Electronic Supplement to
Performance of Several Low-Cost Accelerometers

by J. R. Evans, R. M. Allen, A. I. Chung, E. S. Cochran, R. Guy, M. Hellweg, and J. F. Lawrence

Table S3. Summary of Sensor Performance

Model Box Flip*
(rank; details)		 Transfer Function (rank; description) Clip and Linearity(rank;
description)		 Wide-band Noise (bits PTP over ± 2 g) Double Integration (rank; description) General Comments Sum of ranks and viability for ANSS
Droid‡
 Rank 3;
sens. 0.1 / 0.8 / 1.3;
off. 0.1 / 0.3 / 3.7;
orient. 0.15 / 1.04 / 1.88;
polarities not typical		 Rank 5;
asinusoidal waveforms;
generally reasonable†
responses;
used time-domain rms		 Rank 3;
asinusoidal waveforms;
used time-domain rms;
X and Y responses reasonable but Z anomalous		 Rank 5;
13.7		 Rank 5;
unrecognizable waveform and large wander		 Needs reliable, higher sample rates (100 to200 sps);
must not drop samples		 Rank sum 21;
Unacceptable inpresent form mostly because of failed integration and low,uneven sample rates
GCDC Rank 4;
sens. 0.3 / 2.0 / 2.5;
off. 0.1 / 0.4 / 2.4;
orient. 0.91 / 1.68 / 2.59;
polarities not typical		 Rank 3;
reasonable response but sample-rate is unstable;
used time-domain rms		 Rank 3;
X and Y reasonable;
Z anomalous		 Rank 1;
17.1		 Rank 3;
mostly good waveforms but two examples of badbaseline excursions,probably due to dropped samples during motion		 Needs reliable, higher sample rates (100 to200 sps);
must not drop samples		 Rank sum 15;
Unacceptable inpresent form because of dropped samples and low sample rate
JWF14 Rank 1; for 2 g only:
sens. 0.1 / 0.9 / 2.1;
off. 0.1 / 0.9 / 1.4 (Z large);
orient. 0.45 / 0.79 / 1.18;
polarities are typical		 Ranks all 1 (ties);
reasonable responses, flat midband and probably to 0 Hz;
soft shouldered at high frequencies,but within 3 dB to~20 Hz (80 Hz for Phidgets)		 Rank 2;
plausible responses but quite variable roll-off frequencies		 Rank 4;
14.6		 Rank 4;
excessive velocity hysteresis;
waveform generally fair;
baseline wander somewhat high		 Double integration test poor (large baseline instability and hysteresis) Rank sum 12;
Unacceptable in present form principally for low resolution and poor integration
O-NaviA and B Rank 5;
sens. 0.0 / 1.0 / 2.1;
off. 1.1 / 1.9 / 5.3 (Z large);
orient. 0.37 / 0.66 / 1.31;
polarities not typical		 Rank 1;
reasonable responses but noticeably low sensitivities		 Rank 2;
16.1		 Rank 1 (tie);
acceptable waveforms and baseline wander;
some hysteresis		 Hysteresis and peak sample rates should be improved;
in other respects viable		 Rank sum 9;
Acceptable
Phidget Rank 2;
sens. 0.1 / 0.3 / 2.4;
off. 0.3 / 0.5 / 1.5;
orient. 0.12 / 0.90 / 2.28;
polarities not typical		 Rank 1;
reasonable responses but with somewhat variable sensitivities		 Rank 3;
15.2		 Rank 1 (tie);
acceptable waveforms and baseline wander;
slight hysteresis		 Only model tested that reaches ≥
200 sps with available software		 Rank sum 8;
Acceptable
Note: Ranks range from 1 (best) to 5 (worst) for each test. In cases of ties or generally acceptable performance, not all ranks are used but the range remains the same.
* Absolute errors are given in that order least / median / worst values of sensitivity (%FS), offset (%FS), and orientation (degrees). From Table S2. Typical polarities are positive up, north, and east.
† Reasonable meaning typical of accelerometers with correct damping: flat from 0 Hz to some corner where the amplitude response rolls off smoothly. Reasonable > plausible.
‡ Siting conditions are not addressed in this paper but should be considered with regard to record usefulness to ANSS.

[ Back ]