AN5851

Application note

Water and liquid level monitoring using VL53L4CD Time-of-Flight high accuracy
proximity sensor

Introduction
This document describes how a user can design a system to measure the level of a liquid in a container with ST Time-of-Flight
(ToF) sensors.
The ST ToF sensors work by emitting a pulsed cone of light, which hits the target and bounces back. The sensor detects the
returning signal and measures the photon travel time. The travel time divided by two and multiplied by the speed of light gives
the distance. In most cases, this is a very robust system.
In this application note, the VL53L4CD ToF sensor is used to measure the level of water in a container. All ST single zone
Time-of-Flight sensors (including VL53L3CX and VL53L0X) can be used in the same manner.
Background
The VL53L4CD is a ToF sensor especially designed to provide very accurate distance measurements from only 1 mm up to
1300 mm. A new generation laser emitter with 18° diagonal FoV improves performances under ambient light, with ranging
speed up to 100 Hz.

AN5851 - Rev 1 - October 2022 www.st.com

For further information contact your local STMicroelectronics sales office.
 AN5851
Observations on water

1 Observations on water

The basic principle of a ToF sensor is to measure the distance to the target based on the time for emitted photons
to be reflected. In most applications, photons travel through the air. But, to measure the level of a liquid, photons
travel through air and water. As a result, the light path is impacted by the different refractive indices.
In other words, there is part of the signal that is reflected from the surface, and another part returned from the
container’s base (traveling through the liquid), and another part bounces from the liquid itself. Thus, the final
ranging distance is impacted with respect to accuracy (see Figure 1. Different light paths when ToF signal aimed
at liquid). However, this document describes some techniques and methods to minimize the ranging error, using
the nonlinearity correction algorithm.

Figure 1. Different light paths when ToF signal aimed at liquid

AN5851 - Rev 1 page 2/20

 AN5851
ST recommendations

2 ST recommendations

2.1 Container properties

The position of the VL53L4CD sensor and the container size impact the accuracy in ranging because the sensor’s
field of view (FoV) is cone shaped (18° for VL53L4CD). The ToF should be mounted so that the signals are
reflected only from the water surface, and not from the edge of the container.
If the container is narrow, with a diameter less than the ToF’s FoV (as illustrated in Figure 2. FoV bigger than
container diameter) then the signals are reflected from the edge, which induces noise. This impacts the ranging
accuracy.
Therefore, ideally, the container diameter should be greater than the ToF’s FoV (as illustrated in Figure 3. FoV
smaller than container diameter).

Figure 2. FoV bigger than container diameter

Figure 3. FoV smaller than container diameter

In addition, there should be a reasonable gap (minimum 30 mm) between the sensor and the maximum water
level. This avoids any obstruction of the ToF’s light path, in the case water drops bounce while the water is
pouring, or if the container is shaken.

AN5851 - Rev 1 page 3/20

 AN5851
Container reflectivity

2.2 Container reflectivity

It is recommended to use containers with a base made of low reflective material (see Figure 4. Low reflective
base) rather than fully transparent or high reflective (see Figure 5. Transparent or high reflective base). This
avoids receiving mixed signals from base and liquid surface when the liquid level is low. Figure 19. Water level
measurement with different colored base and no compensation illustrates the ranging behavior with different
background base.

Figure 4. Low reflective base

Figure 5. Transparent or high reflective base

AN5851 - Rev 1 page 4/20

 AN5851
ST solution

3 ST solution

3.1 Nonlinearity correction algorithm

There are many containers with different shapes and sizes available on the market, and it is not possible to test
them all. Therefore, if the ranging result is nonlinear, then the nonlinearity correction algorithm can be used to
improve the ranging accuracy.
Figure 7. Nonlinear ranging results, shows that the ranging result is higher than expected when the water level is
at the minimum, is steady in the middle range, and decreases when the water level is at the maximum. Hence,
the offset mechanism can help reduce the nonlinearity of the results.
As the algorithm works in a defined setup, the input to the algorithm is the total distance of the container, which
is stored in the host memory. This algorithm maintains a range of distance in %. The reason the distance is
calculated in percentage is that liquid level application shows level of indicators that can be easily correlated with
% value. For example if a container’s height is 1000 mm, and there are five levels of indicators, the indicator is
incremental of 200 mm.
At the start of the program, it calculates the distance based on a list of percentages. As the water level starts
increasing, the program gets the ranging distance, and compares it to the actual distance it has stored. By
iterating this experiment with different water levels, the error rate that needs to be offset later is identified. Finally,
a list of error rates corresponding to the water level is established. These error rates are compensated by the
offset values that need to be programmed in an algorithm to improve the ranging accuracy. Figure 6. Level
indicators based on water level explains how the zones are divided based on water level and apply the offset
values to reduce nonlinear impact on distance.

