cd00211314

AN2834
Application note
How to get the best ADC accuracy in STM32Fx Series and
STM32L1 Series devices
Introduction
The STM32Fx Series and STM32L1 Series microcontroller families embed up to four
advanced 12-bit ADCs (depending on the device). A self-calibration feature is provided to
enhance ADC accuracy versus environmental condition changes.
In applications involving analog-to-digital conversion, ADC accuracy has an impact on the
overall system quality and efficiency. To improve this accuracy, you need to understand the
errors associated with the ADC and the parameters affecting them.
ADC accuracy does not depend on ADC performance and features alone, it depends on the
overall application design around the ADC.
This application note aims to help understanding of ADC errors and how to enhance ADC
accuracy. It is divided into three main parts:
 a simplified description of ADC internal structure to aid understanding of ADC operation
and related ADC parameters
 explanations of the different types and sources of ADC errors related to the ADC design
and to external ADC parameters such as the external hardware design
 recommendations on how to minimize these errors, focusing on hardware and software
methods
This document applies to the products listed in Table 1 which are referred to as STM32x
throughout this document.
Table 1. Applicable products
Product family
Microcontrollers
September 2013
Part numbers
–
–
–
–
–
–
STM32F0xx
STM32F1xx
STM32F2xx
STM32F3xx
STM32F4xx
STM32L1xx
DocID15067 Rev 2
1/45
www.st.com
Contents
AN2834
Contents
1
ADC internal principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.1
2
ADC errors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.1
2.2
3
2/45
SAR ADC internal structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
ADC errors related to the ADC itself . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.1.1
Offset error . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.1.2
Gain error . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.1.3
Differential linearity error . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.1.4
Integral linearity error . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.1.5
Total unadjusted error . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
ADC errors related to its environment . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.2.1
Reference voltage noise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.2.2
Reference voltage / power supply regulation . . . . . . . . . . . . . . . . . . . . . 15
2.2.3
External reference voltage parameters . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.2.4
Analog input signal noise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.2.5
ADC dynamic range bad match for maximum input signal amplitude . . 16
2.2.6
Effect of the analog signal source resistance . . . . . . . . . . . . . . . . . . . . 16
2.2.7
Effect of source capacitance and parasitic capacitance of the PCB . . . 17
2.2.8
Injection current effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.2.9
Temperature influence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.2.10
I/O pin crosstalk . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.2.11
EMI-induced noise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
How to get the best ADC accuracy . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.1
Reduce the effects of ADC-related ADC errors . . . . . . . . . . . . . . . . . . . . 20
3.2
Minimize ADC errors related to external environment of ADC . . . . . . . . . 20
3.2.1
Reference voltage / Power supply noise minimization . . . . . . . . . . . . . 20
3.2.2
Reference voltage / Power-supply regulation . . . . . . . . . . . . . . . . . . . . 22
3.2.3
Analog-input signal noise elimination . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.2.4
Adding white noise or triangular sweep to improve resolution . . . . . . . . 22
3.2.5
Matching the ADC dynamic range to the maximum signal amplitude . . 23
3.2.6
Analog source resistance calculation . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.2.7
Source frequency condition vs. source and parasitic capacitors . . . . . . 27
3.2.8
Temperature-effect compensation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
DocID15067 Rev 2
AN2834
Contents
3.3
3.4
3.2.9
Minimizing injection current . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.2.10
Minimizing I/O pin crosstalk . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.2.11
EMI-induced noise reduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.2.12
PCB layout recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.2.13
Component placement and routing . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
Software methods to improve precision . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.3.1
Averaging samples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.3.2
Digital signal filtering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.3.3
FFT for AC measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.3.4
ADC calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.3.5
Minimizing internal CPU noise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
High impedance source measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.4.1
ADC input stage problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.4.2
Explanation of the behavior . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.4.3
Minimizing added errors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.4.4
Source of described problem - ADC design . . . . . . . . . . . . . . . . . . . . . 41
4
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
5
Revision history . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
DocID15067 Rev 2
3/45
3
List of figures
AN2834
List of figures
Figure 1.
Figure 2.
Figure 3.
Figure 4.
Figure 5.
Figure 6.
Figure 7.
Figure 8.
Figure 9.
Figure 10.
Figure 11.
Figure 12.
Figure 13.
Figure 14.
Figure 15.
Figure 16.
Figure 17.
Figure 18.
Figure 19.
Figure 20.
Figure 21.
Figure 22.
Figure 23.
Figure 24.
Figure 25.
Figure 26.
Figure 27.
Figure 28.
Figure 29.
Figure 30.
Figure 31.
Figure 32.
Figure 33.
Figure 34.
Figure 35.
Figure 36.
Figure 37.
Figure 38.
Figure 39.
4/45
Basic schematic of SAR switched-capacitor ADC (example for 10-bit ADC) . . . . . . . . . . . . 5
Sample state . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Hold state . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Step 1: Compare with VREF/2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Step 2: If MSB = 1, then compare with ¾ VREF. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Step 2: If MSB = 0, then compare with ¼ VREF. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Positive offset error representation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Negative offset error representation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Positive gain error representation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Negative gain error representation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Differential linearity error representation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Integral linearity error representation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Total unadjusted error. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Input signal amplitude vs. ADC dynamic range . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
Analog signal source resistance effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
Analog input with RAIN, CAIN and Cp . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
Effect of injection current . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
Crosstalk between I/O pins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
EMI sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Power supply and reference decoupling for 100- and 144-pin packages . . . . . . . . . . . . . . 21
Power supply decoupling for 36-, 48- and 64-pin packages . . . . . . . . . . . . . . . . . . . . . . . . 21
Simple quasi-triangular source using a microcontroller output . . . . . . . . . . . . . . . . . . . . . . 23
Selecting the reference voltage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
Preamplification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
Worst case error: VAIN = VREF+. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
Recommended values for RAIN and CAIN vs. source frequency FAIN . . . . . . . . . . . . . . . . . . . . . . 27
Crosstalk between I/O pins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
Shielding technique . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
Separating the analog and digital layouts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
Separating the analog and digital supplies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
Typical voltage source connection to ADC input . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
Noise observed on ADC input pin during ADC conversions . . . . . . . . . . . . . . . . . . . . . . . . 35
ADC simplified schematic of input stage - sample and hold circuit. . . . . . . . . . . . . . . . . . . 36
ADC input pin noise spikes from internal charge during sampling process . . . . . . . . . . . . 36
Effect of sampling time extension . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
Charging the external capacitor with too short time between conversions . . . . . . . . . . . . . 38
Implementation of sampling switch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
Parasitic capacitances of sampling switch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
Parasitic current example inside ADC structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
DocID15067 Rev 2
AN2834
ADC internal principle
1
ADC internal principle
1.1
SAR ADC internal structure
The ADC in STM32x microcontrollers uses the SAR (successive approximation register)
principle, by which the conversion is performed in several steps. The number of conversion
