METHOD AND DEVICE FOR DETECTING CONTACT OF A PIPETTE TIP WITH A LIQUID AS WELL AS A LABORATORY SYSTEM WITH SUCH A DEVICE

Abstract

A method for detecting contact of a pipette tip with a surface of a liquid in a pipetting device is disclosed. The method consists of moving the pipette tip in the direction of the surface of the liquid thereby measuring an absolute capacitance between the pipette tip and a reference potential and generating a sampled output signal. A predicted momentary sample value of the output signal is generated based on a plurality of past sample values of the output signal. A contact signal indicative of the pipette tip being in contact with the surface of the liquid is generated based on comparing a momentary sample value of the output signal with the predicted momentary sample value of the output signal. A corresponding pipetting device capable of performing the method as a laboratory system, in particular an automated liquid processing system, with one or more such pipetting devices are proposed.

Claims

1. A method for detecting contact of a pipette tip (4) with a surface (7) of a liquid (6) in a pipetting device (1), the method comprising the steps of: moving (m) the pipette tip (4) in direction of the surface (7) of the liquid (6); measuring an absolute capacitance (C) between the pipette tip (4) and a reference potential (GND), such as a grounded base plate (16) upon which a container (5) with the liquid (6) is arranged, and generating a sampled output signal (s.sub.out[n]); predicting a momentary sample value of the output signal (s.sub.out[n]) based on a plurality of past sample values of the output signal (s.sub.out[n]) and generating a predicted momentary sample value of the output signal (s.sub.out[n]); generating a contact signal (D.sub.L) based on comparing a momentary sample value of the output signal (s.sub.out[n]) with the predicted momentary sample value of the output signal (s.sub.out[n]), in particular by comparing a difference between a momentary sample value of the output signal (s.sub.out[n]) and the predicted momentary sample value of the output signal (s.sub.out[n]) with a predetermined threshold (TH), in particular a user/manually selectable and/or adjustable threshold, the contact signal (D.sub.L) being indicative of the pipette tip (4) being in contact with the surface (7) of the liquid (6).

2. The method of claim. 1, wherein generating the output signal (s.sub.out[n]) comprises the step of: applying samples of the measured absolute capacitance (C) to a median filter (12), in particular a median filter (12) of order in the range from three to six, more particularly a third order median filter (12).

3. The method of claim 1, wherein measuring the absolute capacitance (C) is performed at a first sampling rate and the sampled output signal (s.sub.out[n]) is generated at a second sampling rate by means of down sampling, in particular by a factor of N in the range from 2 to 61, more particularly in the range from 2 to 8, wherein down sampling is in particular performed by determining a sum of or an average over N consecutive samples to generate a sample of a down sampled output signal (s.sub.d[n]).

4. The method of claim 1, wherein predicting comprises applying one or more of the following to the output signal: curve fitting; finite impulse response (FIR) filtering; infinite impulse response (IIR) filtering; unweighted or weighted, e.g., exponentially weighted, (windowed, recursive) moving average filtering; linear or polynomial prediction; linear or polynomial regression; Kalman filtering.

5. The method of claim 1, wherein predicting comprises applying the output signal (s.sub.out[n]) to a sliding window averaging filter, wherein the window length (L) is in particular in the range from 8 to 64, more particularly in the range from 16 to 32.

6. The method of claim 1, wherein the step of predicting comprises applying the output signal (s.sub.out[n]) to a first and a second sliding window averaging filter, wherein the window length (L.sub.1, L.sub.2) of the first and second filter is in particular identical and in particular in the range from 8 to 64, more particularly in the range from 16 to 32.

7. The method of claim 6, wherein a slope of the output signal (s.sub.out[n]) is determined based on an output of the first sliding window averaging filter and an output of the second sliding window averaging filter, in particular wherein the predicted momentary sample value is based on linearly extrapolating the output of the first sliding window averaging filter and the output of the second sliding window averaging filter.

