 Hello, this video is a third part of a series on practical RF impedance matching and filtering of the STN32WL. In this part, we focus on the low-pass filter, RF switch, and the antenna matching network. The first three blocks of transmitted design steps were discussed in the part 2. In this part, we continue with the remaining blocks. We start with the low-pass filter and RF switch. The main purpose of this block is to suppress higher harmonics. The first step is initial design of the filter. The low-pass filter can be calculated. Design parameters can be corner frequency about 1400 MHz, third-order Chebyshev approximation with pi topology, and matching 50 ohms in the band of interest. The filter design can be also reused from existing designs if the designs are similar. The calculation is not necessary. The next step is fine-tuning of the filter. This step includes impact of all parameters of a particular PCB which were not included in the previous steps. Fine-tuning may be done by simulation or measurement. In this video, calculation and measurement are shown. For 868 MHz, we can calculate the following initial values of the filter. C13 is the DC block. The calculated values are ideal, they can be fine-tuned in the real design. Typical ranges of the real components are as follows. The values depend on neighboring blocks, PCB layout, etc. The DC block capacitor has usually fixed value. Here we can see simulated transmission of the low-pass filter, only with idle components. In the detailed view, we can see that the transmission is the best in the working band and the corner frequency is about 1400 MHz. And simulation of S11, which is around 50 ohms in the working band. For practical reasons, it's usually easier to measure the low-pass filter together with the R-switch. Otherwise, it must be removed from the board. In our case, the measurement points of the low-pass filter are before the filter and on the SMA connector. The R-switch and the RF path have impact to the measurement. Anyway, this influence can help to better fine-tune the low-pass filter. Here is the low-pass filter measurement, shown in the schematic. We can see the low-pass filter circuit, the RF switch and the antenna matching network. It is not used. The serial component is zero ohm resistor. The parallel capacitors are not used. One measurement point is at this point through the pigtail. The second measurement point is on the SMA connector. The measured RF path looks like this. The values of the low-pass filter were fine-tuned a bit. C8 was decreased a bit. This is mainly due to the parasitic capacitance of the RF switch. C14 was decreased as well. This is mainly due to impact of the pigtail. Impact of the PCB layout is included as well. Here we can see how the low-pass filter measurement looks in reality. One port of a PNA is connected through the pigtail. The second port is connected to the SMA connector. To do the measurement, we need to set the RF switch correctly. In our case, set the connection between RF3 and RFIN. Because there is no dedicated AT command to control only the RF switch, we can use a command for unmodulated carrier. It switches the RF switch but also enables signal from the RF or LP PIN. Because the signal is not needed, it can be set to low power. For example, 9 dBm, which is the lowest possible value, which can be set. We can use the following commands. ATT conf to configure the frequency and the output power, and the ATT tone, which enables the carrier and sets the RF switch. Here are the measurement results of the low-pass filter with fine-tuned values and the RF switch. It is quite good matched at both ends. In the transmission chart, we can see slightly bigger insertion loss in the working band, which is caused by the RF switch and the RF path. The markers show attenuation of 868 MHz harmonics. For example, attenuation of the 5th harmonic is about minus 41 dB. In the next design step, we connect the low-pass filter with the previous design steps and fine-tune them, if necessary. In the previous step, we designed the low-pass filter, which was isolated from the previous blocks. In this step, we connect it to the blocks, which were designed before. We will use an unmodulated carrier signal from the RF or LP output. We will measure the transmitted signal on the SMA connector. The connection usually slightly changes the parameters we measured before, so the low-pass or not filters might be fine-tuned if needed. There are several reasons, for example. All separated blocks were measured with pigtail, which is now missing. It has some parasitic impedance. The pigtail and how it is connected to the ground have also impact to the measured results. Here we can see the path of the transmitter we are measuring now. The antenna matching network is bypassed by zero resistor. We use spectrum analyzer, which is connected to the SMA connector through the RF cable. Here is the measured spectrum. The higher harmonics have been attenuated. The fundamental harmonic is still at a good level, about 13.49 dBm. The current is about 28 mA, which is more than our target of about 25 mA. We can fine-tune the current. C14 is a shared capacitor between the notch filter and the low-pass filter. Both blocks were first fine-tuned separately, now are matched together. When measured separately, the pigtail was used. Unfortunately, it also adds some parasitic impedance, which impacts the measurement. The pigtail is missing now. We must fine-tune this change. In our case, I've