steps is equal to the number of bits in the ADC converter. Each step is driven by the ADC
clock. Each ADC clock produces one bit from result to output. ADC internal design is a
switched-capacitor type.
The following figures (Figure 1 to Figure 6) explain the principle of ADC operation. The
example given below shows only the first steps of approximation but the process continues
till the LSB is reached.
Figure 1. Basic schematic of SAR switched-capacitor ADC (example for 10-bit ADC)
VREF
VIN
Sa
S1
S2
S3
S4
S5
S6
S7
S8
S9
S10
S11
C
C/2
C/4
C/8
C/16
C/32
C/64
C/128
C/256
C/512
C/512
Sb
A
D PR Q
ADC Data
CLK
CLR
ADC Clk
ai17097
1. Basic ADC schematic with digital output.
DocID15067 Rev 2
5/45
44
ADC internal principle
AN2834
Figure 2. Sample state
VREF
VIN
Sa
S1
S2
S3
S4
S5
S6
S7
S8
S9
S10
S11
C
C/2
C/4
C/8
C/16
C/32
C/64
C/128
C/256
C/512
C/512
D PR Q
A
Sb
ADC Data
CLK
CLR
ADC Clk
VCOMP= 0
2C
VIN
A
Equivalent circuit:
ai17098
1. Sample state - capacitors are charging to VIN voltage. Sa switch to VIN, Sb switch closed during sampling
time.
Figure 3. Hold state
VREF
VIN
Sa
S1
S2
S3
S4
S5
S6
S7
S8
S9
S10
S11
C
C/2
C/4
C/8
C/16
C/32
C/64
C/128
C/256
C/512
C/512
A
Sb
D PR Q
ADC Data
CLK
CLR
ADC Clk
VCOMP= –V IN
2C
Equivalent circuit:
A
ai17099
1. Hold state - input is disconnected - capacitors hold input voltage. Sb switch is open, then S1-S11 switched
to ground and Sa switched to VREF.
6/45
DocID15067 Rev 2
AN2834
ADC internal principle
Figure 4. Step 1: Compare with VREF/2
VREF
VIN
Sa
S1
S2
S3
S4
S5
S6
S7
S8
S9
S10
S11
C
C/2
C/4
C/8
C/16
C/32
C/64
C/128
C/256
C/512
C/512
D PR Q
A
Sb
ADC Data
CLK
CLR
ADC Clk
VCOMP= –V IN+ V REF/2
C
VREF
A
C
Equivalent circuit:
ai17800
1. First approximation step. S1 switched to VREF.
Figure 5. Step 2: If MSB = 1, then compare with ¾ VREF
VREF
VIN
Sa
S1
S2
S3
S4
S5
S6
S7
S8
S9
S10
S11
C
C/2
C/4
C/8
C/16
C/32
C/64
C/128
C/256
C/512
C/512
A
Sb
D PR Q
ADC Data
CLK
CLR
ADC Clk
VCOMP= –V IN+ ¾ V REF
C/2
VREF
Equivalent circuit:
3C/2
A
ai17801
1. Compare with ¾VREF; if MSB =1. S1 switched back to ground. S2 switched to VREF.
DocID15067 Rev 2
7/45
44
ADC internal principle
AN2834
Figure 6. Step 2: If MSB = 0, then compare with ¼ VREF
VREF
VIN
Sa
S1
S2
S3
S4
S5
S6
S7
S8
S9
S10
S11
C
C/2
C/4
C/8
C/16
C/32
C/64
C/128
C/256
C/512
C/512
A
Sb
D PR Q
ADC Data
CLK
CLR
ADC Clk
VCOMP= –V IN+ ¼ V REF
3C/2
VREF
Equivalent circuit:
C/2
A
ai17802
1. Compare with ¼ VREF; if MSB =0. S1 remained switched to ground. S2 switched to VREF.
8/45
DocID15067 Rev 2
AN2834
2
ADC errors
ADC errors
This section lists the main errors which have an effect on A/D conversion accuracy. These
types of error occur in all A/D converters and conversion quality depends on eliminating
them. Values for these errors are specified in the ADC characteristics section of the STM32x
device datasheets.
Different accuracy error types are specified for the STM32x ADC. Accuracy errors are
normally expressed as multiples of 1 LSB for easy reference. The resolution in terms of
voltage depends on the reference voltage. The error in terms of voltage is calculated by
multiplying the number of LSBs with the voltage corresponding to 1LSB (1LSB = VREF+/212
or VDDA/212).
2.1
ADC errors related to the ADC itself
2.1.1
Offset error
It is defined as the deviation between the first actual transition and the first ideal transition.
The first transition occurs when the digital output of the ADC changes from 0 to 1. Ideally,
when the analog input is between 0.5 LSB and 1.5LSB, the digital output should be 1.
Ideally still, the first transition occurs at 0.5 LSB. The offset error is denoted by EO. Offset
error can easily be calibrated by the application firmware.
Example
For the STM32x ADC, the smallest detectable incremental change in voltage is expressed
in terms of LSBs:
1 LSB = VREF+/4096 (on some packages is VREF+ = VDDA).
If VREF+ = 3.3 V, ideally, the input of 402.8 µV (0.5 LSB = 0.5 × 805.6 µV) should lead to the
generation of a digital output of 1. In practice, however, the ADC may still show the reading
as 0. If a digital output of 1 is obtained from an analog input of 550 µV, then:
Offset error = Actual transition – Ideal transition
EO = 550 µV – 402.8 µV = 141.2 µV
EO = 141.2 µV / 805.6 µV = 0.17 LSB
When an analog input voltage greater than 0.5LSB generates the first transition, the offset
error is positive. Figure 7 shows a positive offset error.
DocID15067 Rev 2
9/45
44
ADC errors
AN2834
Figure 7. Positive offset error representation
Digital output
Ideal transfer curve
EO > 0
2
Actual transfer curve
1
0
VAIN
0.5LSB
ai15475
When an analog input voltage of less than 0.5LSB generates the first transition, the offset
error is negative. Figure 8 shows a negative offset error.
If the analog input voltage VAIN = VSSA and the ADC generates a nonzero digital output, the
offset error is negative. This means that a negative voltage generates the first transition.
Figure 8. Negative offset error representation
Digital output
Ideal transfer curve
EO > 0
Actual transfer curve
2
1
0
0.5LSB
VAIN
ai15476
10/45
DocID15067 Rev 2
AN2834
2.1.2
ADC errors
Gain error
Gain error is defined as the deviation between the last actual transition and the last ideal
transition. Gain error is represented as EG.
The last actual transition is the transition from FFEh to FFFh. Ideally, there should be a
transition from FFEh to FFFh, when the analog input is equal to VREF+ – 0.5LSB. So for
VREF+= 3.3 V, the last ideal transition should be at 3.299597 V.
If the ADC provides the FFFh reading for VAIN < VREF+ – 0.5LSB, then we have a negative
gain error.
Example
Gain error EG = Last actual transition – ideal transition
If VREF+ = 3.3 V and VAIN = 3.298435 V generates a transition from FFE to FFF then,
EG = 3.298435 V – 3.299597 V
EG = –1162 µV
EG = (–1162 µV / 805.6 µV) LSB = –1.44 LSB
If we do not get a full scale reading (FFFh) for VAIN equal to VREF+, the gain error is positive.
This means that a voltage greater than VREF+ will cause the last transition. Figure 9 shows a
positive gain error and Figure 10, a negative gain error.
Figure 9. Positive gain error representation
Digital output
4095
EG > 0
Ideal transfer curve
Actual transfer curve
4094.5
LSB
VAIN
ai15477
Figure 10. Negative gain error representation
Digital output
4095
EG < 0
Ideal transfer curve
Actual transfer curve
4094.5
LSB
DocID15067 Rev 2
VAIN
ai15478
11/45
44
ADC errors
2.1.3
AN2834
Differential linearity error
The differential linearity error (DLE) is defined as the maximum deviation between the actual
and ideal steps. Here ‘ideal’ is not used for the ideal transfer curve but for the resolution of
the ADC. The DLE is denoted by ED. It is represented in Figure 11.
ED = Actual step width – 1LSB
Ideally, an analog input voltage change of 1LSB should cause a change in the digital code. If
an analog input voltage greater than 1LSB is required for a change in digital code, the ADC
has a differential linearity error. The DLE therefore corresponds to the maximum additional
voltage that is required to change from one digital code to the next.
The DLE is also known as the differential non-linearity (DNL) error.
Example
A given digital output should correspond to an analog input range. Ideally, the step width
should be 1LSB. Let us assume that we get the same digital output over an analog input
voltage range of 1.9998 V to 2.0014 V, the step width will be 2.0014 V – 1.9998 V = 1.6 mV.
ED is thus the voltage difference between the higher (2.0014 V) and lower (1.9998 V)
analog voltages less the voltage corresponding to 1LSB.
Figure 11. Differential linearity error representation
Actual
step
width
Digital output
1LSB
ED > 0
1LSB
Actual
step
width
ED < 0
Ideal transfer curve
Actual transfer curve
VAIN
ai15479
If VREF+ = 3.3 V, an analog input of 1.9998 V (9B1h) can provide results varying between
9B0h and 9B2h. Similarly, for an input of 2.0014 V (9B3h), the results may vary between
9B2h and 9B4h.
Thus, the total voltage variation corresponding to the 9B2h step is:
9B3h – 9B1h, that is, 2.0014 V – 1.9998V = 1.6 mV (1660 µV)
ED= 1660 µV – 805.6 µV
ED= 854.4 µV
ED= (854.4 µV/805.6 µV) LSB
ED= 1.06 LSB
Let us assume that no voltage greater than 2.0014 V will result in the 9B2h digital code.
When the step width is less than 1LSB, ED is negative.
12/45
DocID15067 Rev 2
AN2834
2.1.4
ADC errors
Integral linearity error
The integral linearity error is the maximum deviation between any actual transition and the
endpoint correlation line. The ILE is denoted by EL. It is represented in Figure 12.
The endpoint correlation line can be defined as the line on the A/D transfer curve that
connects the first actual transition with the last actual transition. EL is the deviation from this
line for each transition. The endpoint correlation line thus corresponds to the actual transfer
curve and has no relation to the ideal transfer curve.
The ILE is also known as the integral non linearity error (INL). The ILE is the integral of the
DLE over the whole range.
Figure 12. Integral linearity error representation
Digital output
4095
EL
2
Actual transfer curve
1
0
550 μV
3.298435 V
VAIN
ai15480
Example
If the first transition from 0 to 1 occurs at 550 µV and the last transition (FFEh to FFFh)
occurs at 3.298435 V (gain error), then the line on the transfer curve connecting the actual
digital codes 1h and FFFh will be the endpoint correlation line.
DocID15067 Rev 2
13/45
44
ADC errors
2.1.5
AN2834
Total unadjusted error
The total unadjusted error (TUE) is defined as the maximum deviation between the actual
and the ideal transfer curves. It is a parameter that specifies the total errors that may occur,
causing maximum deviation between the ideal digital output and the actual digital output. It
is the maximum deviation recorded between the ideal expected value and the actual value
obtained from the ADC for any input voltage.
The TUE is denoted by ET. It is represented in Figure 13.
The TUE is not the sum of EO, EG, EL, ED. The offset error affects the digital result at lower
voltages whereas the gain error affects the digital output for higher voltages.
Example