8. The method of claim 1, further comprising the step of: predicting a further momentary sample value of the output signal (s.sub.out[n]) based on a plurality of past sample values of the output signal (s.sub.out[n]) and generating a further predicted. sample value of the output signal (s.sub.out[n]), wherein predicting the further momentary sample value is performed differently than predicting the momentary sample value, in particularly based on one or more features different than that or those used for predicting the momentary sample value, and wherein the step of generating the contact signal (D.sub.L) is further based on comparing the momentary sample value of the output signal (s.sub.out[n]) with the further predicted momentary sample value of the output signal (s.sub.out[n]), in particular by comparing a difference between the momentary sample value of the output signal with the further predicted momentary sample value of the output signal with a further predetermined threshold, in particular a user/manually selectable and/or adjustable threshold, the further predetermined threshold in particular being the same as the predetermined threshold (TH).

9. The method of claim 1, wherein the past sample values of the output signal (s.sub.out[n]) are at least older by two sampling periods of the output signal than the momentary sample value, more particular older by three sampling periods.

10. The method of claim 1, wherein measuring the absolute capacitance (C) comprises charging and discharging a capacitor formed between the pipette tip (4) and the reference potential (GND) and determining a charging time, in particular an elapsed time for charging the capacitor from a first charging state to a second charging state, the charging time in particular being determined by a time-to-digital converter, wherein a measuring frequency is equal to a charging/discharging rate, and in particular in the range from 100 kHz to 500 kHz, more particularly in the range from 200 kHz to 400 kHz.

11. The method of claim 10, wherein charging the capacitor is performed by a voltage source (U) or current source via a charging resistor (R.sub.C), and wherein discharging is in particular performed via a switch (S.sub.D), such as a transistor, in particular a field effect transistor, in particular via a discharging resistor (R.sub.D).

12. A pipetting device (1) capable of detecting contact of a pipette tip (4) with a surface (7) of a liquid (6), the pipetting device (4) comprising: a pipette (3) with the pipette tip (4) adapted to aspirate and/or dispense the liquid (6); a transport unit (2), to which the pipette (3) is attached, adapted and configured to move (m) the pipette tip (4) in direction of the surface (7) of the liquid (6); a capacitance measuring unit (8) adapted and configured to measure an absolute capacitance (C) between the pipette tip (4) and a reference potential (GND), such as a grounded base plate (16) upon which a container (5) with the liquid (6) is arranged, and to generate a sampled output signal (s.sub.out[n]); a prediction unit (13) adapted and configured to predict a momentary sample value of the output signal (s.sub.out[n]) based on a plurality of past sample values of the output signal (s.sub.out[n]) and to generate a predicted momentary sample value of the output signal (s.sub.out[n]); a detection unit (14) adapted and configured to generate a contact signal (D.sub.L) based on comparing a momentary sample value of the output signal (s.sub.out[n]) with the predicted momentary sample value of the output signal (s.sub.out[n]), in particular by comparing a difference between a momentary sample value of the output signal (s.sub.out[n]) and the predicted momentary sample value of the output signal (s.sub.out[n]) with a predetermined threshold (TH), in particular a user/manually selectable and/or adjustable threshold, the contact signal (D.sub.L) being indicative of the pipette tip (4) being in contact with the surface (7) of the liquid (6).

13. The pipetting device (1) of claim 12, further comprising, in particular as part of the capacitance measuring unit (8), a median filter (12) adapted and configured to perform median filtering on samples of the measured absolute capacitance (C) and to generate the samples of the output signal (s.sub.out[n]), in particular a median filter (12) of order in the range from three to six, more particularly a third order median filter (12).

14. The pipetting device (1) of claim 12, further comprising, in particular as part of the capacitance measuring unit (8), a down sampling unit (11) adapted and configured to perform down sampling from a first sampling rate at which measuring the absolute capacitance (C) is performed to a second sampling rate at which the sampled output signal (s.sub.out[n]) is generated, in particular by a factor of N in the range from 2 to 64, more particularly in the range from 2 to 8, wherein down sampling is in particular performed by determining a sum of or an average over N consecutive samples to generate a sample of a down sampled output signal (s.sub.d[n]).

