System Accuracy - Philips KMZ51 Application Note

Philips Semiconductors
Electronic Compass Design using
KMZ51 and KMZ52

10. SYSTEM ACCURACY

Table 2 summarizes the main error influences on a compass system together with methods for calibration, as
discussed in the respective sections of this paper. In all equations, α is the real azimuth.
Error source
Offsets Vox and Voy
at SCU outputs
Sensitivity difference
∆S between SCU
channels
Non-orthogonality, i.e.
sensors mounted at
90°±β
Interference field
Hix,Hiy
("hard iron" effect)
Declination λ
Tilt τ
During trimming for compensation of ∆S, soft iron interference fields and tilt should be avoided, as these effects
would also appear as a difference in SCU output amplitudes.
In systems using a microcontroller, further errors occur, mainly due to quantization and non-linearity of the A/D
converter.
As the individual errors described above are random quantities and independent of each other, the total system
error can be found by applying the error propagation law:
= +
2 2
E E E
tot 1 2
Table 2 Summary of errors and calibration methods
Resulting azimuth error
(α - α
)
measured
V sin
α max
= −
arctan
V cos
max
V
[ ]
ox y ,
≈ °
0 8 . / %
V
max
=
V V
Assumption:
ox oy
 ∆

S
= α − +

arctan  1


S
∆
[ ]
S
≈ °
0 3 . / %
S
(
α β
+
sin
α
= −
arctan
α
cos
[ ]
β
≈ ° °
1 /
α
H sin
e
= α −
arctan
α
H cos
e
H
[ ]
ix y ,
≈ °
0 8 . / %
H
e
[ ]
= λ
°
(
δ τ
= ⋅
arctan tan sin
east
δ: inclination angle
+ +
2
... E
n
Calibration
α
+
Continuous in-circuit calibration by
V
oy
flipping combined with electronic
α
+
V
ox
feedback or mathematical correction.
Trimming of SCU amplification to


α

tan 
achieve equal amplitudes of Vout;


if required, in-circuit T-drift
compensation by electro-magnetic
feedback.
)
Measure β as phase difference
between SCU output voltages and
store it. Mathematical correction
during operation.
+
Determine interference fields, e.g.
H
iy
using bi-directional calibration
+
H
ix
method. Correction mathematically
or using compensation coils.
Shift zero-point of azimuth reading
by λ.
)
Continuous in-circuit calibration by
−
west
using 3-dimensional compass, pitch
and roll sensors and mathematical
correction.
(16)
34
Application Note
AN00022
See
section
5.2, 5.5
5.3, 5.5
5.4
7
8
9