If VREF+ = 3.3 V and VAIN = 2 V, the ideal result is 9B2h. But if, on conversion, we get the
result 9B4h, the deviation may result from the offset since the DLE and ILE errors occur
simultaneously.
TUE = absolute (actual value – ideal case value) = 9B4h – 9B2h = 2h = 2LSB
Figure 13. Total unadjusted error
Digital output
ET
Ideal transfer curve
Actual transfer curve
VAIN
ai15481
14/45
DocID15067 Rev 2
AN2834
ADC errors
2.2
ADC errors related to its environment
2.2.1
Reference voltage noise
As the ADC output is the ratio between the analog signal voltage and the reference voltage,
any noise on the analog reference will cause a change in the converted digital value. The
VDDA analog power supply is used on some packages as the reference voltage (VREF+) so
the quality of this power supply has influence to ADC error.
For example, with an analog reference of 3.3 V (VREF+ = VDDA) and a 1 V signal input, the
converted result is (1/3.3) × 4095 = 4D9h
But with a 40 mV peak-to-peak ripple in the analog reference, the converted value becomes
(1/3.34) × 4095 = 4CAh (with VREF+ at its peak).
Error = 4D9 – 4CA = 15 LSB
The SMPS (switch-mode power supply) normally has internal fast-switching power
transistors. This introduces high-frequency noise in the output. The switching noise is in the
range of 15 kHz to 1 MHz.
2.2.2
Reference voltage / power supply regulation
Power supply regulation is very important for ADC accuracy since the conversion result is
the ratio of the analog input voltage to the VREF+ value.
If the power supply output decreases when connected to VDDA or VREF+ due to the loads on
these inputs and to its output impedance, an error will be introduced in the conversion result.
N
V AIN  2 – 1 
Digital code = ---------------------------------, where N is the resolution of the ADC (in our case N = 12).
V REF+
If the reference voltage changes, the digital result changes too.
For example:
If the supply used is a reference voltage of 3.3 V and VAIN = 1 V, the digital output is:
12
1  2 – 1
Digital output = ---------------------------------- = 4D9h
3.3
If the voltage supply provides a voltage equal to 3.292 V (after its output connection to
VREF+), then:
12
1  2 – 1
Digital output = ---------------------------------- = 4DCh
3.292
The error introduced by the voltage drop is: 4DC – 4D9 = 3LSB .
DocID15067 Rev 2
15/45
44
ADC errors
2.2.3
AN2834
External reference voltage parameters
In case of usage external source for reference voltage (on VREF+ pin) there are important
parameters of this external reference source. It is important to look at three reference
voltage specifications: temperature drift, voltage noise, long term stability.
2.2.4
Analog input signal noise
Small but high-frequency signal variation can result in big conversion errors during sampling
time. This noise is generated by electrical devices, for example, motors, engine ignition,
power lines, and so on, and affects the source signal (for example a sensor) by adding an
unwanted signal. As a consequence, the conversion results of the ADC are not accurate.
2.2.5
ADC dynamic range bad match for maximum input signal amplitude
It is very important that the ADC dynamic range matches the maximum amplitude of the
signal to be converted to have the maximum ADC conversion precision. Let us assume that
the signal to be converted varies between 0 V to 2.5 V and that VREF+ is equal to 3.3 V. The
maximum signal value converted by the ADC is 3102 (2.5 V) as shown in Figure 14. In this
case, there are 993 unused transitions (4095 – 3102 = 993). This implies a loss in the
converted signal accuracy.
See Section 3.2.5: Matching the ADC dynamic range to the maximum signal amplitude on
page 23 for details on how to have the ADC dynamic range matching the maximum input
signal amplitude.
Figure 14. Input signal amplitude vs. ADC dynamic range
V
VREF+ = 3.3 V
(4095h)
unused ADC transition range
2.5 V
(3102h)
t
2.2.6
ai15499
Effect of the analog signal source resistance
The impedance of the analog signal source, or series resistance (RAIN), between the source
and pin causes a voltage drop across it because of the current flowing into the pin. The
charging of the internal sampling capacitor CADC is controlled by switches with resistance
RADC.
With the addition of source resistance (with RADC), the time required to fully charge the hold
capacitor increases. Figure 15 shows the analog signal source resistance effect.
The effective charging of CADC is governed by RADC + RAIN, so the charging time constant
becomes tc = (RADC+RAIN) × CADC. If the sampling time is less than the time required to fully
charge the CADC through RADC + RAIN (ts < tc), the digital value converted by the ADC is
less than the actual value.
16/45
DocID15067 Rev 2
AN2834
ADC errors
Figure 15. Analog signal source resistance effect
STM32
VAIN
R AIN
RADC
AIN x
CADC
VSSA
12-bit ADC
Vc
VAIN
Vc
t
tc
ai15482
1. tc is the time taken by the CADC capacitor to fully charge: Vc = VAIN (with max.1/2 lsb error)
Vc: capacitor (CADC) voltage
tc = (RADC + RAIN) × CADC
2.2.7
Effect of source capacitance and parasitic capacitance of the PCB
When converting analog signals, it is necessary to account for the capacitance at the source
and the parasitic capacitance seen on the analog input pin (refer to Figure 16). The source
resistance and capacitance form an RC network and the ADC conversion results may not be
accurate unless the external capacitor (CAIN + Cp) is fully charged to the level of the input
voltage. The greater value of (CAIN + Cp), the more limited the source frequency.
The external capacitance at the source and the parasitic capacitance are denoted by CAIN
and Cp, respectively.
Figure 16. Analog input with RAIN, CAIN and Cp
STM32
Source
VAIN
RAIN
CAIN
AINx
CP
ai15483
DocID15067 Rev 2
17/45
44
ADC errors
2.2.8
AN2834
Injection current effect
A negative injection current on any analog pin (or a closely positioned digital input pin) may
introduce leakage current into the ADC input. The worst case is the adjacent analog
channel. A negative injection current is introduced when VAIN < VSS, causing current to flow
out from the I/O pin. This is illustrated in Figure 17.
Figure 17. Effect of injection current
VAIN0
STM32
RAIN0
AIN0
leakage
current
AIN1
Injection
current
VAIN < VSS
VSSA
2.2.9
ai15484
Temperature influence
The temperature has a major influence on ADC accuracy. Mainly it leads to two major
errors: offset error drift and gain error drift. Those errors can be compensated in the
microcontroller firmware (refer to Section 3.2.8 for the temperature-compensation methods).
2.2.10
I/O pin crosstalk
Switching the I/Os may induce some noise in the analog input of the ADC due to capacitive
coupling between I/Os. Crosstalk may be introduced by PCB tracks that run close to each
other or that cross each other.
Internally switching digital signals and I/Os introduces high-frequency noise. Switching highsink I/Os may induce some voltage dips in the power supply caused by current surges. A
digital track that crosses an analog input track on the PCB may affect the analog signal (see
Figure 18).
Figure 18. Crosstalk between I/O pins
Analog
in
STM32
Digital
I/O
Analog
in
STM32
Digital
I/O
Case 1
Case 2
ai15485
Case 1: Digital and analog signal tracks that pass close to each other.
Case 2: Digital and analog signal tracks that cross each other on a different PCB side.
18/45
DocID15067 Rev 2
AN2834
2.2.11
ADC errors
EMI-induced noise
Electromagnetic emissions from neighboring circuits may introduce high-frequency noise in
an analog signal because the PCB tracks may act like an antenna (See Figure 19.).
Figure 19. EMI sources
Electromagnetic
noise
I/O
coupled
noise
STM32
ADC
Noise induced
from PCB tracks
Internal
noise
ai15486
DocID15067 Rev 2
19/45
44
How to get the best ADC accuracy
AN2834
3
How to get the best ADC accuracy
3.1
Reduce the effects of ADC-related ADC errors
The TUE is not the sum of all the errors EO, EG, EL, ED. It is the maximum deviation that can
occur between the ideal and actual digital values. It can result from one or more errors
occurring simultaneously.
As the ILE is the integral of the DLE, it can be considered as the indicator of the maximum
error. Do not add the DLE and ILE together to calculate the maximum error that may occur
at any digital step.
The maximum error values specified in the datasheet are the worst error values measured
in laboratory test environment over the given voltage and temperature range (see device
datasheet).
The ILE and DLE are dependent on the ADC design. It is difficult to calibrate them. They
can be calibrated by the measured ADC curve stored in the microcontroller memory but this
needs calibration of each individual device in final application.
Offset and gain errors can be easily compensated using the STM32x ADC self-calibration
feature or by microcontroller firmware.
3.2
Minimize ADC errors related to external environment of ADC
3.2.1
Reference voltage / Power supply noise minimization
Power supply side
Linear regulators have a better output in terms of noise. The mains must be stepped down,
rectified and filtered, then fed to linear regulators. It is highly recommended to connect the
filter capacitors to the rectifier output. Please refer to the datasheet of the used linear
regulator.
If you are using a switching power supply, it is recommended to have a linear regulator to
supply the analog stage.
It is recommended to connect capacitors with good high-frequency characteristics between
the power and ground lines. That is, a 0.1 µF and a 1 to 10 µF capacitor should be placed
close to the power source.
The capacitors allow the AC signals to pass through them. The small-value capacitors filter
high-frequency noise and the high-value capacitors filter low-frequency noise. Ceramic
capacitors are generally available in small values (1 pF to 0.1 µF) and with small voltage
ratings (16 V to 50 V). It is recommended to place them close to the main supply (VDD and
VSS) and analog supply (VDDA and VSSA) pins. They filter the noise induced in the PCB