Figure 6. Level indicators based on water level

AN5851 - Rev 1 page 5/20

 AN5851
Nonlinearity correction algorithm

Figure 7. Nonlinear ranging results

Example
If a container’s height is 1000 mm and it has five indicator levels. The distance corresponds to the indicators: 200
mm, 400 mm, 600 mm, 800 mm, and 900 mm. These correspond to 20%, 40%, 60%, 80%, and 90%.
When the water level starts pouring, note the ranging distance at each level.

Table 1. Ranging results example

Water level in % Expected (mm) Actual (mm) Error in %

20 800 840 5
40 600 620 3
60 400 380 5
80 200 160 20
90 100 70 30

You can reiterate the test to get a clearer view of the deviation in ranging with regard to the water level.
Apply these ranging errors to the algorithm to improve the ranging accuracy. For example, if the program gets 840
mm in 20% water level, then apply 5% offset to the algorithm to minimize the error.

AN5851 - Rev 1 page 6/20

 AN5851
Ranging result characterization

3.2 Ranging result characterization

Figure 8. Nonlinear impact on ranging result illustrates the characterization of the ranging result on the different
liquid levels. This experiment assumes a transparent plastic jar, with a width of 100 mm and height of 189 mm.
The measurement points are broadly divided into three zones (low, medium, and high) to define the ranging
behavior within the different zone. The ranging curve changes at the inflexion points.

Figure 8. Nonlinear impact on ranging result

AN5851 - Rev 1 page 7/20

 AN5851
System characterization using ST tool

3.3 System characterization using ST tool

ST provides a characterization tool (see Figure 9. Characterization tool) to better understand the signal reflectivity
of the container and liquid. The tool can be found in the STSW39_L4CD package. This improves the ranging
accuracy in deviation level.

Figure 9. Characterization tool

It is recommended to use a dust free cover glass on top of the container and to measure more than one
similar setup (minimum 10 sets) to validate the consistency in measurement. Once the same nonlinear curve
with ranging deviation is found across all setups, you can adopt a nonlinearity algorithm to reduce nonlinearity in
ranging.
This algorithm uses a lookup table that is created based on noted data points and ranging deviation on different
water levels. The lookup table can be easily modified based on the container’s size or setup, and the data
received. The only input to this algorithm is the total distance from the sensor to the container’s base, which
needs to be stored in the host memory (NVM). This algorithm reads this value before the start of ranging, and
calculates how much offset to apply dynamically on different levels.
Alternatively, the user can apply a linear regression model to all data points to find the intercept and coefficient for
the best fit line. These two parameters can be used in ranging to compensate the ranging deviation.
Following is a snippet of characterization tool; it allows the user to perform a defined number of times the
characterization setup. For example, in the characterization plot below, the red lines show the characterization of
the same setup and liquid three times. The green line is the mean derived from all characterized values. This can
be used to create a lookup table (Figure 10. Data for lookup table).

AN5851 - Rev 1 page 8/20

 AN5851
System characterization using ST tool

Figure 10. Data for lookup table

The final data points are created in FinalResult.csv (Figure 11. Sample data points of characterization).
The characterization process begins by continuously pouring liquid at each level and clicking the “Characterize”
button until all levels are finished. This process continues until the defined number of setups or iterations are
characterized. At the end, the user can see the final characterization graph by using a nonlinearity correction
algorithm. For the experiment, it includes a linear regression model as well.

Figure 11. Sample data points of characterization

AN5851 - Rev 1 page 9/20

 AN5851
Nonlinearity correction algorithm flow

3.4 Nonlinearity correction algorithm flow

The following flow chart depicts the flow of the nonlinearity correction algorithm. The total height of the container
is stored in the host memory and it is provided as input for the algorithm.

