Melexis Triaxis Back-End Calibration Sensors User Guide
- October 30, 2023
- Melexis
Table of Contents
Triaxis Back-End Calibration Sensors
User Guide
Introduction
The back end calibration is a feature of the Triaxis position sensors that allows the user to compensate for non-linearity error in the application, improving the linearity performance. It consists of defining a transfer function between the sensor’s measurements (after front end error compensation) and the output. Consequently, it defines the behavior of the output vs the magnet position in the application. This is done as part of the End-of-Line programming of the sensor. Different Triaxis products allow for different types of back end calibration methods. Also, the way a certain method works may vary between the different Triaxis generations. Therefore, the purpose of this document is to help the user select the appropriate sensor and method for their application. Note, however, that this guide is dedicated to the calibration methods typically used in rotary or linear displacement applications. Different methods are available for joystick applications, but those are beyond the scope of this document.
Calibration Methods
The Triaxis position sensors currently offer four different calibration
methods. In general, all methods consist of storing a set of values in the
sensor’s non-volatile memory which will be used to convert the sensor’s angle
measurement into an output in %VDD (analog), % duty cycle (PWM) or LSB (SENT
or PSI5). This chapter presents an overview of these methods.
2.1. 4-point Calibration
2.1.1. Description
The first generation of Triaxis devices offered a 3 points / 4-slope
calibration. With the launch of MLX90360 Melexis introduced a 4 points /
5-slope calibration. It allows the user to define the output transfer curve
through a combination of points and slopes. To do so, the user should move the
magnet in the application to a number of positions and set the target output
for each using the software library provided by Melexis. Depending on the
generation the user has the ability to program positive and negative slopes
per point, e.g. MLX90316-BCG offers a negative and positive slope only for
slope C whilst all other slopes per point must be positive. Later generations
offer both options per point.
Not all of the points have to be set by the user. Customers looking for better
nonlinearity compensation should use at least three calibration points.
Customers looking for a cost-effective calibration set-up and shorter
calibration time will prefer two-point calibrations. However, the above
statement is to be taken into consideration together with the desired angle
range i.e. within small-angle ranges 2-points calibration can be as effective
as e.g. 3 points or 4 points calibration.
The calibration points can be placed anywhere, which means the user can
achieve a better linearization by placing the points where the non-linearity
is the highest, as shown in Figure 1. One might notice two extra points in
this figure, these are called virtual points. It’s the first and last point
and they are used typically by the solver to calculate LNR_S0 and LNR_D_S (or
LNR_C_S in case of 3 points).
Figure 1
Example of Non-Linearity Error before 4-point calibration; Blue curve
resembles the measured angle by the sensor: reality (nonlinear behavior coming
from magnet or system); Yellow points resemble the calibration points. First
and last points are virtual points typical used by the solver to calculate the
first and last slope
Figure
2 Example of residual error after 4-point calibration. Blue resembles the
error without back-end calibration; Green resembles the error after 4 points
calibration.
This method can also be used to define transfer curves with different shapes,
including discontinuities (e.g. a saw-tooth curve, take note that enable
scaling must be set for defining “negative” output values) as shown in Figure
3, by setting the slopes independently from the points. Not applicable to Gen
I. Triaxis. For more information, please consult the datasheet.
Figure 3 Example of the
output transfer curve that can be defined with the 4-points calibration
2.1.2. Typical applications
Typical applications include (but are not limited to):
- Electronic Throttle Body with 2-point calibration;
- Ride height sensors;
- Accelerator pedals;
- Linear actuators;
- Linear displacement applications with limited stroke;
2.2. 8-point Calibration
2.2.1. Description
The 8-point calibration method was introduced in the third Triaxis generation
and is very similar to the 4point method.
It allows the user to define up to 8 calibration points, also in arbitrary
positions. However, the user cannot define the slopes independently, so the
segments between two points are only calculated by linear interpolation.
Different from the 4-point calibration is that the two virtual points are
fixed at VP1{0,0%} and VP2{360,100%}. For more information please consult the
datasheet.
Having a greater number of calibration points, combined with the freedom to
place the points anywhere, this method allows the user to achieve a better
linearity performance, as shown in Figure 4.
Figure 4 Example of Non-Linearity Error before 8-point
calibration; Blue curve resembles the measured angle by the sensor: reality
(nonlinear behavior coming from magnet or system); Yellow points resemble the
calibration points. The first and last points are virtual points fixed by the
sensor.
Figure 5 Example of residual error after 8-point calibration.