tracks. Small capacitors can react fast to current surges and discharge quickly for fastcurrent requirements.
Tantalum capacitors can also be used along with ceramic capacitors. To filter low-frequency
noise, you can use high-value capacitors (10 µF to 100 µF), which are generally electrolytic.
It is recommended to put them near the power source.
To filter high-frequency noise, you can use a ferrite inductance in series with the power
supply. This solution leads to very low (negligible) DC loss unless the current is high
20/45
DocID15067 Rev 2
AN2834
How to get the best ADC accuracy
because the series resistance of the wire is very low. At high frequencies, however, the
impedance is high.
STM32x side
In most STM32x microcontrollers the VDD and VSS pins are placed close to each other. So
are the VREF+ and VSSA pins. A capacitor can therefore be connected very close to the
microcontroller with very short leads. For multiple VDD and VSS pins, use separate
decoupling capacitors.
The VDDA pin must be connected to two external decoupling capacitors (10 nF Ceramic +
1 µF Tantalum or Ceramic). See Figure 20 and Figure 21 for decoupling examples.
For STM32x devices delivered in 100/144-pin packages, it is possible to improve the
accuracy on low-voltage inputs by connecting a separate external ADC reference voltage
input on VREF+ (refer to Section 3.2.5). The voltage on VREF+ may range from 2.4 V to
VDDA. If a separate, external reference voltage is applied on VREF+, two 10 nF and 1 µF
capacitors must be connected on this pin. In all cases, VREF+ must be kept between 2.4 V
and VDDA.
Figure 20. Power supply and reference decoupling for 100- and 144-pin packages
VDDA
VREF+
1 μF // 10 nF
1 μF // 10 nF
VDDA
1 μF // 10 nF
VREF+
VREF–
VREF–
VSSA
VSSA
VREF+ not connected to VDDA
VREF+ connected to VDDA
ai15487
Figure 21. Power supply decoupling for 36-, 48- and 64-pin packages
VDDA
1 μF // 10 nF
VSSA
ai15488
DocID15067 Rev 2
21/45
44
How to get the best ADC accuracy
3.2.2
AN2834
Reference voltage / Power-supply regulation
The power supply should have good line and load regulation since the ADC uses VREF+ or
VDDA as the analog reference and the digital value is the ratio of the analog input signal to
this voltage reference. VREF+ must thus remain stable at different loads.
Whenever the load is increased by switching on a part of the circuit, the increase in current
must not cause the voltage to decrease. If the voltage remains stable over a wide current
range, the power supply has good load regulation.
For example: for the LD1086D2M33 voltage regulator, the line regulation is 0.035% typical
when VIN varies from 2.8 V to 16.5 V (when Iload = 10 mA), and the load regulation is 0.2%
when Iload varies from 0 to 1.5 A (please refer to the LD1086 series datasheet for details).
The lower the line regulation value, the better the regulation. Similarly, the lower the load
regulation value, the better the regulation and the stability of the voltage output.
It is also possible to use a reference voltage for VREF+, for instance the LM236, which is a
voltage reference diode of 2.5 V (refer to LM236 datasheet for more details).
3.2.3
Analog-input signal noise elimination
Averaging method
Averaging is a simple technique where you sample an analog input several times and take
the average of the results by software. This technique is helpful to eliminate the effect of
noise on the analog input in case of an analog voltage that does not change often.
The average has to be made on several readings that all correspond to the same analog
input voltage. Make sure that the analog input remains at the same voltage during the time
period when the conversions are done, otherwise you will add up digital values
corresponding to different analog inputs, and you will introduce errors.
Adding an external filter
Adding an external RC filter eliminates the high frequency. An expensive filter is not needed
to deal with a signal that has frequency components above the frequency range of interest.
In this case, a relatively simple lowpass filter with a cutoff frequency fC just above the
frequency range of interest will suffice to limit noise and aliasing. A sampling rate consistent
with the highest frequency of interest will suffice, typically 2 to 5 times fC.
Note:
The R and C that form the external filter should have values that match the conditions
described in Section 3.2.4 and Section 3.2.7.
3.2.4
Adding white noise or triangular sweep to improve resolution
This method combines hardware and software techniques to improve precision. From a
software point of view, this method uses averaging (oversampling) and from a hardware
point of view, it uses signal modification/spreading/dithering.
Averaging can be used in cases where the input signal is noisy (some signal change is
necessary in order to be able to calculate an average) and the requirement is to obtain the
mean value of a signal. A problem appears when the input signal is a very stable voltage
without noise. In this case, when the input signal is measured, each data sample is the
same. This is because the input signal level is somewhere between two ADC word levels
(e.g. between 0x14A and 0x14B). Therefore it is not possible to determine the input voltage
level more precisely (e.g. if the level is near to 0x14A or near to 0x14B level).
22/45
DocID15067 Rev 2
AN2834
How to get the best ADC accuracy
The solution is to add noise or some signal change (with uniform signal distribution e.g.
triangular sweep) to the input signal which pushes its level across 1-bit ADC level (so that
the signal level changes below 0x14A and above 0x14B level). This causes the ADC results
to vary. Applying software averaging to the different ADC results, produces the mean value
of the original input signal.
Example of implementation of this method is done by using a triangular generator with RC
coupling to the input signal (white noise generation is more complicated). Care must be
taken not to modify the mean value of the original input signal (so, capacitive coupling must
be used).
A very simple implementation of the quasi-triangular source which is generated directly by
the STM32x microcontroller is on Figure 22.
Figure 22. Simple quasi-triangular source using a microcontroller output
Uout
VDD
t
R1
OUT
C
R2
Ain
Vin
MCU
Uin
Vin
t
3.2.5
ai170803
Matching the ADC dynamic range to the maximum signal amplitude
This method improves accuracy by a proper selection of the reference voltage or by using a
preamplifier stage to obtain the maximum possible resolution using the full ADC output
range.
Selecting a reference voltage (method for devices delivered in 100-/144-pin
packages only)
The reference voltage is selected in the expected range of the signal to be measured. If the
measured signal has an offset, then the reference voltage should also have a similar offset.
If the measured signal has a defined maximum amplitude, then the reference voltage should
also have a similar maximum value. By matching this reference voltage to the measurement
signal range, we obtain the maximum possible resolution using the full ADC output range.
In STM32x devices delivered in 100- and 144-pin packages, the ADC reference voltage is
connected to the external VREF+ and VREF- pins that should be tied to ground. This makes it
possible to match the reference voltage and the measured signal range.
For example, if the measured signal varies between 0 V and 2.5 V, it is recommended to
choose VREF+ = 2.5 V, possibly using a reference voltage like LM235 (see LM235 datasheet
for more details). Figure 23 illustrates these conditions.
DocID15067 Rev 2
23/45
44
How to get the best ADC accuracy
Note:
AN2834
The voltage on VREF+ may range between 2.4 V and VDDA.
Figure 23. Selecting the reference voltage
V
V
VREF+ = 3.3 V
(4095h)
VREF+ = 2.5 V
(4095h)
Change VREF+
value from
3.3 V to 2.5 V
unused ADC
transition range
2.5 V
(3102h)
Digital
Output
4095V in
= -----------3.3
t
Digital
Output
4095= -----------V in
2.5
t
ai15600
Using a preamplifier
If the measured signal is too small (in comparison with the ADC range) then it is good to use
an external preamplifier. This method can be implemented with STM32x devices delivered
in all packages, and more specifically so in packages that do not have the VREF+ input.
For example, if the measured signal varies between 0 V to 1 V and VDDA is 3 V, the signal
can be amplified so that its peak-to-peak amplitude is similar to the VDDA value. The gain
will then be equal to 3. Figure 24 illustrates this example.
This amplifier can adapt the input signal range to the ADC range. It can also insert offsets
between the input signal and the ADC input. Care must be taken in the preamplifier design
not to generate additional errors (for example additional offset, amplifier gain stability,
linearity, frequency response, etc.).
Figure 24. Preamplification
V
V
VDDA = 3 V
(4095h)
Input signal max
value is 3 V
(4095h)
unused ADC
transition range
1V
(1365h)
G=3
t
t
Before amplification
Digital
24/45
Output
4095= -----------V in
3
After amplification
Digital
DocID15067 Rev 2
Output
=
4095 G
3
V in
ai15601
AN2834
3.2.6
How to get the best ADC accuracy
Analog source resistance calculation
Let us assume that the maximum error allowed is equal to 1/2 LSB. We will calculate the
maximum source resistance allowed.
Vc is the voltage across the internal CADC capacitor (refer to Figure 15).
1
Then we have:
Error = V AIN – V c = --- LSB
2
Figure 25. Worst case error: VAIN = VREF+
Vc
VAIN = VREF+
Error(VAIN = VREF+)
Error(VAIN < VREF+)
VAIN < VREF+
0V
ts
t
Error(VAIN = VREF+) > Error(VAIN < VREF+)
ai15602
Let ts be the sampling time.
ts = TS/fADC, where Ts is the sampling time evaluated by cycles
(1)
For a given ts, the error corresponding to VAIN = VREF+ is greater than the error
corresponding to VAIN < VREF+ because the CADC capacitor takes more time to charge from
0 V to VAIN when VAIN = VREF+ than it takes when VAIN < VREF+ (refer to Figure 25). So VAIN
= VREF+ is the worst case to be taken into account in the demonstration of the maximum
source resistance.