Figure 12. Algorithm flow

AN5851 - Rev 1 page 10/20

 AN5851
Reference code for nonlinearity correction algorithm

3.5 Reference code for nonlinearity correction algorithm

The code snippet below determines the distances between the sensor and the liquid level, by increments of 10%.
/*Find water level position based on real ranging distance and apply offset value to correct deviation
Input : totaldistance= height of container 2. rangevalue = received by sensor
Output : rangevalue_out= After compensate value (og_val) which you received from characterization method applied
*/
void Liquidlevelmeasureerrorcomponsate(uint16_t totaldistance,uint16_t rangevalue,uint16_t *rangevalue_out)
{
uint16_t pos=0;
uint16_t i=0; uint16_t value;
if (totaldistance ==0)
totaldistance=TOTAL_HT_CONTAINER; // Test setup container's height
// In this test, MAX_LABEL =10
for (i=0;i<MAX_LABEL;i++)
{
if (rangevalue <totaldistance+5)
{
if (rangevalue < ogalgo_data_inst[i].expected_res)
{

// Check first
if (i >0)
{
value = (ogalgo_data_inst[i].expected_res + ogalgo_data_inst[i-1].expected_res)/2;
}
else
value = ogalgo_data_inst[i].expected_res;

if(rangevalue <= value)

{
/* Find water level position based on ranging value */
if (i ==0 )
{
pos=1;
break;
}
pos=1+(i-1); // adding one bcz position starts from 1
break;
}
else
{
pos=1+i;
break;
}
}
else
{
if (rangevalue > totaldistance+5)
pos=99;
else
pos=6; // Need to find which one has max OG value or store it directly here
}
}
else
pos=99;
}

Based on the liquid level, the algorithm detects if it is over ranging or under ranging. It then uses a lookup table to
apply offset to improve ranging accuracy. The offset compensation value applied depends on the environment or
setup.
// position = water level position, og_val= got compensated value from characterization method
switch(pos)
{
case 1:
*rangevalue_out=rangevalue + ogalgo_data_inst[pos].og_val; // Category Underranging,
break;
case 2:
*rangevalue_out=rangevalue + ogalgo_data_inst[pos].og_val; // Category Underranging, 30mm deviation b/w C3&C2
break;
case 3:
*rangevalue_out=rangevalue + ogalgo_data_inst[pos].og_val; // Category Underranging, 25mm deviation b/w C4&C3
break;
case 4:
*rangevalue_out=rangevalue + ogalgo_data_inst[pos].og_val; // Category Underranging, 12mm deviation b/w C2&C2
break;
case 5:
*rangevalue_out=rangevalue - ogalgo_data_inst[pos].og_val; // Category Overranging, 4mm deviation b/w C2&C2
break;
case 6:
*rangevalue_out=(uint16_t)(rangevalue - ogalgo_data_inst[pos].og_val); // Category Overranging, 1mm deviation
break;
case 7:
*rangevalue_out=(uint16_t)(rangevalue - ogalgo_data_inst[pos].og_val); // Category Overranging, 2mm deviation
break;
case 8:
*rangevalue_out=(uint16_t)(rangevalue - ogalgo_data_inst[pos].og_val); // Category Overranging, 4mm deviation

AN5851 - Rev 1 page 11/20

 AN5851
Reference code for nonlinearity correction algorithm

break;
case 9:
*rangevalue_out=(uint16_t)(rangevalue - ogalgo_data_inst[pos].og_val); // Category Overranging, 4mm deviation
break;
default:
printf("Not meet condition");
}

AN5851 - Rev 1 page 12/20

 AN5851
Test setup and results

4 Test setup and results

The test setup uses a narrow transparent glass bottle, with a width of 220 mm and radius of 50 mm. The sensor is
mounted slightly above the bottle, but all the measurements are relative to the initial height. The test is conducted
on the water level with a timing budget of 50 ms. This application is not usually time critical, so the rolling average
along with a bigger timing budget can be used to allow the ToF to accumulate more signals that ultimately
enhance ranging performance.

Figure 13. Water bottle with black mat background

4.1 Test results without nonlinearity correction algorithm and without cover glass
The following graph shows the ranging data with different water levels. It can be observed that there is a
nonlinearity between expected and actual measurement in low water level.

Figure 14. Without nonlinearity algorithm and without cover glass

AN5851 - Rev 1 page 13/20

 AN5851
Test results with offset calibration, without nonlinearity correction algorithm, and with cover glass

4.2 Test results with offset calibration, without nonlinearity correction algorithm,
and with cover glass
The following graph illustrates the data with offset calibration and with cover glass, but without nonlinearity
correction algorithm. The offset calibration is done with the same setup, maintaining 100 mm distance from sensor
to water level. The results are more accurate, yet over range persists in low water levels. Offset calibration is done
using the feature available and described in the VL53L4CD UM2931.

