For a MIDAS measurement, sensitivity may be defined as the ability to meaningfully register small amounts of ferromagnetic material present in a sample. The MIDAS instrument responds to the overall magnetic properties of a sample, using this response to estimate the quantity of ferrous material included. There are a number of theoretical considerations in applying an inductive type instrument which have a bearing in interpretation of results. Here consideration is limited to those aspects relating to confidence that the reading taken is a meaningful measure of the targeted magnetic properties of the sample.
It has been established that MIDAS responds in a very linear fashion to the mass of ferromagnetic present (in the form of steel spheres). This linear response is fully characterised by the slope of the response and where the response meets the axis. The slope of MIDAS' response can be checked using the calibration 'sample' tube included with each instrument. The calibration sample reading is very consistent and sets a calibration factor or sensitivity in units of mg per litre of steel (specifically that in the sample tube). This condition is discussed further in the page 'calibration issues'.
At the low end of the scale we are more concerned with a number of difficulties which combine to reduce reliability of a reading. Perhaps the most significant of these are, in no particular order:
Response to dielectric properties of the materials involved- In contrast many fluids which may be used in MIDAS have very different dielectric (electrostatic or capacitive) properties. This implies that they will show differing influence on any electric field which is associated with the sensor (stray capacitance). In MIDAS an electrostatic shield is used to confine this electric field away from the sensed region, so that these capacitive effects on the sensor coil are negligible. This is apparent from the very small difference between MIDAS zero point for a range of fluids with widely divergent dielectric constant, such as oils, ethanol and water with dielectric constants of about 2-3, 24 and 80 respectively. Some 'magnetic' debris instruments are not similarly immune from dielectric differences in the sample, and practical sensitivity of readings suffer as a result.
Temperature effects and drift- Any effect on instrument stability due to changes of temperature will be detrimental to realisable sensitivity. Due to a self resetting aspect of the MIDAS design only the effects of temperature changes in a measurement period of less than one second may affect the datum point or zero setting, and any such change is normally so small as to be negligible. The electronics inherently compensate for temperature effects on instrument gain, across a wide operating temperature range.
Random system noise- Random noise from a number of sources may affect reading repeatability, and thus limit the useful sensitivity. Primary amongst random noise sources is random electronic noise in the sensor and related electronic circuitry. Output levels at the lower end of the range are extremely small and noise levels must be lower than this not to obscure real signals. This is achieved in MIDAS by appropriate low noise circuit techniques and construction.
Reading resolution- Resolution is not properly a part of sensitivity but an overly large reading step could obscure significance in readings if not otherwise limited. The 1 µg per ml resolution chosen for MIDAS is both convenient and does not normally mask significance in the data. Too high a display resolution would tend give rise to an erratic last display digit providing little or no additional information.
Interference effects- Any undue influence on results from interference of any sort can be viewed as detrimental to sensitivity. Here interference may be from external electromagnetic fields or the presence of metallic and / or magnetic objects nearby the sensor. A further possible source of such interference is from 'body' or 'hand' effects. The MIDAS sensor region is internal to the instrument and inherently well screened form these external interference sources.
The combination of the listed sources (and
any other significant sources) of uncertainty in low end readings will in
general combine to limit the achievable sensitivity of the unit. As
independent random sources they will in general combine in a mean square
manner. The square root of this error or root mean square (RMS), that is
its standard deviation, gives a useful measure of the practical
sensitivity limit of the unit. In MIDAS this is typically less than 0.5
µg / ml. An even lower effective error from these sources can be achieved
by averaging results from a series of readings.