ts

– --------------------------------

R max C ADC
V REF+
 = 1
---  -----------------Error = V REF+ – V REF+  1 – e


N
2
2




, where:
Rmax = (RAIN + RADC)max
(2)

N is the ADC resolution (in our case N = 12)
ts
– -------------------------------
R max C ADC
1
This gives: e
= ---------- . Thus: R max
N+1
2

t
s
= ----------------------------------------N+1
C ADC  ln  2

(3)
By combining equations (1), (2) and (3), we obtain final formula:
Ts
R AINmax = ------------------------------------------------------------------- – R ADCmax
N+1

f ADC  C ADC  ln  2
DocID15067 Rev 2
25/45
44
How to get the best ADC accuracy
AN2834
Example for STM32F1 parameters:
fADC = 14 MHz, CADC = 8 pF, RADCmax = 1 k and for Ts = 7.5, the maximum source
resistance allowed for an error equal to 1/2 LSB is:

7.5
R AINmax = ------------------------------------------------------------------------------------- – 1k
6
– 12
12 + 1
14  10  8  10
 ln  2


That is:
RAINmax = 6.4 k
Note:
The use of a follower amplifier can reduce the resistance of the source effect because of its
high input impedance and its very low output impedance. It isolates RAIN from RADC.
However, the amplifier introduces an offset error that should be taken into account.
In case of longer sampling times and reduced number of ADC clocks, better results can be
obtained. It is possible to further increase the allowed external resistance by decreasing the
ADC clock frequency or selecting a lower resolution. Refer to the datasheet of your device
to obtain the exact values of the RC parameters.
26/45
DocID15067 Rev 2
AN2834
Source frequency condition vs. source and parasitic capacitors
The external capacitance will not allow the analog input voltage to be exactly the same as
VAIN if the capacitor is not fully charged by the analog source.
If the analog input signal changes, then the analog signal frequency (FAIN) should be such
that the time period of this analog signal is at least: 10 × RAIN × (CAIN + Cp).
TAIN = analog signal time period = 1/FAIN.
We have: T AIN  10  R AIN   C AIN + C P  
1
Therefore: F AIN  --------------------------------------------------------------------- 
10  R AIN   C AIN + C P 
For example:
For RAIN = 25 k, CAIN = 7 pF, Cp = 3 pF, this gives:

1
F AINmax = ---------------------------------------------------------------------------------- 
3
– 12
10  25  10   7 + 3   10
Thus, the maximum frequency of the source will be: F AINmax = 400 kHz .
So for the above defined source characteristics (capacitance and resistance), the frequency
of the source must not exceed 400 kHz, otherwise the ADC conversion result will be not
accurate.
Figure 26. Recommended values for RAIN and CAIN vs. source frequency FAIN
1000
CAIN 10 nF
CAIN 22 nF
100
CAIN 47 nF
Max. RAIN (k )
3.2.7
How to get the best ADC accuracy
10
1
0.1
0.01
0.1
1
FAIN(kHz)
DocID15067 Rev 2
10
ai15489
27/45
44
How to get the best ADC accuracy
3.2.8
AN2834
Temperature-effect compensation
One method would be to fully characterize the offset and gain drift and provide a lookup
table in memory to correct measurement according to temperature change. This calibration
involves additional cost and takes time.
The second method consists in recalibrating the ADC when the temperature change
reaches given values, by using the internal temperature sensor and the ADC watchdog.
3.2.9
Minimizing injection current
Check the application to verify whether any digital or analog input voltage can be less than
VSS or VSSA. If it is the case, a negative injection current will flow from the pins. The effect
on the accuracy will be greater if a digital input is close to the analog input being converted.
Negative current injection on any of the standard (non-robust) analog input pins should be
avoided as this would significantly reduce the accuracy of the conversion being performed
on another analog input.
It is recommended to connect a Schottky diode between VSSA and the I/O pin that can give
birth to the negative injection current.
The ADC accuracy will not be affected by positive injection currents within the limits
specified for IINJ(PIN) and ΣIINJ(PIN) (refer to the appropriate STM32x datasheet, I/O port
characteristics section).
3.2.10
Minimizing I/O pin crosstalk
The noise produced by crosstalk can be reduced by shielding the analog signal by placing
ground tracks across it. Figure 27 shows the recommended grounding between signals.
Figure 27. Crosstalk between I/O pins
Analog
in
STM32
Digital
I/O
ai15490
28/45
DocID15067 Rev 2
AN2834
3.2.11
How to get the best ADC accuracy
EMI-induced noise reduction
You can reduce EMI noise using proper shielding and layout techniques. The possible
sources of emission must be physically separated from the receptors. They can be
separated electrically by proper grounding and shielding.
Shielding technique
Placing ground tracks alongside sensitive analog signals provides shielding on the PCB.
The other side of the two-layer PCB should also have a ground plane. This prevents
interference and I/O crosstalk affecting the signal. See Figure 28.
Signals coming from distant locations (like sensors, etc.) should be connected to the PCB
using shielded cable. Care should be taken to minimize the length of the paths of these
types of signal on the PCB.
The shield should not be used to carry the ground reference from the sensor or analog
source to the microcontroller. A separate wire should be used as ground. The shield should
be grounded at only one place near the receiver such as the analog ground of the
microcontroller. Grounding the shield at both ends (source and receiver) might lead to the
creation of ground loops, with the result of current flowing through the shield. In this case,
the shield acts like an antenna and the purpose of the shielding is lost.
The shielding concept also applies to grounding the chassis of the application if it is metallic.
And it also helps remove EMI and EMC interference. In this case the mains earth ground is
used to shield the chassis. Similarly DC ground can be used for shielding in case of the
earth ground not being available.
Figure 28. Shielding technique
Not recommended
Shielded cable
Sensor
Recommended
Sensor
ADC
ADC
Ground
loop
Do not ground the shield at both ends
Ground the shield at the receiver end only
DocID15067 Rev 2
ai15491
29/45
44
How to get the best ADC accuracy
3.2.12
AN2834
PCB layout recommendations
Separating the analog and digital layouts
It is recommended to separate the analog and digital circuitry on the PCB (see Figure 29).
This also avoids tracks crossing each other. The tracks carrying digital signals may
introduce high-frequency noise in analog signals because of coupling.
The digital signals produce high-frequency noise because of fast switching.
Coupling of a capacitive nature is formed due to the metal connections (tracks) separated
by the dielectric provided by the PCB base (glass, ceramic or plastic).
It is recommended to use different planes for analog and digital grounds. If there is a lot of
analog circuitry then an analog ground plane is recommended. The analog ground must be
placed below the analog circuitry.
Figure 29. Separating the analog and digital layouts
Power
supply
Digital
circuitry
(Noise generator)
Analog
circuitry
STM32
(Affected by noise)
Analog
ground
plane
Digital
ground
plane
ai15492
Separating the analog- and digital-circuit power supplies
It is desirable to have separate analog and digital power supplies in cases where there is a
lot of analog and digital circuits external to the microcontroller (see Figure 30). Depending
on the STM32x package, different analog and digital power supply and ground pins are
available. The VDDA/VREF+ and VDD pins can be powered from separate power supplies.
If you use a switching-type power supply for the digital circuitry, you should use a separate
linear supply for the analog circuit. Also, if you expect a lot of noise on the DC power supply
due to I/O switching etc., it is preferable to use a separate supply for the analog circuit.
Figure 30. Separating the analog and digital supplies
Vout1
GND SMPS
Vout2
Linear
regulator
Star
Network
V SS V DD
V DDA
V REF+
Analog
Circuit
STM32
V REFV SSA
ai15493
30/45
DocID15067 Rev 2
AN2834
How to get the best ADC accuracy
It is also recommended to connect the analog and digital grounds in a star network. This
means that you must connect the analog and digital grounds at only one point. This
prevents the introduction of noise in the analog power supply circuit due to digital signal
switching. This also prevents current surges from affecting the analog circuit.
Using separate PCB layers for the supply and ground