15. The pipetting device (1) of claim 12, wherein the prediction unit (13) comprises one or more of the following to which the output signal is applied: a curve fitting unit; a finite impulse response (FIR) filter; an infinite impulse response (IIR) filter; an unweighted or weighted, e.g., exponentially weighted, (windowed, recursive) moving average filter; a linear or polynomial prediction unit; a linear or polynomial regression unit; a Kalman filter.

16. The pipetting device (1) of claim 12, wherein the prediction unit (1) comprises a sliding window averaging filter to which the output signal (s.sub.out[n]) is applied, wherein the window length (L) is in particular in the range from 8 to 64, more particularly in the range from 16 to 32.

17. The pipetting device (1) of claim 12, wherein the prediction unit (13) comprises a first and a second sliding window averaging filter in series to which the output signal is applied, wherein the window length (L.sub.1, L.sub.2) of the first and second filter is in particular identical and in particular in the range from 8 to 61, more particularly in the range from 16 to 32.

18. The pipetting device (1) of claim 17, wherein the prediction unit (13) is adapted and configured so determine a slope of the output signal (s.sub.out[n]) based on an output of the first sliding window averaging filter and an output of the second sliding window averaging filter, in particular wherein the predicted momentary sample value is based on linearly extrapolating the output of the first sliding window averaging filter and the output of the second sliding window averaging filter.

19. The pipetting device (1) of claim 12, wherein the prediction unit (13) is adapted and configured to predict a further momentary sample value of the output signal (s.sub.out[n]) based on a plurality of past sample values of the output signal (s.sub.out[n]) and generating a further predicted momentary sample value of the output signal, wherein predicting the further momentary sample value is performed differently than predicting the momentary sample value, in particularly based on one or more features different than that or those used for predicting the momentary sample value, and wherein the prediction unit (13) is adapted and configured to generate the contact signal (D.sub.L) further based on comparing the momentary sample value of the output signal (s.sub.out[n]) with the further predicted momentary sample value of the output signal (s.sub.out[n]), in particular by comparing a difference between the momentary sample value of the output signal with the further predicted momentary sample value of the output signal (s.sub.out[n]) with a further predetermined threshold, in particular a user/manually selectable and/or adjustable threshold, the further predetermined threshold in particular being the same as the predetermined threshold (TH).

20. The pipetting device (1) of claim 12, wherein the past sample values of the output signal (s.sub.out[n]) are at least older by two sampling periods of the output signal (s.sub.out[n]) than the momentary sample value, more particular older by three sampling periods.

21. The pipetting device (1) of claim 11, wherein the capacitance measuring unit (8) comprises a charging unit (9) and a discharging unit (10) adapted and configured to charge and discharge a capacitor formed between the pipette tip (4) and the reference potential (GND) and a time measurement unit adapted and configured to determine a charging time, in particular an elapsed time for charging the capacitor from a first charging state to a second charging state, the time measurement unit in particular being a time-to-digital converter, wherein a measuring frequency is equal to a charging/discharging rate, and in particular in the range from 100 kHz to 500 kHz, more particularly in the range from 200 kHz to 400 kHz.

22. The pipetting device (1) of claim 21, wherein the charging unit (9) comprises a voltage source (U) or a current source and a charging resistor (R.sub.C), and wherein the discharging unit (10) in particular comprises a switch (S.sub.D), such as a transistor, in particular a field effect transistor, and in particular a discharging resistor (R.sub.D).

