidity. Formulations designed for interlock use have the highest resistance to temperature.
There are two commonly used methods, a “wet” gas and a “dry” gas.
A wet gas uses a liquid calibration standard in a breath alcohol “simulator”. The most popular simulator now is probably the range made by Guth Labs in the USA. We supply certified liquid standards, mostly for the UK market, as export is expensive owing to the large weight of the glass bottle and the water content. They are used in some countries as a traceable reference for locally made standards.
A dry gas uses a compressed alcohol/air mixture in a gas cylinder. Note that the fuel cell sensor reads dry and wet gases differently, the difference typically being about 6% higher for the wet, and so a correction – the “wet/dry factor” – must be made to adjust the dry result up. Also with a dry gas, a correction must be made for variation from standard barometric pressure as the dry gas will expand more into a low pressure atmosphere than a high pressure and so have a lower concentration.
The output is linear through zero so single point calibration is normal, the chosen value usually being the legal limit in the jurisdiction of the purchaser. For additional accuracy (for police use) calibration at three temperatures is usual.
For consumer units the 11mm sensor is suitable. We recommend the 16mm sensor for evidential and police use. The 32 mm sensor is only needed for old instruments taking that size.
The physical difference is that the premium carries a higher load of platinum (and so it costs more). This gives it greater calibration stability over repeat samples.
In practical terms, if you want to meet approval standards such as NF, you choose the premium.
If low cost is your priority, choose the economy and be more careful with the thinner contact wires.
Some interlock manufacturers are satisfied with the performance of the premium 11 mm sensor.
A rigorous treatment would be full integration of the response curve, but this has not been found necessary for adequate accuracy, repeatability and calibration stability. Use a zero load transimpedance circuit, and terminate the integration at some fixed fraction of the peak value (for example, 5%). An example circuit is available on request. Your electronics must be designed to be low-noise for this method to be effective.
There is a sharp peak and its magnitude depends on (for example) sample volume, alcohol content, age of sensor. It would not normally exceed around 1 mA. Diagnostic values are peak height, peak time, integration time and integral value. Peak values do decline in the early weeks and shelving during this period is common, but integral values are very stable over prolonged periods.
Breath alcohol sensors are almost always operated with a snap sample taken near the end of an exhalation, and signal processing may be classified according to whether it is by peak voltage or a current integration method. The course of breath expiration is generally detected with the aid of a pressure switch, a heated thermistor, or a microphone.
Recent work suggests that the best way to operate the pump is to take the sample, hold for about 200 mS, then reset. Alcohol vapour is taken up very rapidly in to the sensor wafer. Drawing in first, then blowing out, is the preferred method.
A suitable pump can be bought directly from our office in Shenzhen, small quantities can be purchased on this website and shipped from our UK factory.
The environmental alcohol sensor works by diffusion, so it just needs to be exposed to the environment to be sampled.
The mouthpiece design is not critical, there are many patterns. Basically it consists of a simple tube with a hole in the side to accommodate the inlet to the sensor. If a pressure switch is used to detect breath flow, then a constriction is placed downstream to build up back-pressure.
The minimum current is zero for a sample containing no alcohol.
Under practical conditions the sensor has no difficulty reading down to 5 µg/100 ml, 0.005 BAC, which is a commonly chosen cut-off point.
Under ideal conditions less than 1 ppm is possible, but the humidity of human breath and possible low level CO in smokers give transients which become significant at low levels.