Antenna transmission line length and impedance matching

Thread Starter

gggccc

Joined Aug 15, 2022
11
Hi,

I am not very experienced with RF and I am trying to design an RF module. I've done antenna design before so I know a little about impedance matching. However, there is one thing I am not sure about. I'm waiting for your help.

In microstrip antennas, we calculate the length of the feedline to match the feedline. This length usually has a length of λ/4. I can't understand how this happens on a PCB. In the RF Module, there is a path from the processor to the antenna, and there are various components on this path. There is also a matching circuit near the antenna. What I'm wondering is:
1) Is the matching circuit sufficient to match the entire transmission line from the processor to the antenna, and if there is a matching circuit, is there a need to calculate the length of the transmission line?

2) Should we calculate the matching circuit according to the antenna, or according to the transmission line before the matching circuit?

Thanks in advance.
 

Ya’akov

Joined Jan 27, 2019
9,070
Welcome to AAC, nice to have you join us.

In microstrip antennas, we calculate the length of the feedline to match the feedline.
Is the second “feedline” supposed to be impedance? Antenna? Something else?

You would probably understand this a lot better if you step back and learn about the transmitter—feedline—antenna system in the context of why impedance matching is required. Fundamentally, impedance matching is to reduce reflected signals caused by a mismatch to a minimum.

Each element in the system needs to have a connection matching its characteristic impedance. So, for example, if there is a transmither whose output impedance is 50Ω and it needs to couple to a 75Ω feedline, something has to be done to ensire that on tramistter side it looks like the desired 50Ω and on the other 75Ω.

If not, some portion of the signal from the output stage of the transmitter will reflect from that connection potenitally causing enough heat to destroy the output stages amplifier. The same is true at the feedline-to-antenna connection. Most feedline, even on PCBS, is by convention 50Ω. This isn‘t always the case but it mostly so. Things like ladder line have a 300Ω or 450Ω impedance, and the coaxial cable used in cable TV distribution like RG8U is 75Ω.

Each of these were chosen for practical reasons, compromising on cost vs. performance. 50Ω is a compromise between the 77Ω of free space and the expensive to produce 30Ω cable that would have the best power handling capabilities. 300Ω was very convenient because it matched the 4:1 baluns used to match 300Ω ladder line to the characteristic impedance of a resonant ¼λ dipole mounted ~¼λ above the ground, which is 75Ω. (The relationship of this number to the cable number is coincidence).

So you have to provide a match anywhere there is a change in characteristic impedance. This is any any such interface, transmitter—feedline, feedline—antenna, or anything else that might be in the signal path. The film from AT&T’s tech channel below is very helpful in developing an intuition for this. Definitely recommended as foundation.

Another very helpful film produced by the U.S. Army SIgnal Corps for technicians at the start of WWII discusses in general terms the behavior of radio waves the transmitter—feedline—antenna system. I found their explanation quite lucid and it provides additional tools for intuition.

I fear that despite the large number of words I have written, I’ve not directly answered your questions.

1) Is the matching circuit sufficient to match the entire transmission line from the processor to the antenna, and if there is a matching circuit, is there a need to calculate the length of the transmission line?
The answer to this depends on whether there is a match between the transmission line and the transmitter(/receiver) of the module. In practice most transmission lines on PCBs will be laid out to achieve a 50Ω impedance and most radios in ICs are designed for 50Ω, by convention.

The length of a transmission line does not change its characteristic impedance. Longer lines have more loss, but that’s all. However, in practice the antenna is likely to need a matching network. This is because the characteristic impedance of the antenna will be different depending on frequency. You can match an antenna by adjusting its length and/or distance from the ground, but it is often not practical to do that.

For example, vertical antenna inside a radome, nominally ¼λ, of the type frequently used for small devices cannot easily be physically trimmed. And in any case, compromises in size mean the antenna may rely on the mat hing network to be shorder to accommodate the physical constraints of a practical design.

2) Should we calculate the matching circuit according to the antenna, or according to the transmission line before the matching circuit?
I think it should be clear by now that the match is the interface between the two and so is calculated according to both. This detailed tutorial from Maxim om PCB transmission lines is a very good resource. It covers much more than you probably need to know. This application note from NXP on antenna matching is also full of good stuff.

I believe it will make understanding of the moving parts of this area much easier if you stand back and spend a bit of time on the fundamentals in order to understand why the different bits are needed. It will make things less mysterious. I hope this was helpful.
 

Thread Starter

gggccc

Joined Aug 15, 2022
11
Thank you so much for giving such a detailed answer to my question and for your support.