23. A laboratory system, in particular an automated liquid processing system, comprising one or more pipetting devices (1) according to claim 12.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0069] The present invention is further explained below by means of non-limiting specific embodiments and with reference to the accompanying drawings, which show the following:

[0070] FIG. 1 a schematic illustration of a pipetting device according to the present invention;

[0071] FIG. 2 an exemplary graph of the progression over time of a measured absolute capacitance as a pipette tip is lowered towards and then into a liquid;

[0072] FIG. 3 an exemplary graph of the progression over time of a measured absolute capacitance as a pipette tip is lowered towards and then into a liquid a) without and b) with applying a median filter according to the present invention;

[0073] FIG. 4 a schematic illustration of down sampling according to the present invention;

[0074] FIG. 5 a) an exemplary graph of the progression over time of a measured absolute capacitance as a pipette tip is lowered towards and then into a liquid, b) an exemplary graph of the progression over time of a difference signal between a predicted momentary sample value and the momentary sample value of the measured absolute capacitance when applying a first prediction technique according to the present invention;

[0075] FIG. 6 a) an exemplary graph of the progression over time of a measured absolute capacitance as a pipette tip is lowered towards and then into a liquid, b) an exemplary graph of the progression over time of a difference signal between a predicted momentary sample value and the momentary sample value of the measured absolute capacitance when applying a second prediction technique according to the present invention;

[0076] FIG. 7 a schematic illustration of the sample values of the measured capacitance as used in an embodiment according to the present invention;

[0077] FIG. 8 a) an exemplary graph of the progression over time of a measured absolute capacitance as a pipette tip is lowered towards and then into a liquid, b) an exemplary graph of the progression over time of a difference signal between a predicted momentary sample value and the momentary sample value of the measured absolute capacitance when applying the second prediction technique, and c) when alternatively applying the first prediction technique according to the present invention;

[0078] FIG. 9 a) an exemplary graph of the progression over time of a measured absolute capacitance as a pipette tip is lowered towards and then into a liquid when applying down sampling by a factor of four, b) an exemplary graph of the progression over time of a difference signal between a predicted momentary sample value and the momentary sample value of the measured absolute capacitance when applying down sampling by a factor of four and the second prediction technique, and c) alternatively applying the first prediction technique according to the present invention;

[0079] FIG. 10 a schematic illustration of a method according to the present invention for generating the liquid contact signal based on employing a combination of the first and second prediction technique.

[0080] In the figures, like reference signs refer to like parts.

DETAILED DESCRIPTION OF THE INVENTION

[0081] FIG. 1 shows a schematic illustration of a pipetting device 1 according to the present invention. The pipetting device 1 comprises a pipette 3 (a pipette tube) onto which a pipette tip 4 adapted to aspirate and/or dispense a liquid 6 present in a container 5 is mounted. The pipette tip 4 may be a disposable tip intended for single use (and replaced thereafter with a new one) or may be formed integrally with the pipette tube 3. The pipette 3 is attached to a transport unit 2 which can move the pipette tip 4 towards and into the liquid 6 (indicated by the arrow marked with an m to the right of the pipette tip 4), and subsequently remove the pipette tip 4 out of and away from the liquid 6 after aspirating a sample of the liquid 6 (the up and down movements being indicated by the double arrow next to the transport unit 2). The pipetting device 1 according to the present invention is capable of detecting contact of the pipette tip 4 with the surface 7 of the liquid 6. To achieve this, the pipetting device 1 comprises a capacitance measuring unit 8 adapted and configured to measure an absolute capacitance C between the pipette tip 4 and a reference potential, such as ground (GND), e.g., a grounded base plate 16 upon which a container 5 with the liquid 6 is arranged, e.g., held by a carrier 15. The absolute capacitance C depends on (i.e., is a function f of) a tip capacitance C.sub.T, a liquid capacitance and a carrier (or labware) capacitance C.sub.C (including the capacitance of the container) which is indicated by the series connection of the three capacitors C.sub.T, C.sub.L and C.sub.C in FIG. 1. It is to be noted that additional stray/parasitic capacitances may further contribute to the absolute capacitance C.

