A more controlled experiment was conducted to test the response of the force sensitive resistors. In this case, force was applied to a sensor (FSR + rubber buffer) using a hinge mechanism. The experiment tested the effect/variance of the resting height of the hinge lid on the sensor, the force applied to the lever/system, and the (binary) distance from force application to the fulcrum. For each combination lid height and distance to fulcrum (measure at the ‘front’ side and ‘back’ side of the sensor), three measurements were taken: one each at approximately 1kg, 3kg, and 6kg (hand washing pump force is likely to be 3.5-4.5kg). Statistical data has been compiled and is presented below as an ANOVA general linear model. In this instance, the force applied is represented in Kg, the distance from the force application to the fulcrum is represented by ‘f/b’ (1 being the side of the sensor closeset to the fulcrum), the resting height of the lid is represented by the categorical ‘lid’ variable — in position 0, the hinge lies at a lower height than the sensor, in position 5 the hinge lies at a lower height than the sensor, and at position 3 they are roughly at the same height. each position is approximately 0.3″.
General Linear Model: ohm versus lid, f/b Factor Type Levels Values lid fixed 6 0, 1, 2, 3, 4, 5 f/b fixed 2 0, 1 Analysis of Variance for ohm, using Adjusted SS for Tests Source DF Seq SS Adj SS Adj MS F P kg 1 3602646 3609772 3609772 38.93 0.000 lid 5 308095 305901 61180 0.66 0.657 f/b 1 799933 799933 799933 8.63 0.007 Error 28 2596131 2596131 92719 Total 35 7306805 S = 304.498 R-Sq = 64.47% R-Sq(adj) = 55.59% Term Coef SE Coef T P Constant 1034.3 103.5 10.00 0.000 kg -155.52 24.92 -6.24 0.000 lid 0 -33.8 113.5 -0.30 0.768 1 -73.0 113.6 -0.64 0.526 2 -72.9 113.6 -0.64 0.527 3 -33.1 113.6 -0.29 0.773 4 18.2 114.8 0.16 0.875 f/b 0 -149.30 50.83 -2.94 0.007 Unusual Observations for ohm Obs ohm Fit SE Fit Residual St Resid 25 1805.00 1040.10 143.29 764.90 2.85 R 31 2210.00 1269.33 154.55 940.67 3.59 R R denotes an observation with a large standardized residual.
This data suggests that it is highly likely that the force applied and the distance from the force application to the fulcrum are good predictors for sensor resistance. This makes sense. More insightful is the suggestion that the resting height of the lid is not a good predictor for sensor resistance (suggested by the correlation coefficients of the lid variable). The R2 value for this analysis seems reasonable enough with only 36 data points.
This scatterplot illustrates the difference in sensor response at two different locations (the left panel at the ‘front’ of the sensor, furthest from the fulcrum, and the right panel closest to the fulcrum). The right scatterplot indicates generally the type of distribution we would like to see, but with some sort of characteristic in the force range we would like (3.5-4.5kg). The left scatterplot represents desirable low variance.
On Thursday I will perform another test that does not vary lid height, continually varies fulcrum distance (rather than taking only two measurements), and takes more continual forces in the range that we are concerned with. If these results show enough promise, the hinge design will be confirmed at more vigorous work can be completed on this puck design.