Blue resembles the error without back-end calibration; Green resembles the
error after 8 points calibration
2.2.2. Typical applications
Typical applications are the same as listed in paragraph 2.1.2
2.3. 17-point Calibration
2.3.1. Description
The 17-point calibration method (often referred to as 16-points calibration)
has also been available since the first Triaxis generation (including
MLX90324, excluding MLX90316). It allows for a better linearity performance
when compared to the previous methods, as shown in Figure 4.
Figure 6 Example of Non-Linearity Error before
17-point calibration. Blue curve resembles the measured angle by the sensor:
reality (nonlinear behavior coming from magnet or system); Yellow points
resemble the calibration points. 
Figure 7 Example of residual error
after 17-point calibration. Blue resembles the error without back-end
calibration; Green resembles the error after 16 points calibration.
In this method, the transfer curve is defined by 17-points, which are
equidistantly spread over the angular working range of the sensor.
There are two ways to define the points. The first is to enter the calibration
points using the dedicated solver, which is a part of the software library
provided by Melexis. In this case, the targets entered by the user are not
directly stored in the memory but are interpolated in order to determine the
equivalent targets at the equidistant angles. The user can enter as many
calibration points as desired, and this will affect the calibration time as
well as the linearity performance.
The second way to determine the points is to pre-program the sensor with a
default output characteristic, and then characterize the output by comparing
it with a reference. Afterwards, the 17 points can be directly calculated and
stored in the memory. This method is not as simple to implement but allows for
very fast programming, suitable for production timing.
2.3.2. Typical applications
Typical applications include (but are not limited to):
- Through-shaft rotary applications;
- Rotary knob selector;
- Linear displacement applications with long strokes;
2.4. 32-point Calibration
2.4.1. Description
The 32-point calibration method was introduced in the third Triaxis generation
and is very similar to the 17- point method. However, by having twice as many
points, it allows for an even better linearity performance, as shown in Figure
8.
Figure 8
Example of Non-Linearity Error before 32-point calibration. Blue curve
resembles the measured angle by the sensor: reality (nonlinear behavior coming
from magnet or system); Yellow points resemble the calibration points
In the 32-point method, the points are also equidistantly spread over the
angular working range of the sensor. Just as the 17-point method, this method
also offers two ways to define the points, one using the dedicated solver and
the other calculating the points directly by comparing the sensor’s output to
a reference. 
Figure 9: Example of
residual error after 32-point calibration. Blue resembles the error without
back-end calibration; green resembles the error after 32 points calibration
Therefore, customers that are already familiar with the 17-point method can
easily use this method as well, achieving the best linearity performance, as
shown in Figure 9, with a reasonable programming time.
Other than the number of points, the main difference this method has is the
fact that the points are no longer stored in the memory as absolute values.
Instead, a slope is calculated based on the desired output behavior and then
only the difference between the calibration points and the values given by the
slope are stored in the memory.
The resulting transfer function is formed by these 32 differential points,
plus an additional point that is equal to the value given by the slope and
that can be either at the start or in the middle of the working range,
depending on the product.
2.4.2. Typical applications
Typical applications include (but are not limited to):
- Transmission range sensors;
- Brake pedals;
- Linear displacement applications with long strokes;
- Arc-shaped displacement applications;
Parameters
Table 1 shows the programmable parameters related to back-end calibration for
the various methods and Triaxis generations. These parameters can be defined
either by the solver algorithm or directly by the user.
Note that the parameter names may also vary between generations.
Table 1 Programmable parameters related to back end calibration
Parameter| 3-pt Gen I| 4-pt Gen II & III| 8-pt Gen III|
16-pt Gen I & II| 16-pt Gen III| 32-pt Gen III
---|---|---|---|---|---|---
Clockwise (CW)| | | | | |
Discontinuity Point (DP)| | | | | |
LNRAX| | | | | |
LNRBX| | | | | |
LNRCX| | | | | |
LNRDX| | | | | |
LNRAY| | | | | |
LNRBY| | | | | |
LNRCY| | | | | |
LNRDY| | | | | |
LNRS0| | | | | |
LNRAS| | | | | |
LNRBS| | | | | |
LNRCS| | | | | |
LNRDS| | | | | |
LNRX0…LNRX7| | | | | |
LNRY0…LNRY7| | | | | |
LNRY8…LNR16| | | | | |
LNR_DELTAY01…
LNR_DELTAY32| | | | | |
Working Range (W)| | | | | |
Clamp Low| | | | | |
Clamp High| | | | | |
Anchor Mid| | | | | |
LNR_DELTA_Y_EXP| | | | | |
WORK_RANGE_GAIN| | | | | |
Selection Guide
Table 2 shows the calibration methods available for each Triaxis product.