[0082] The absolute capacitance C is measured periodically at a certain sampling rate. This can be done in a variety of different ways. The measuring capacitor can for instance be excited by a sinusoidal voltage signal whereby the magnitude and phase of the response will be dependent on the capacitance. Alternatively, the measuring capacitor can be part of an oscillator circuit which consists of a coil and the measuring capacitor and in which the resonant frequency of the oscillator circuit is dependent on the measuring capacitor. According to FIG. 1 the capacitance measuring unit 8 comprises a charging unit 9 and a discharging unit 10 (all within dash-clotted lines) which charge and discharge the measuring capacitor formed between the pipette tip 4 and the reference potential GND. A time measurement unit (not explicitly shown in FIG. 1) determines a charging time (e.g., an elapsed time for charging the capacitor from a first charging state to a second charging state). The time measurement unit can for instance be realised by a time-to-digital converter such as the TDC7200 from Texas Instruments. The measuring frequency is then equal to a charging/discharging rate of the measuring capacitor and is for instance in the range from 100 kHz to 500 kHz, more particularly in the range from 200 kHz to 400 kHz. The charging unit 9 comprises a voltage source U and a charging resistor R.sub.C. The charging time is dependent on the value of the charging resistor R.sub.C and the current value of the measuring capacitor. The absolute capacitance C can therefore be determined from the measured charging time. The discharging unit 10 comprises a switch S.sub.D, which can for instance be implemented with a transistor, such as a field effect transistor (FET), and a discharging resistor R.sub.D. The discharging time is dependent on the value of the discharging resistor R.sub.D and the current value of the measuring capacitor C. The absolute capacitance may therefore also be determined from the measured discharging time. The value of the discharging resistor R.sub.D may be chosen to be smaller than the value of the charging resistor R.sub.D in order to achieve very rapid discharging which in turn allows to increase the measuring frequency. The voltage source U provides a constant charging voltage during charging and 0V (or the charging unit 9 is disconnected from the pipette tip 4) during discharging.

[0083] FIG. 2 shows an exemplary graph of the progression over time of the measured absolute capacitance C as the pipette tip 4 is lowered towards the surface 7 and then into the liquid 6. As can be seen in this graph the absolute capacitance C gradually increases as the pipette tip 4 gets closer to the liquid 6 (during which the various contributing capacitances C.sub.T, C.sub.L & C.sub.C are changing due to the varying geometry) and when the pipette tip 4 touches the surface 7 of the liquid 6 an abrupt jump in the value of the measured absolute capacitance C occurs. This sudden increase of the measured absolute capacitance C is due to the dominating effect of the capacitance C.sub.L of the liquid 6 when the pipette tip 4 is immersed in (i.e., in contact with) the liquid 6.

[0084] The aim of the present invention is to reliably and quickly determine when such a capacitance jump occurs to provide a contact signal indicating that the pipette tip 4 has touched the surface 7 of a liquid 6. Reliable detection of this contact event implies a low rate of false detections caused by interferences such as noise and cross coupling (e.g., due to capacitive coupling from neighbouring pipetting devices). Rapid detection of this contact event is necessary because moving of the pipette tip 4 to far into the liquid should be prevented. Thereby, the detection delay negatively impacts the (maximum) delivery speed by which the pipette tip 4 may be lowered into the container 5

[0085] According to the present invention contact detection is based on comparing a momentary (i.e., a current/present) sample value of the output signal s.sub.out[n] of the capacitance measuring unit 8 with a predicted momentary sample value of this output signal s.sub.out[n]. Hereby, the predicted momentary sample value of the output signal s.sub.out[n] is based on a plurality of past sample values of the output signal s.sub.out[n]. This prediction is performed by the prediction (or estimation) unit 13. The contact signal D.sub.L which indicates when the pipette tip 4 comes into contact with the surface 7 of the liquid 6 is for instance based on comparing a difference between the momentary sample value of the output signal s.sub.out[n] and the predicted momentary sample value of the output signal s.sub.out[n] with a predetermined threshold TH. The threshold TH may be a user/manually selectable and/or adjustable threshold. Generation of the liquid contact signal (indicator) D.sub.L is accomplished by the detection unit 14.

[0086] In order to decrease the false alarm rate of the liquid contact detection a number of optional measures may be taken. The purpose of these measures is to increase detection immunity against various sources of interference.