Two-layer PCBs
For two-layer PCBs, it is recommended to provide a maximum ground plane area. The
power supply (VDD, VDDA) should run through thick tracks. The two layers can have
their ground shorted together via multiple connections in the overlap region if the two
layers feature the same ground signals. The unused PCB area can be used as the
ground plane.
The other convention is to connect the unused PCB area on one layer to the positive
supply (VDD) and the unused area on the other layer, to ground. The advantage is a
reduced inductance for power and ground signals. The maximum ground area provided
for ground on the PCB results in a good shielding effect and reduces the
electromagnetic induction susceptibility of the circuit.

Multilayer PCBs
Wherever possible, try to use multilayer PCBs and use separate layers on the PCB for
power and ground. The VDD and VSS pins of the various ICs can be directly connected
to the power planes, thus reducing the length of track needed to connect the supply
and ground. Long tracks have a high inductive effect. The analog ground can be
connected at one point to this ground plane. If so, it should be close to the power
supply.
A full ground plane provides good shielding and reduces the electromagnetic induction
susceptibility of the circuit.

Single-layer PCBs
Single-layer PCBs are used to save cost. They can be used only in simple applications
when the number of connections is very limited. It is recommended to fill the unused
area with ground. Jumpers can be used to connect different parts of the PCB.
3.2.13
Component placement and routing
Place the components and route the signal traces on the PCB so as to shield analog inputs.
Components like resistors and capacitors must be connected with very short leads. You can
use surface-mounted device (SMD) resistors and capacitors. You can place SMD capacitors
close to the microcontroller for decoupling purposes.
Use wide tracks for power, otherwise the series resistance of the tracks would cause a
voltage drop. Indeed, narrow power tracks have a non-negligible finite resistance, so that
high load currents through them would cause a voltage drop across them.
Quartz crystals must be surrounded by ground tracks/plane. The other side of the two-layer
PCB below the crystal should preferably be covered by the ground plane. Most crystals
have a metallic body that should be grounded. You should also place the crystal close to the
microcontroller. You can use a surface-mounted crystal.
DocID15067 Rev 2
31/45
44
How to get the best ADC accuracy
3.3
Software methods to improve precision

Averaging samples:
–




averaging decreases speed but can give improved accuracy
Digital filtering (50/60 Hz suppression from DC value)
–
set proper sampling frequency (the trigger from timer is useful in this case)
–
perform software post-processing on sampled data (e.g.: comb filter - for 50 Hz
noise and its harmonics suppression)
Fast Fourier Transform (FFT) for AC measurements:
–
to show harmonic parts in measured signal
–
slower, due to the use of more computation power
Calibration of ADC: offset, gain, bit weight calibration
–
decreases internal ADC errors
–
internal ADC structure must be known
Minimizing internal noise generated by CPU
–
3.3.1
AN2834
design firmware to use minimum disturbance from microcontroller during ADC
conversion - digital silence.
Averaging samples
The principle of this method is to increase ADC precision but decrease ADC conversion
speed (oversampling). If the measured analog signal produces unstable ADC values, then
the mean value of the given input signal can be obtained by averaging a set of values.
Variation can be caused by signal noise or noise generated by the microcontroller itself
(high speed digital signals capacitively coupled to the analog input signal).
Averaging is performed by choosing an appropriate number of samples to be averaged.
This number depends on the required precision, minimum conversion speed and the level of
other ADC errors (if another error has a greater influence on ADC precision, then increasing
the number of averaging values has no effect on total measurement precision).
The advantage of this method is improving precision without any hardware changes. The
disadvantage is lower conversion speed and therefore also lower frequency response (it is
equivalent to decreasing effective sampling frequency).
3.3.2
Digital signal filtering
This method uses digital signal processing techniques.
In principle, averaging is also a simple digital filter with a specific frequency response. But if
the noise frequency spectrum is known, a digital filter can be designed which minimizes
noise influence and maximizes ADC frequency response. For example, if the noise in the
measured signal is coming from the 50Hz power lines, then an appropriate digital filter
suppresses only the 50 Hz frequency and delivers data signal without this noise.
The disadvantage of this method is that it needs appropriate microcontroller processing
power and resources: CPU speed, data/program memory usage.
32/45
DocID15067 Rev 2
AN2834
3.3.3
How to get the best ADC accuracy
FFT for AC measurement
In some specific cases the application needs to know the amplitude of an AC signal with a
given frequency. In this case the effective value of an AC signal can also be obtained by
using a relatively slow sampling speed (in comparison to the measured signal frequency).
For example, when measuring an AC mains signal (which is near-to-sinusoidal and has
relatively low harmonics content), it is sufficient to choose a sampling frequency 32 times
greater than the mains frequency (50 Hz). In this case you can obtain harmonics of up to the
15th order. The amplitude of 15th harmonics in the main signal is very small (the next order
harmonics can be neglected). The calculated effective value of the mains signal is obtained
with high precision because the effective values of harmonics are added to the total AC
harmonic value as:
U ef =
2
2
2
U1 + U2 +  + Un
So if the 15th harmonics amplitude is only 1% (0.01) from the 1st harmonics (50 Hz) then its
contribution to the total effective value will be only 0.01% (because of theabove equation square addition: 0.012 = 0.0001).
The principle of this method is therefore to sample the AC signal with a known frequency
and then perform post-processing onthe FFT for each measured period. Because the
number of sampling points per measured signal period is small (32 points for example) then
the performance needed for FFT processing is not so high (only 32-point FFT for example).
Advantages: this method is good for AC measurement of signals with lower distortion. 
The disadvantage is the requirement for precise signal sampling:

The frequency of the measured signal must be known and the ADC sampling
frequency must be set exactly as a 2n multiplier of the measured frequency.

The input signal frequency is measured by another method.