increased the value of the capacitor C14 from 1.5 picofarads to 3 picofarads. After this change, the current has decreased to 24.16 mA. The spectrum has also changed a bit. The fundamental harmonic has decreased to 13.18 dBm, which is still acceptable. The high harmonics have also changed a bit. All harmonics are below the Etsy regulatory limit, but some harmonics seem to be stronger, closer to the limit. This is usually caused by the PCB layout. On the PCB, we can see the RF switch, transmitter part, and the receiver part. Here is a position on the low-pass filter. The strong signal from the RF output may couple to other layout structures on the PCB before the low-pass filter. The coupling is usually better for higher frequencies and strongly depends on the particular PCB layout. This coupled signal flows outside the low-pass filter. This is the reason why some harmonics appear after the low-pass filter. Depending on the coupling path, they can be attenuated in the receiver part or in the antenna matching network. It's usually also a low-pass filter. In this example, we will show additional filtering by it. Let's focus to the antenna matching network. In our case, we expect 50 ohms on the SMA connector, so from the matching point of view, it matches impedance from 50 ohms to 50 ohms. More important is filtering feature of this structure. We use it as an additional low-pass filter. The filter is symmetrical, matched at both ends in the 868 MHz band. The filter calculation was based on the value of the inductor that was available at the moment. The calculated component values are ideal. The filter is not very sharp, but in our case it helps. The performance can be improved more if needed. For our case to show the principle, it is good. Up to now, we made measurements with bypass antenna matching network. Serial zero ohm resistor was used. Position of parallel capacitors were empty. Now, the calculated values of the antenna matching network were fitted. Here is the final measurement with the antenna matching network. The higher harmonics were attenuated, and we have better margin to the certification limit. The fundamental harmonic has decreased a bit. The final value is about 13.09 dBm. The current has decreased a bit as well. If needed, the antenna matching network or the transmitter part can be fine-tuned again to get a bit better results. For our example, these results are okay. Here is the picture of the measured board, as we use it in this example. Typical insertion loss of the transmitter path with the RF switch is about 1 dB. In our case, we set 14 dBm in the firmware and measured the following values on the individual blocks. The final output power measured on the SMA connector is 13.09 dBm. Our insertion loss is about 0.8 dBm. Here is the table of components we used in this example. Now, we are almost at the end. The last important step is verification that the design transmitter path works as expected in all working conditions, which can be used. If something is not good, then it must be improved. To find a good design is an interactive process. But now, with knowledge of each building block, it should be easier. You should know what to focus on. Here are some points, which is good to check. The first one is the RF spectrum. In the example, we used one frequency and one power level. It is good to check what is the spectrum at different frequencies and power levels, which can be used. The unmodulated carrier can be used for these tests. The next step may be to test radio communication if the FSK or lower-modulated signal works as expected. Tolerance of components may have also impact. It is good to check parameters on different samples and the impact of temperature. The target environment can affect antenna parameters, such as the radiation pattern, but also the radiated power and power consumption. The details of the last two topics are described in the following two slides. To have good and stable parameters of the transmitter path, components with good tolerance are recommended. Here is the simulation of the notch filter. The red curve is based on nominal values. The blue curves are based on values with a tolerance of plus minus five percent. The difference may be quite big. Attenuation of the second harmonic may differ several dB. If the operating temperature range is wide, it is a good idea to check the transmitter performance depending on the temperature. Some variations always occur. It is necessary to have a good margin from the certification limits. Load of the transmitter path has an impact to the optimal loading impedance. Different loading impedance may change the output power and RF current consumption. In the Smith chart, we can see a simple example of the matching network from 50 ohms to the optimal loading impedance. If we change the load around the circle of constant SWR, which is 1.5, we can see how the load impedance into which the power amplifier operates is changing. The load is typically an antenna. The impedance of the antenna depends on the surrounding environment. To have the best parameters, fine-tuning in the target working environment is needed. Brief conclusion of the transmitter RF path matching. PCB layout has impact to the overall results. Fine-tuning of the transmitter RF path is an interactive process. It's better to divide it into smaller parts. Typical insertion loss of the transmitter RF path is about 1 dB. Verification is another important step after the matching. Thank you for your attention.