[0087] The samples of the measured absolute capacitance C may optionally be applied to a median filter 12, for example of order three. This median filter 12 selects the median value of three consecutive samples (i.e., the one with the middle value between the minimum and maximum value of the three sample) as its output. In this way outliers caused by erratic interference, such as electrostatic discharge (ESD) spikes, are removed from the output signal. This is clearly illustrated in FIG. 3 which shows an exemplary graph of the progression over time of a measured absolute capacitance as a pipette tip is lowered towards and then into a liquid a) when not applying the median filter 12 when a prominent ESD spike is present which would lead to a false liquid contact detection and b) when applying the median filter 12 which eliminates the ESD spike.

[0088] The signal-to-noise ratio (SNR) of the measured absolute capacitance C may optionally be increased by down sampling the measured absolute capacitance values, where measuring the absolute capacitance C is performed at a first sampling rate (to generate the sampled output signal s.sub.out[n′]) and the (down) sampled output signal s.sub.out[n] is generated at a second sampling rate by means of down sampling, for instance by a factor of N=4. As shown in FIG. 4 down sampling is for instance performed by determining a sum of or an average over N=4 consecutive measurement samples s[n′] to generate a sample of the down sampled output signal s.sub.d[n]. This is achieved by the down sampling unit 11 shown in FIG. 1. The effect of down sampling in terms of noise reduction (and hence increased SNR) is clearly apparent when comparing FIG. 8 a) showing all the samples of the measured absolute capacitance signal and FIG. 9 a) showing the down sampled (by a factor of N=4) version of the measured absolute capacitance signal which is much smoother and will therefore less likely lead to false liquid contact detections.

[0089] FIGS. 5 & 6 illustrate two alternative techniques for predicting the momentary sample value of the output signal s.sub.out[n] from a plurality of past sample values of the output signal s.sub.out[n]. As mentioned above, prediction may be performing in several ways with different levels of complexity. The employed prediction technique may for instance comprise one or more (in combination) of curve fitting, FIR or IIR filtering, unweighted or weighted, e.g., exponentially weighted, (windowed, recursive) moving average filtering, linear or polynomial prediction, linear or polynomial regression, and Kalman filtering. FIG. 10 for instance illustrates a combination of the two prediction techniques employed in FIGS. 5 & 6 which yields an even more reliable liquid contact detection.

[0090] A “first” exemplary prediction technique illustrated in FIG. 5 is based on averaging over 32 past sample values to predict/estimate the momentary/current sample value. This can be achieved by employing a 32-tap FIR filter, which simply calculates the sum over the 32 sample values (with all taps weighted by 1). The mean value is then the sum divided by 32. This mean value is taken as an estimate (i.e., the predicted value) of the momentary/current sample value. As shown in FIG. 7 the three samples immediately preceding the current sample are not taken in account for predicting the current sample—the prediction is based on the 32 samples preceding these three disregarded samples. The reason for doing so is to mask out the possible ascent in the capacitance curve towards the lump in capacitance. In this way the difference between the current sample value which is increased due to the onset of the lump in the capacitance curve and the predicted/estimated value, which is determined based on values that are still distant from this onset will be greater and consequently result in a correct indication of a liquid contact detection. This difference signal shown in FIG. 5 b). The contact signal D.sub.L will indicate a contact of the pipette tip 4 with the surface 7 of the liquid 6 as soon as the difference signal exceeds the threshold TH.

[0091] A problem of this prediction technique is that it is unable to take into account gradual increases of the slope of the capacitance curve which is often the case when the labware being used produces considerable capacitive coupling. In order to overcome this shortcoming of the “first” prediction technique, FIG. 6 illustrates an improved “second” exemplary prediction technique which takes into consideration the slope of the capacitance curve by extrapolating two mean values in order to predict the momentary/current sample value. The two mean values are derived from the same 32 past sample values as with the “first” prediction technique, whereby the first mean value is calculated from the 16 oldest sample values and the second mean value is calculated from the 16 newest sample values within the 32-tap FIR filter. The effect of the “first” and “second” prediction technique on the difference signal (i.e., current value minus predicted value) can be seen in FIGS. 8 & 9 when comparing the difference signal b) when employing the “second” prediction technique (i.e., averaging over 2×16 samples+extrapolation) with the difference signal c) when employing the “first” prediction technique (i.e., simple averaging over 1×32 samples). Note that in FIG. 8 no down sampling is employed and that in FIG. 9 down sampling by a factor N=4 is employed leading to an increased SNR. Clearly, the slope (preceding the peak) in the difference signal resulting from applying the “first” prediction technique seen in FIGS. 8 c) & 9 c) will cause false contact detection alarms if the threshold is not set to a high level (which increases the detection delay).

