RF PCB layout help - design 50ohm tracks to carry 25W of RF power

[YaSaRa_N]

Hi EEVblog community,
I am a newbie Engineer who is doing his first RF project. I have learned to design RF circuits mostly by myself by following various tutorials and guides available on the Internet. however, I am stuck on the RF PCB layout part.

What is my project:

I am designing a Radio communication device ( much similar to a walki-talki) that works on both VHF and UHF  (137MHz to 174MHz and 400MHz to 470MHz) and has 25W of RF output power
I am using two separate RF chains for VHF and UHF. (I tried designing with a single chain but realized that it is almost impossible to match the impedances for such a broad frequency range) I am connecting the VHF and UHF RF chains near the antenna by a PIN diode SPDT switch.
Both RF chains are pretty much identical in structure and they look like this:



I have used Keysight ADS to design the circuit and simulate it. ( I have the non-linear model for the PA) I designed all the subcircuits as modules and simulated them separately and all connected together.  I am very happy with the simulation results.

all the sub circuits of the RF chain ( Driver amp, PA, all the filters ) are all 50 ohms matched. (both inputs and outputs). I have only considered microstrips in the impedance matching of the PA. here is a picture of the PA and a harmonic filter.




Where I am stuck at:

According to the simulations I have created the schematics in Altium but I am now stuck with The layout. I was not able to find a good tutorial on RF Layout which would answer the questions I have. Hopefully, with your guys' help, I will be able to find answers.

What are my questions:

• Can any 50ohm track handle 25W of RF power? If we consider JLCPCB, with their two available PCB stack-up options we need to draw either an 11.6 mil track or 5.8 mil track for a 50ohm trace. Can a 5.8 mil track handle 25W of RF power?   
• How should I place the components and do the layout ?
  I am currently thinking of placing components of each subcircuit as close as possible and connecting them using wider but shorter PCB tracks. After that connect the input and outputs of each subcircuit with 50ohm tracks (as short as possible). Is this the right way to do this?   
• I have seen in many places that if your track length is shorter than 1/10 of the wavelength of the highest frequency (in my case for 470HMz it's 63.8 mm) you do not need to really care about the track impedance. Does this valid for only individual tracks on the PCB or for the whole RF chain. (In my case I am pretty sure that none of the tracks will exceed 63mm but for sure the whole RF chain would be longer than 63mm)   

hopefully, with all your responses I would be able to learn how to do the layout properly.
Thanks in advance.
Cheers
 

[BigBoss]

If you work with small signal analysis only, you will be very regretful at end of the day.
Small signal analysis is necessary as a starting point but not sufficient.
If a High Power RF Amplifier wants to be designed , Load Pull, Source Pull, Nonlinear Gain, Nonlinear Stability, Efficiency etc. ALL have to be carefully inspected and analyzed. PCB layout should also be simulated with related components in EM environment.
Otherwise a strong disappointment will be inevitable.
I recommend you to search about RF Power Amplifiers then try to implement it.

Note : ADS can be used for PCB Layout.

[radiolistener]

For what reason you're using harmonic filter on RX path? Antenna is linear device and doesn't add distortions, isn't it?

Also I think it's better to put harmonic filter between driver and PA.

[richard.cs]

To be clear did you say you intend to use PIN diodes to switch TX/RX? At this power level that's... unconventional. Not impossible to make work but difficult and probably not the best solution. Diode switching is normally used for signal levels.

25 W at 50 Ohms is 700 mA. That's pushing it in a 6 thou track at DC (perhaps a 20 C temperature rise), at 470 MHz when the skin depth is only about 20% of the track thickness it'll run crazy hot. And that's only in the areas where it is actually 50 Ohms, around your filters some areas will have wildly different impedances and higher (or lower) currents.

[JohnG]

To be clear did you say you intend to use PIN diodes to switch TX/RX? At this power level that's... unconventional. Not impossible to make work but difficult and probably not the best solution. Diode switching is normally used for signal levels.

PIN diodes are used for RF power switches. My first job out of college was designing RFPAs for mobile radio (when mobile meant you could drive it around in your vehicle). All the PAs designed by our group used PIN diode T/R switches, and the designs ranged from 15-35W at 800 MHz to 60W low band (~30 MHz).

They were robust, reliable, low cost, and small. Not without challenges, though. Tuning out the parasitics was a bear at 800 MHz. I had to fight to get us to qualify a surface mount PIN diode because you could not control the parasitic inductance of a leaded diode in a production environment. That should give you some idea how long ago this was...

One nice thing about leaded parts was that they did not drop off the PCB when the solder melted due to high VSWR!

John

[richard.cs]

PIN diodes are used for RF power switches. My first job out of college was designing RFPAs for mobile radio (when mobile meant you could drive it around in your vehicle). All the PAs designed by our group used PIN diode T/R switches, and the designs ranged from 15-35W at 800 MHz to 60W low band (~30 MHz).

Sounds like I am mistaken, at work we almost exclusively used fet based switching in that power and frequency range. Don't you need amps of DC to keep the diodes hard switched under those conditions?

[geggi1]

Microsemi have this publication called "THE PIN DIODE CIRCUIT DESIGNERS’ HANDBOOK"
https://www.ieee.li/pdf/essay/pin_diode_handbook.pdf
This will get you started on designing PIN diode switching circuits.

[JohnG]

Sounds like I am mistaken, at work we almost exclusively used fet based switching in that power and frequency range. Don't you need amps of DC to keep the diodes hard switched under those conditions?

To be honest, I don't remember the current. Pretty sure it was not amps, though, maybe a hundred mA or a few? These were push-to-talk radios and they were only forward biased during transmit, so generally is was not a power consumption issue overall.

The thing with PIN RF switches is that they seemed to be really robust and the off-state capacitance was really small.

John

[RoV]

Sounds like I am mistaken, at work we almost exclusively used fet based switching in that power and frequency range. Don't you need amps of DC to keep the diodes hard switched under those conditions?

PIN diodes, in conduction state, can normally handle an RF current much higher than the DC bias current: this is the magic of stored charge.
In very simplified terms: if the diode has a storage time which is N times half period of the RF waveform, it can handle an RF current approaching N times the DC current applied.