The ADC sampling frequency is tuned by programming the prescaler and MCU master
clock selection (if sampling is performed with an inaccurate clock an interpolation can
be used to obtain samples at the required points).
DocID15067 Rev 2
33/45
44
How to get the best ADC accuracy
3.3.4
AN2834
ADC calibration
This method requires knowledge of internal ADC structure and how the ADC converter is
implemented inside the microcontroller. This knowledge is necessary in order to design a
physical/mathematical model of the ADC implementation.
A proper physical model (which is usually a schematic diagram) is used as the base for
describing it mathematically. From the mathematical model, each element in the model can
be obtained by a set of equations (for example, resistor/capacitor values which represent bit
weights). To solve these equations, it is necessary to perform a set of practical
measurements and obtain a set of solvable equations.
From the measured values and mathematical computation of the model, all known values of
model elements (resistors, voltages, capacitors,...) can be put into the schematic diagram.
So instead of the ADC schematic with the designed values, you obtain an ADC schematic
with the real values for a given microcontroller.
Computed model parameters are stored in the microcontroller memory after calibration and
used in post-processing to correct ADC values.
3.3.5
Minimizing internal CPU noise
When the CPU operates, it generates a lot of internal and external signal changes which are
transferred into the ADC peripheral through capacitive coupling. This disturbance influences
ADC precision (unpredictable noise due to different microcontroller operations).
To minimize influences of the CPU (and of other peripherals) on ADC, it is necessary to
minimize digital signal changes during sampling and conversion time (digital silence). This is
done using one of the following methods (applied during sampling and conversion time):
34/45

minimizing I/O pin changes

minimizing internal CPU changes (CPU stop, wait mode)

stopping clock for unnecessary peripherals (timers, communications...)
DocID15067 Rev 2
AN2834
3.4
How to get the best ADC accuracy
High impedance source measurement
This section describes the ADC measurement behavior of STM32x ADC when a signal
source with high internal impedance is used. It explains how to design an application to
reach the requested precision and provides workarounds.
3.4.1
ADC input stage problem
The ADC in STM32x devices is a switched-capacitor ADC type. Switched capacitors work
also as sampling capacitors (see Section 1.1 for a detailed explanation).
When a signal comes from a voltage source with high internal impedance (for instance,
150 k), an added error can be seen in measurement results. Error signals have also been
observed on the ADC input pin, as shown in Figure 33 (if the voltage source has zero
voltage: Uin = 0 V, Rin = 150 k, Cext = 0 pF):
Figure 31. Typical voltage source connection to ADC input
ADC input
R in
U in
STM32
C ext
ai17903
Figure 32. Noise observed on ADC input pin during ADC conversions
ADC input signal during conversion:
an ADC noise is injected to the input.
DocID15067 Rev 2
ai17904
35/45
44
How to get the best ADC accuracy
3.4.2
AN2834
Explanation of the behavior
The explanation of this added pin noise and added measurement error (in case a signal
source with high internal impedance is used) comes from the internal ADC structure: its
input sampling circuit.
Figure 33 shows a simplified schematic of the input stage (sample and hold circuit).
Figure 33. ADC simplified schematic of input stage - sample and hold circuit
S1
ADC input
Csh
ai17905
The spikes (noise) present on ADC input pin during conversions are related to the sampling
switch (S1). If the switch is closed, some charge (coming from the sample and hold
capacitor Csh or caused by another effect) is transferred to the input pin. Then this charge
starts discharging through the source impedance (Rin). The discharge process ends at the
end of the sampling time (tS) when the switch S1 is opened. The remaining undischarged
voltage remains on the capacitor Csh and ADC measures this voltage. If the sampling time
(tS) is too short, the remaining voltage does not drop under 0.5 lsb and ADC measurement
shows an added error. Figure 34 shows this process.
Added error
Uinput
Figure 34. ADC input pin noise spikes from internal charge during sampling process
t
Sampling time (tS)
=
Discharging time
Conversion time
(tC)
ai17906
Note that a non-zero external capacitance Cext (parasitic pin capacitance) also exists, so
during conversion time the pin capacitance is discharged through source impedance Rin.
36/45
DocID15067 Rev 2
AN2834
Minimizing added errors
Workaround for high impedance sources
To solve the added error problem, you can increase the sampling time (TS) configuring ADC
settings in MCU firmware, so that you discharge the Csh charge through the source
impedance Rin. The time constant (Rin * Csh) is the reference for choosing the sampling
time. To choose the sampling time cycles, use this formula (for max. error of 1/2 lsb, see
also Section 3.2.6):
T S  f ADC   R in  C sh   ln  2
N+1

 cycles 
The clock for ADC (fADC) is another important factor; slowing down the ADC clock increases
the sampling time.
Figure 35. Effect of sampling time extension
Added error < 0.5 lsb
Original added error
U input
3.4.3
How to get the best ADC accuracy
t
Original sampling
time (tS)
Extended sampling
time (tS)
Conversion time
(tC)
ai17907
If the maximum register value of the sampling time (TS) setting is reached and the problem
is still present, you need a more complex solution which is applicable also for
measurements of source with extra high internal impedance (see the section Workaround
for extra high impedance sources below).
Note that for this application you must take into account not only the internal sampling
capacitance, but also any external parasitic capacitance (in parallel to Cext), such as: pin
capacitance, PCB path capacitance, etc.
Do not add any external capacitor (Cext) to the input pin when applying this above
workaround. Its capacity will increase the timing constant (Rin * Csh || Cext) and the problem
will remain.
DocID15067 Rev 2
37/45
44
How to get the best ADC accuracy
AN2834
Workaround for extra high impedance sources
This workaround consists of 2 parts: a hardware change and a software change.
Hardware change:
The hardware change consists of adding a large external capacitor (Cext) to the input pin.
The capacity size connected to the input pin must reach the value that causes the
discharging of the internal sampling capacity Csh to the external capacitor Cext without
increasing the voltage on Cext to more than 0.5 lsb. 
Example:
If the internal capacitor (Csh = 16 pF) is charged to full scale (Umax, which corresponds to
4096 lsb), then the external capacitor Cext must be charged at maximum 0.5 lsb voltage
level (Ulsb) after discharging Csh to it. The capacity of Cext will then be:
U max
4096
C ext  C sh  -------------- = 16pF  -------------  131nF
U lsb
0.5
The closest larger standard value chosen here is: Cext = 150 nF.
If the internal sampling capacitor Csh is not charged to full voltage range (4096 level) before
sampling, you can calculate the Cext value replacing “4096” in the formula above.
Calculating with 4096 level gives precise measurement results also in the case of ADC input
channels switching (Csh was charged from different ADC input in the previous
measurement).
A side effect of this hardware workaround is the cyclical charging of Cext which must be
taken into account. Each ADC conversion transfers charge from Csh to Cext. One transfer
charges the Cext below 0.5 lsb, as described above, but more transfers can charge Cext to
larger values if it is not discharged between two conversions. Figure 36 shows an example
of this scenario where the ADC measurement is performed faster.
Added error < 0.5 lsb
Uinput
Added error > 0.5 lsb
Figure 36. Charging the external capacitor with too short time between conversions
Too short
tC
t
Recommended
tC
ai17908
38/45
DocID15067 Rev 2
AN2834
How to get the best ADC accuracy
Software change:
The side effect mentioned above can be solved by software. The objective is to create a
delay in order to let Cext discharge through Rin (not measure so often) giving enough
“discharge time” between ADC conversions. The “discharge time” (tC) is equal to the
transferred charge from Csh to Cext (charging) and from Cext to Rin (discharging). The
assumption is that Cext >> Csh.
Q ch arg ing = Q sh = C sh  U max
tC
Q disch arg ing
t
– ------------------U lsb
R in C ext
= -----------   e
dt
R in
0
where: 
Ulsb ....... 0.5 lsb voltage level
Umax ..... 4096 lsb voltage level (worst case)
Qcharging = Qdischarging
tC
C sh  U max
t
------------------U lsb
R in C ext
= -----------   e
dt
R in
0
Simplification of the above formula gives the final formula for the required waiting time
between conversions:
C sh U max
t C = –  R in  C ext   In 1 – ----------- -------------C ext U lsb
This final formula shows dependency between the external capacitor Cext and the required
waiting time between two conversions if the precision Ulsb is needed.
From the same formula you can see that the argument in logarithm must be positive and
therefore there is a condition for the minimal value of Cext:
C sh U max
1 – ----------- --------------  0
C ext U lsb
C sh U max
1  ----------- -------------C ext U lsb
U max
C ext  C sh  -------------U lsb
Choosing a larger Cext decreases more the time between conversions (tC).
U
U lsb
max
- ) enables sampling more often.
An extra large Cext (Cext>> Csh  -------------
However, increasing Cext limits the frequency bandwidth of measurement signal (increasing
the “external” timing constant Rin . Cext).
DocID15067 Rev 2
39/45
44
How to get the best ADC accuracy
AN2834
The formulas below show how to choose the optimal Cext value: signal bandwidth in
correlation with sample time. Signal bandwidth is characterized by an “external” timing
constant, so optimal solution is to charge Cext during tC:
 R in  C ext  = t C
C sh U max
 R in  C ext  = –  R in  C ext   In 1 – ----------- -------------C ext U lsb
C sh U max
– 1 = In 1 – ----------- -------------C ext U lsb
e
–1
C sh U max
= 1 – ----------- -------------C ext U lsb
After simplification we obtain the final formula for optimal Cext:
C ext
U max
C sh -------------U lsb
U max
-  1 58  C sh ------------= ----------------------–1
U lsb
1–e
and the corresponding waiting time between conversions:
1 -  R  C 
t C  –  R in  C ext   In 1 – -----------in
ext
1 58
Practically the firmware must not program the ADC in continuous mode but only in single
mode and must ensure that there will be a time gap between conversions with duration
equal to tC. This adding of waiting time is the software change which must be applied
together with the hardware change (adding an external capacitor Cext).
Without implementation of tC waiting time in software (for instance, running a conversion just
after the first one) the external capacitor Cext will be cyclically charged from the Csh
capacitor. After a lot of cycles the voltage on Cext will reach a quite high error value (as
previously shown in Figure 36).
A practical example of implementation for STM32L1xx ADC is shown below:
Csh = 16 pF
..... ADC property
Rin = 150 k 
..... signal source property
Umax = 4096 lsb
..... ADC property
Ulsb = 0.5 lsb
..... required precision
U max
4096
C ext = 1 58  C sh -------------- = 1 58  16pF  -------------  207nF  220nF
U lsb
0.5
C sh U max
16pF 4096
t C = –  R in  C sh   In 1 – ----------- -------------- = –  150k  220nF   In 1 – ----------------- -------------  29891s  30ms
C ext U lsb
220nF 0.5
40/45
DocID15067 Rev 2
AN2834
3.4.4
How to get the best ADC accuracy
Source of described problem - ADC design
The following sections list some possible causes for the charging of the internal sampling
capacitor Csh. This is not an exhaustive list; only the main possible sources of the ADC
design are listed.
Parasitic switch capacitance effect
The sampling switch inside ADC sampling circuit (see Figure 33) is not ideal. In reality the
sample and hold switch (S1) is designed as 2 transistors (PMOS and NMOS, see
Figure 37):
Figure 37. Implementation of sampling switch
S1
C sh
U in
H = switch on
L = switch off
ai17909
The switch is controlled by the gate voltages of transistors (inverted signal on PMOS
transistor). This design is a standard bidirectional switch (for rail to rail range of input Uin
voltages). Both transistors have parasitic capacitances between gate and source.
If those capacitances are charged (close to the switch), then their charge can be transferred
to the sampling capacitor (see Figure 38).
Figure 38. Parasitic capacitances of sampling switch
i
i
S1
C sh
U in
C ext
L --> H switching on
ai17910
This charging and discharging currents (PMOS and NMOS asymmetric capacitances) can
cause charge transfer to sampling capacitor Csh.
DocID15067 Rev 2
41/45
44
How to get the best ADC accuracy
AN2834
Internal charging of sampling capacitor
It is possible that after the conversion process (successive approximation process in SAR
type of ADC) the sample and hold capacitor Csh is charged to some voltage. The reason
can be:

