How conducted EMI influences touch sensors
As the touch sensing devices are getting thinner (fewer or thinner substrates) and capacitive touch sensing technologies are getting implemented in a growing number of new environments, new challenges arise for touch sensor designers. One major challenge is to provide proper EMI (Electromagnetic interference) immunity levels to touch sensors. Providing proper EMI immunity can be a difficult process for engineers since EMI (noise) depends on a lot of factors, such as the application, the environment of installation, the product specifications and more. To evaluate EMI, SNR is used as an acceptable performance metric for capacitive touchscreen systems.
What is SNR?
As the name suggests, SNR is the ‘Signal to Noise Ratio’ and is defined as the ratio of the power of a signal (meaningful information) to the power of background noise (unwanted signal). In principle, although this value depends on a lot of factors a touch sensor designer generally wants to achieve a high SNR.
In the pursuit of a high Signal-to-noise ratio, designers have to either increase the signal power or reduce the noise or do both at the same time. In real life, though, many times there is a tradeoff; by trying to increase the signal power, an increase in the noise level is happening as well. To build robust touch sensors, engineers need to take into account the environment and the possible ways that the end user might interact with the product.
In this post we will discuss how simulation can be used to evaluate the optimal touch sensor design for achieving the desired noise immunity levels. For this purpose, it is important for the engineer to have an understanding of what noise is, where it comes from and how it affects capacitive touch sensors.
What is EMI?
EMI is an unwanted disturbance in an electrical signal and it is produced by several different effects. In a basic arrangement, a noise source is producing the noise and that noise is coupled to a device through one or more of the possible different coupling modes.
There are four basic noise mechanisms: conductive, capacitive, inductive and radiative. Those mechanisms describe the different ways by which a device can be coupled and influenced by a noise source.
Naturally, capacitive touch sensors are influenced by all of them in different ways.
What are the most common types of noise in touch sensors?
Capacitive touch sensors are generally vulnerable to the so called radio frequency interference. By radio frequency we mean sources oscillating from a few kHz to some hundreds GHz. There are two mechanisms for RF interference: conducted and radiated. The distinction is based on the frequency of the noise being tested. At lower frequencies (10s of kHz to 10s of MHz), noise is interpreted as conducted and at higher frequencies it is interpreted as radiated.
Because most capacitive-sensing circuits are influenced by noise sources that operate at less than 80 MHz, the conducted EMI is the most relevant noise to capacitive touch sensors (Texas Instruments).
What is conducted EMI?
Conducted EMI is classified into two types according to the conduction mode:
The first type is the differential mode noise. In this mode the noise usually comes from the power supply source, but can also occur within the load or the wires. It is called “differential mode” because the noise is conducted on the signal and GND lines in opposite directions. The most common sources of differential mode noise are switching actions from power electronic devices on the power supply side.
The second type is common mode noise which is conducted on all lines in the same direction. In common mode noise the current that has leaked via a stray capacitance passes through ground and returns to the power supply line. In a touch sensing system, the main source of common mode noise is introduced by the finger of the user. The user is coupled to the sensor by a stray capacitance due to the different ground references of the sensor and the human body. For example, in a sensor that is influenced by common-mode noise, the entire system “moves” relative to earth ground as it follows the common-mode noise. This is usually not a problematic state, until a finger “opens” a conductive path to that same earth ground. That touch creates a path for charge flow, which is equivalent to a noise signal injected exactly at the finger location (Cypress).
Common ways to improve noise immunity
There are many ways with which noise immunity can be pursued in a capacitive touch sensor system. Some have to do with the way the microcontroller deals with the signal itself, some have to do with proper touch sensor layout design and some are a combination of both. STMicroelectronics suggests the following methods:
- Active shield
- Spread spectrum
- Detection threshold
- SW filter
- Frequency hopping
- Channel blocking
- Impedance path to earth
In the end, no matter how potent the microcontroller is at dealing with unwanted noise signals, it will always be limited by the layout design of the touch sensor. A bad sensor design will reduce the SNR level to the point that the wanted and unwanted signal will no longer be able to be interpreted properly by the controller. This will lead to no touch response or to unwanted touch events. This means that good layout design is the cornerstone for noise immunity.
Improving noise immunity through standard testing
The IEC61000-4-6 standard specifies the test procedure to evaluate the noise immunity of a touch sensor when influenced by conducted noise sources. The test consists of a noise generator which is used to inject modulated noise signals into the touch sensor power supply lines. By varying the frequency and the level of the injected signal, we are able to characterize situations where the touch device becomes unreliable.
According to the standard, the minimum frequency step recommended is 1% of the preceding frequency value in a range from 150 kHz to 80 MHz.
In real life for the majority of the touch sensing systems those steps are simply not enough to properly identify the frequency with the highest impact. That is because most of the time a noise source will have a high influence on a frequency that is between those steps. It is then important to set smaller steps to identify noise in critical frequencies. The challenge is that most noise signal generators are not able to do so.
The solution that we will present in the use case below is to perform the standard test using the simulation software SENSE and cover the whole spectrum of frequencies with steps as small as desired.
USE CASE: Evaluate noise immunity levels of capacitive sensors using SENSE
We are going to evaluate the noise immunity levels of a simple capacitive touch sensor using the simulation software SENSE and IEC 61000-4-6 standard for the common mode noise case.
The steps we are going to follow are:
- Design a few simple capacitive touch sensor layouts using the design capabilities of SENSE.
- Perform the Parasitic Extraction using SENSE.