Figure 15. With offset calibration, without nonlinearity correction algorithm, and with cover glass

4.3 Test results with offset calibration, with nonlinearity correction algorithm, and
with cover glass
The following figure shows that the measurements are quite linear between expected and actual measurement
after adopting the nonlinearity correction algorithm and offset calibration. It may be further fine tuned depending
on the specific container setup.

Figure 16. With offset calibration, with nonlinearity correction algorithm, and with cover glass

AN5851 - Rev 1 page 14/20

 AN5851
Non transparent liquid results

4.4 Non transparent liquid results

Tests on different types of non transparent liquids such as milk and syrup show much better results than
transparent liquid. It is recommended to adopt the nonlinearity algorithm for other transparent liquids like oil,
gasoline, soda to get high accuracy in ranging measurement.

AN5851 - Rev 1 page 15/20

 AN5851
Description of the experiment context

5 Description of the experiment context

5.1 Hardware required

The VL53L4CD standard package is used as a proof of concept.
It can be found on st.com and includes:
• X-NUCLEO-53L4A1 expansion board for NUCLEO boards
• P-NUCLEO-53L4A1 expansion pack including the X-NUCLEO-53L4A1 and the NUCLEO-F401RE

Figure 17. VL53L4CD expansion board with NUCLEO F401RE board

5.2 Cover glass

ST recommends using a light blocker cover glass to get better results compared to one window standard
cover glass. Furthermore, the light blocker cover glass helps to maintain good performance results in a humid
environment.

Figure 18. Light blocker cover glass by Hornix

5.3 Software requirements

The GUI, ST characterization tool, and application example code described in this application note are available
on st.com under the reference STSW-IMG039_L4CD. ST recommends using the GUI for evaluation.

AN5851 - Rev 1 page 16/20

 AN5851
Known limitations

6 Known limitations

The following known limitations have been identified and should be considered for final setup:
• Ranging is not reliable if the distance between the sensor and the water level is less than 20 mm.
• A highly reflective bottom impacts the accuracy for low water levels.
The lower reflectivity base has better linearity in ranging than high reflectivity. It is recommended to
use a low reflective base, like black mat reflector when selecting a container. Figure 19. Water level
measurement with different colored base and no compensation shows the effect of the base's reflectivity on
the ranging measurement without compensation.
• Ranging results are not reliable if water or liquid is in motion (for example ripples on the surface), or if
bubbles form on the surface.

Figure 19. Water level measurement with different colored base and no compensation

AN5851 - Rev 1 page 17/20

 AN5851

Revision history
Table 2. Document revision history

Date Version Changes

27-Oct-2022 1 Initial release

AN5851 - Rev 1 page 18/20

 AN5851
Contents

Contents
1 Observations on water. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
2 ST recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3
2.1 Container properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.2 Container reflectivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
3 ST solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5
3.1 Nonlinearity correction algorithm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
3.2 Ranging result characterization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
3.3 System characterization using ST tool . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
3.4 Nonlinearity correction algorithm flow. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
3.5 Reference code for nonlinearity correction algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
4 Test setup and results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .13
4.1 Test results without nonlinearity correction algorithm and without cover glass . . . . . . . . . . . 13
4.2 Test results with offset calibration, without nonlinearity correction algorithm, and with cover
glass . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
4.3 Test results with offset calibration, with nonlinearity correction algorithm, and with cover
glass . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
4.4 Non transparent liquid results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
5 Description of the experiment context. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .16
5.1 Hardware required . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
5.2 Cover glass . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
5.3 Software requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
6 Known limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .17
Revision history . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .18

AN5851 - Rev 1 page 19/20

 AN5851

IMPORTANT NOTICE – READ CAREFULLY

STMicroelectronics NV and its subsidiaries (“ST”) reserve the right to make changes, corrections, enhancements, modifications, and improvements to ST
products and/or to this document at any time without notice. Purchasers should obtain the latest relevant information on ST products before placing orders. ST
products are sold pursuant to ST’s terms and conditions of sale in place at the time of order acknowledgment.
Purchasers are solely responsible for the choice, selection, and use of ST products and ST assumes no liability for application assistance or the design of
purchasers’ products.
No license, express or implied, to any intellectual property right is granted by ST herein.
Resale of ST products with provisions different from the information set forth herein shall void any warranty granted by ST for such product.
ST and the ST logo are trademarks of ST. For additional information about ST trademarks, refer to www.st.com/trademarks. All other product or service names
are the property of their respective owners.
Information in this document supersedes and replaces information previously supplied in any prior versions of this document.
© 2022 STMicroelectronics – All rights reserved

AN5851 - Rev 1 page 20/20