Table 2 Calibration methods available for each Triaxis product
Back End Calibration Method| 4-point| 8-point| 16-point|
32-point
---|---|---|---|---
MLX90316| | | |
MLX90324| | | |
MLX90333| (x2)| | |
MLX90340| | | |
MLX90360| | | |
MLX90364/MLX90365| | | |
MLX90366/MLX90367 See versions| (ABT/ABU)| | (ABS/ABV/ABX)|
MLX90421| | | |
MLX90422| | | |
MLX90423| | | (arbitrary)|
MLX90425| | | |
MLX90426| | | |
MLX90371| | | |
MLX90372| | | |
MLX90373| | | |
MLX90374| | | |
MLX90378| (x2)| | |
Graphical representation
The graph below shows the relationships between the mechanical position
applied, the target output, the measured angle, and the calibration points.
The example is given for a 16 point calibration. The measured range is divided
into 16 segments of the same length, yielding 17 equidistant points. The
target at these points is what will be programmed inside the memory of the IC.
The red lines show the position where the output clamping starts. The points
before and after are adjusted to set the transition point accordingly. Note
the third quadrant. This quadrant is of particular interest when using the
sensor in analog mode. The error (gain and offset) introduced by the digital
to analog converter (DAC) can be corrected. This is not applicable for PWM and
digital outputs
Figure
9 Relation between mechanical position vs target and sensor output
Useful Documents
- AN_90360_BackEndCalibration_rev3.1.pdf
- MLX90324-90360-AN_16pts_calibrationVs3Pts.pdf
- AN-MLX90371_MLX90372_32-PointCalibration.pdf
See also our video about back-end calibration: https://youtu.be/S-PT4CkWdVQ.
Disclaimer
The information furnished by Melexis herein (”Information”) is believed to be
correct and accurate. Melexis disclaims (i) any and all liability in
connection with or arising out of the furnishing, performance or use of the
technical data or use of the product(s) as described herein (”Product”)
(ii) any and all liability, including without limitation, special,
consequential or incidental damages, and (iii) any and all warranties,
express, statutory, implied, or by description, including warranties of
fitness for particular purpose, non-infringement, and merchantability. No
obligation or liability shall arise or flow out of Melexis’ rendering of
technical or other services.
The Information is provided ‘as is’ and Melexis reserves the right to change
the Information at any time and without notice. Therefore, before placing
orders and/or prior to designing the Product into a system, users or any third
party should obtain the latest version of the relevant information to verify
that the information being relied upon is current. Users or any third party
must further determine the suitability of the Product for its application,
including the level of reliability required and determine whether it is fit
for a particular purpose. The Information is proprietary and/or confidential
information of Melexis and the use thereof or anything described by the
Information does not grant, explicitly or implicitly, to any party any patent
rights, licenses, or any other intellectual property rights. This document, as
well as the Product(s), may be subject to export control regulations. Please
be aware that export might require prior authorization from competent
authorities. The Product(s) are intended for use in normal commercial
applications. Unless otherwise agreed upon in writing, the Product(s) are not
designed, authorized or warranted to be suitable in applications requiring
extended temperature range and/or unusual environmental requirements. High-
reliability applications, such as medical life-support or life-sustaining
equipment are specifically not recommended by Melexis. The Product(s) may not
be used for the following applications subject to export control regulations:
the development, production, processing, operation, maintenance, storage,
recognition or proliferation of
- chemical, biological or nuclear weapons, or for the development, production, maintenance or storage of missiles for such weapons:
- civil firearms, including spare parts or ammunition for such arms;
- defense-related products, or other material for military use or for law enforcement;
- any applications that, alone or in combination with other goods, substances or organisms could cause serious harm to persons or goods and that can be used as a means of violence in an armed conflict or any similar violent situation.
The Products sold by Melexis are subject to the terms and conditions as
specified in the Terms of Sale, which can be found at
https://www.melexis.com/en/legal/terms-and conditions.
This document supersedes and replaces all prior information regarding the
Product(s) and/or previous versions of this document.
Melexis NV © – No part of this document may be reproduced without the prior
written consent of Melexis.
(2021)
IATF 16949 and ISO14001 Certified
Revision 002
Docserver № 390110000007
Read User Manual Online (PDF format)
Read User Manual Online (PDF format) >>