[0092] As can be seen in FIGS. 6 b), 8 b) & 9 b) the difference signal exhibits both a positive and a subsequent negative amplitude shortly before and after the pipette tip 4 touches the surface 7 of the liquid 6. This can have a negative impact when capacitive coupling from neighbouring channels occurs in that this may result in false contact detection. To overcome this problem, it is for example proposed to concurrently apply the “first” as well as the “second” prediction technique according to FIGS. 5 & 6 (i.e., employ two different prediction techniques simultaneously) and to perform contact detection based on the outcome from both, for instance by logically ANDing the two separate liquid contact detection decisions to provide a final joint decision. This yields an even more reliable liquid contact detection. Such a scheme is depicted in FIG. 10. Here an average/mean over the 32 past samples is calculated as a first prediction value which is compared with the momentary/current sample value and if the difference exceeds the threshold TH a first contact detection decision is positive. Concurrently, the same 32 samples are employed to calculate two average/mean values based upon which the second prediction value is determined via extrapolation. This second prediction value too is compared with the momentary/current sample value and if the difference exceeds the threshold TH a second contact detection decision is positive. A logical AND operation of the two contact detection decisions then results in the final liquid contact decision provided by the contact signal D.sub.L. As stated above other alternative predication techniques could be employed instead of the exemplary “first” and “second” prediction technique explained above for illustrative purposes. Moreover, for example a weighted combination of multiple (e.g., even more than two) separate contact decisions may be used instead of the described logical AND operation. Thereby, the weights may be dependent on a “quality measure” associated with a certain prediction/detection technique, e.g., its likelihood of providing an incorrect contact alarm (i.e., a “false positive” probability) or not indicating a contact when one has occurred (i.e., a “false negative” probability).

LIST OF REFERENCE SYMBOLS

[0093] 1 pipetting device

[0094] 2 transport unit

[0095] 3 pipette (tube)

[0096] 4 pipette tip

[0097] 5 container

[0098] 6 liquid in container

[0099] 7 surface of liquid

[0100] 8 capacitance measuring unit

[0101] 9 charging unit

[0102] 10 discharging unit

[0103] 11 down sampling unit

[0104] 12 median filter

[0105] 13 prediction/estimation unit

[0106] 14 detection unit

[0107] 15 carrier for holding container

[0108] 16 grounded base plate

[0109] C absolute capacitance (=f(C.sub.T,C.sub.L,C.sub.C)˜total series capacitance of C.sub.T, C.sub.L & C.sub.C)

[0110] C.sub.C carrier (labware) capacitance (incl. container)

[0111] C.sub.L liquid capacitance

[0112] C.sub.T tip capacitance

[0113] D.sub.L liquid contact signal

[0114] GND ground, reference potential

[0115] L, L.sub.1, L.sub.2 length of (1.sup.st/2.sup.nd) FIR/averaging filter (window length, number of taps)

[0116] m tip delivery motion

[0117] n, n′ sample index

[0118] N down sampling factor

[0119] R.sub.C charging resistor/resistance

[0120] R.sub.D discharging resistor/resistance

[0121] S.sub.D discharging switch.

[0122] s[n′] sampled output signal (at first sampling rate)˜absolute capacitance

[0123] s.sub.d[n] down sampled signal (at second sampling rate)

[0124] s.sub.in(t) input signal (applied to pipette tip)

[0125] s.sub.out(t) output signal (in response to input signal)

[0126] s.sub.out[n] sampled output signal˜absolute capacitance

[0127] t time

[0128] TH detection threshold

[0129] U voltage source