- Automatically generate the equivalent circuit with SENSE.
- Perform a SPICE analysis using IEC 61000-4-6 standard specifications.
Step 1: How to design a touch sensor
First, we need to create a capacitive touch sensor layout. In this use case will use a PCB based touch sensor and explore 3 different designs, using shielding as the key differentiation between the designs. The options are:
- No Shield: 1 design without a shield, that promises high touch sensitivity but poor noise immunity performance.
- Strong Shield: 1 design with very strong shield, that decreases sensitivity but also improves noise immunity.
- Weak Shield: 1 intermediate design that is a trade-off between sensitivity and noise immunity.
Step 2: How to perform parasitic extraction
The purpose of this step is to extract all the capacitance values that are necessary in order to evaluate the performance of the touch sensing system. For this reason we need the capacitance of the button towards the local earth, the capacitance of the button towards the far earth and the capacitance introduced to the system from the finger touching. In order to understand which sensor is more sensitive to touch, we also need the change of self capacitance of the button before and after the touch. The larger the change, the more sensitive the sensor.
In our case we used SENSE to estimate the self capacitance and parasitic capacitance values of the three test layouts.
|In pF||No Shield||Strong Shield||Weak Shield|
Cself No Touch
Table 1. Self capacitance values before and after the touch
|Touch / In pF||No Shield||Strong Shield||Weak Shield|
C towards local earth
C towards far earth
C towards pointer
Table 2. Parasitics of the system
Step 3: How to create the equivalent schematic
Equivalent schematic refers to a realistic representation of the components of an electrical circuit. In this step we need to extract the equivalent schematic of our touch sensor – including the touch chip (controller) and the effect of the human body that interacts with the circuit.
The controller and the human body model are complex if they are to be represented in full detail and far from the scope of this investigation. Instead, we are going to use simplified models in order to (i) depict the Charge Transfer modulation method of a generic touch button controller that converts capacitance to counts and (ii) represent the human body.
In the generic Charge Transfer modulation method, the controller charges the sensor. When the sensor is fully charged, the charge is transferred to an internal capacitor. Each charge and transfer sequence of the capacitor is measured as a “count”. The process is finished once a voltage threshold is reached. Then the total raw counts are measured and the controller identifies if we have a touch or no touch event. If you want to get a better understanding in Charge Transfer modulation method, you can read our post ‘How to configure Charge Phase in capacitive sensing’ where we describe the Charge Transfer modulation method in more detail.
For the purpose of this use case the controller’s functionality can be represented as a voltage source that drives the touch sensor at Vdd. The generic model that represents the human model consists of a resistance in series with a capacitance. Depending on many factors (i.e. age, gender, skin color etc.) those values can vary. In general, 1.5kΩ and 100pF can be considered a good approximation.
In Figure 5, we can see the capacitive coupling of the sensor towards the local earth, the earth far and the pointer.
Figure 5. Equivalent Schematic
Using SENSE the user has the ability to export a ready-to-use Netlist file with the equivalent circuit of a sensor by automatically analysing and combining the R and C values of the whole sensor.
Step 4: How to replicate the IEC 61000-4-6 standard using simulation
Next, we are going to run a SPICE Analysis, using the netlist file as derived from SENSE to replicate the IEC 61000-4-6 standard by introducing a step of 50Hz to the frequency samples. This step is small enough (STMicroelectronics suggests 60Hz) in order to include any possible influences of the noise source that might otherwise be missed due to large sampling steps in the laboratory test. Table 1 describes the input values used in SPICE Analysis:
3V rms 80% amplitude modulated with a 1kHz sine wave
Table 3. SPICE Analysis input values
Running a SPICE Analysis for the 3 different layouts, we get the following results:
There are three readings of the results we’ve extracted in this process:
- The first reading is understanding which design is the most sensitive to touch. As expected, when we do not apply a shielding element around a touch button, the sensitivity of the sensor is improved. We can see that in Table 1 by reading the change of the self capacitance or by reading the change of counts without touch, shown in Table 3, versus those shown in Figure 6.
- The second reading is understanding the frequency spectrum in which those designs are more prone to noise. Reading Figure 6 we can see that in the vicinity of 304KHz we have highs and lows in the counts measurement, which means that noise injected with this frequency will have higher impact in the performance of the sensor.
- Finally, a third reading is identifying which one of those sensors is more sensitive to noise. In Figure 6 we can see that not applying a shield creates peaks at specific frequencies that are larger than when applying a weak or strong shield. Any large deviation from an expected mean value can be rather problematic for a sensor. The reason for that is that the IC measures the counts and uses this value to identify a touch event or a no touch condition. If this measurement fluctuates a lot then a potential touch event might be rejected as a no touch condition.
As expected, the strong shield is more effective in terms of noise immunity. That is due to the position of the shield in close proximity enables a low impedance path for the noise to propagate to earth, in comparison to the weak shield or no shield design.
|Counts||No Shield||Strong Shield||Weak Shield|
Table 4. Counts without touch (noise injected)
|Touch||No Shield||Strong Shield||Weak Shield|
Table 5. Counts with touch (300…310kHz amplitude modulated noise injected by finger)
The results from this case study highlight the importance of understanding how the touch sensor works in order to make the right design choices, bearing in mind the tradeoffs of each case. Adding a strong shield to the system makes it almost immune to conducted noise influences, while this same immunity reduces the performance of the touch sensor. SENSE provides a seamless process for engineers to run as many design iterations they need and get system-level answers for their capacitive touch sensors.