Based on the designs I reviewed; If there is a balun and an LC filter on the path from the chip to the antenna, that's enough to match the transmission line, right? An antenna matching circuit is used for the antenna as well to be compatible with the balun and LC filter, right? Thus, the system can operate at minimum loss. I hope I interpreted it correctly. Ya’akov
 

Ya’akov

Joined Jan 27, 2019
9,070
Thank you so much for giving such a detailed answer to my question and for your support.

Based on the designs I reviewed; If there is a balun and an LC filter on the path from the chip to the antenna, that's enough to match the transmission line, right? An antenna matching circuit is used for the antenna as well to be compatible with the balun and LC filter, right? Thus, the system can operate at minimum loss. I hope I interpreted it correctly. Ya’akov
Yes, that should be a good confguration.
 

Janis59

Joined Aug 21, 2017
1,834
RE:""1) Is the matching circuit sufficient to match the entire transmission line from the processor to the antenna, and if there is a matching circuit, is there a need to calculate the length of the transmission line?""

The any part from output cascade to cabling and antenna had adjusted to the same 50 Ohm impedance. If some of it havent, then apply individual filter - output cascade own, antenna own. Impedance is transmission line independent.

RE:"" 2) Should we calculate the matching circuit according to the antenna, or according to the transmission line before the matching circuit?""

Yes, because all cables have 50 Ohms and 75 Ohm times have restricted to sattelite technique and thats all. Therefore any antenna filter is adjusted to work with 50 Ohm cabling. Any transmitter is adjusted to work with 50 Ohm load.

Add on - in the extreme rare case if power transmission load is shorter than wavelength divide to ten or better 40, then impedances may not be adjusted and nothing bad happens.

Add on two - the symmetrical antennas have inherrent need to balun, simmetrizing the genuine unsimmetric cable signal.
 

UweX

Joined Sep 2, 2020
33
Just an advice of caution: RF may behave unexpected. So doing experiments here needs also some test equipment. If you are tryping to setup a link, it maybe sufficient to do some experiments and just check on the receiver side the RSSI (Received signal strength Indicator). But if it doesn't match your expectations or first calculations, it maybe hard to identify the reason. So I recommend access to a simple VNA for the frequency range of interest. I didn't see any number, what is the frequency range you are designing your equipment ?
Another word of caution: please also keep in mind the government regulations. They are different in different countries and besides the limitations on the power of the emitted signals there may be also restrictions on the RF-level outside range you are using. In your statement I see the word "processor", that usually has digital outputs with a lot of harmonics. So besides any antenna matching structures you may also need a filter to avoid unwanted emissions. To check for that you need something like a spectrum analyzer covering also at least the harmonics of your circuit.
So doing any test in a highly shielded environment or with an attenuator or just low power settings is a good idea.
 

Janis59

Joined Aug 21, 2017
1,834
For sure, UveX. Test equipment here in the kingdom what cant be seen by naked eye is tha all You have. So, first is antenna analyser. I am exploiting two good and cheap. Because Rhode&Schwartz may buy each millioneer, but (however slightly weaker) result may be got from 1) PS200 or narrower bandwidth SA1201 antenna network vector analyser. 2) by NanoVNA vecor analyser. Better is v2 than v1 because frequency. And best is with 10 inch screen not 4 inch screen. 3) for signal strength and harmonics: GQ-EMS390. Functions - calibrated field strength components and by frequencies and (bit weak but still existing) frequency analyser. Alternatives - HackRF v.001 - not calibrated field strength very well wievable by frequencies and very good frequency analyser. Cheaper alternative - SDR USB stick, the same as HackRF but narrower frequency band. Shall not mention the famous TM195 because it not differentiates one frequency from other, thus useless for adjusting works.
 

Thread Starter

gggccc

Joined Aug 15, 2022
11
868MHz frequency band. Actually, Texas MCU was used in the design and we design using the reference design of this module. Our design results in less than the maximum output power promised in the Texas design, so I'm trying to understand the problem. I thought one of the problems might be that the transmission line is not being calculated, and that's why I asked this question.

The design uses a balun circuit followed by an LC filter. That way, they say it's syncing the line from the MCU to the antenna. But we are not using the same antenna, so we need to add a different matching circuit. For this I do many measurements with a VNA. One of them is to measure the impedance to calculate the matching circuit suitable for the antenna we are using. Any suggestions as to what might be causing the underpower or how to find out?

Thank you for your answers.

UweX
Janis59
 

UweX

Joined Sep 2, 2020
33
For 868 MHz standard PCB material (FR4) can be used.
For further debugging: for components like R or C you may find equivalent subcircuit, taking into account parasitic inductance for capacitors and resistors or capacitance for inductors. You can check in the simulator, if switching from an ideal element to a real 0805 or 0603 component with parasitics makes a big difference. If you are using multiple inductors be aware that they may couple through the magnetic field. That give also a big difference between simulation and reality.
 
Top