some leakage current to Csh (parasitic current inside ADC structure, see Figure 39)

residual charge transfer from the switches when ADC structure is switched back to
default state before next conversion

other reasons (related to internal ADC parasitic structures)
Figure 39. Parasitic current example inside ADC structure
Rparasitic
ADC input
S1
Csh
ai17911
42/45
DocID15067 Rev 2
AN2834
4
Conclusion
Conclusion
This application notes describes the main ADC errors and then methods and application
design rules to minimize STM32x ADC errors and obtain the best ADC accuracy.
The choice of method depends on the application requirements and is always a
compromise between speed, precision, enough computation power and design topology.
The published methods lead to a precision improvement and are optimized for the design of
an ADC converter using the SAR (successive approximation register) principle.
DocID15067 Rev 2
43/45
44
Revision history
5
AN2834
Revision history
Table 2. Document revision history
Date
Revision
14-Nov-2008
1
Initial release.
2
Extended to STM32Fx Series and STM32L1 Series devices.
Added Section 1.1: SAR ADC internal structure.
Added Section 3.4: High impedance source measurement.
Added Section 3.3: Software methods to improve precision.
Text improvements and additions.
Changed the Disclaimer on the final page.
16-Sep-2013
44/45
Changes
DocID15067 Rev 2
AN2834

Please Read Carefully:
Information in this document is provided solely in connection with ST products. STMicroelectronics NV and its subsidiaries (“ST”) reserve the
right to make changes, corrections, modifications or improvements, to this document, and the products and services described herein at any
time, without notice.
All ST products are sold pursuant to ST’s terms and conditions of sale.
Purchasers are solely responsible for the choice, selection and use of the ST products and services described herein, and ST assumes no
liability whatsoever relating to the choice, selection or use of the ST products and services described herein.
No license, express or implied, by estoppel or otherwise, to any intellectual property rights is granted under this document. If any part of this
document refers to any third party products or services it shall not be deemed a license grant by ST for the use of such third party products
or services, or any intellectual property contained therein or considered as a warranty covering the use in any manner whatsoever of such
third party products or services or any intellectual property contained therein.
UNLESS OTHERWISE SET FORTH IN ST’S TERMS AND CONDITIONS OF SALE ST DISCLAIMS ANY EXPRESS OR IMPLIED
WARRANTY WITH RESPECT TO THE USE AND/OR SALE OF ST PRODUCTS INCLUDING WITHOUT LIMITATION IMPLIED
WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE (AND THEIR EQUIVALENTS UNDER THE LAWS
OF ANY JURISDICTION), OR INFRINGEMENT OF ANY PATENT, COPYRIGHT OR OTHER INTELLECTUAL PROPERTY RIGHT.
ST PRODUCTS ARE NOT DESIGNED OR AUTHORIZED FOR USE IN: (A) SAFETY CRITICAL APPLICATIONS SUCH AS LIFE
SUPPORTING, ACTIVE IMPLANTED DEVICES OR SYSTEMS WITH PRODUCT FUNCTIONAL SAFETY REQUIREMENTS; (B)
AERONAUTIC APPLICATIONS; (C) AUTOMOTIVE APPLICATIONS OR ENVIRONMENTS, AND/OR (D) AEROSPACE APPLICATIONS
OR ENVIRONMENTS. WHERE ST PRODUCTS ARE NOT DESIGNED FOR SUCH USE, THE PURCHASER SHALL USE PRODUCTS AT
PURCHASER’S SOLE RISK, EVEN IF ST HAS BEEN INFORMED IN WRITING OF SUCH USAGE, UNLESS A PRODUCT IS
EXPRESSLY DESIGNATED BY ST AS BEING INTENDED FOR “AUTOMOTIVE, AUTOMOTIVE SAFETY OR MEDICAL” INDUSTRY
DOMAINS ACCORDING TO ST PRODUCT DESIGN SPECIFICATIONS. PRODUCTS FORMALLY ESCC, QML OR JAN QUALIFIED ARE
DEEMED SUITABLE FOR USE IN AEROSPACE BY THE CORRESPONDING GOVERNMENTAL AGENCY.
Resale of ST products with provisions different from the statements and/or technical features set forth in this document shall immediately void
any warranty granted by ST for the ST product or service described herein and shall not create or extend in any manner whatsoever, any
liability of ST.
ST and the ST logo are trademarks or registered trademarks of ST in various countries.
Information in this document supersedes and replaces all information previously supplied.
The ST logo is a registered trademark of STMicroelectronics. All other names are the property of their respective owners.
© 2013 STMicroelectronics - All rights reserved
STMicroelectronics group of companies
Australia - Belgium - Brazil - Canada - China - Czech Republic - Finland - France - Germany - Hong Kong - India - Israel - Italy - Japan Malaysia - Malta - Morocco - Philippines - Singapore - Spain - Sweden - Switzerland - United Kingdom - United States of America
www.st.com
DocID15067 Rev 2
45/45
45