pcb substrates?

[s2000]

Hi, I've been doing a bit of studying recently on different pcb materials available, the problem is that most of the literature I'm reading is related to working in the high Ghz region. I would like to know what pcb material is best suited for doing a 500/600/700w fm amplifiers? I have a circuit diagram and would try to build it with copper island style with no pcb inductors.

I have had a look at some pallets of the same wattage, some appear to be PTFE (teflon) based and others appear to have used what looks like FR4 (or maybe FR4 TG varient?). Obviously, it would be great if FR4 was possible as it would mean low cost and facilitate a spacious board layout. My concern is that it might not be able to handle the heat involved and possible even the higher voltages/currents? Conversely, if Teflon is advised, it will mean microminiaturising the pcb to keep costs down! What are the limits of FR4 power/heat wise before you need to move to a different substrate?

I have had a look at my pcb house, they do a number of different substrates. FR4, FR4 varients (TG150-160 or TG 170-180), Rogers 4003c or 4350b. The Rogers 4350b one often pops up in various literature, and from what I understand, it's not actually PTFE but a kind of lower cost alternative. I would imagine its pretty expensive though as its the only substrate they don't have live prices for!

I understand that there are a huge number of factors in pcb design and that it's no trivial subject. Having said that, I have become a bit of a fan now I have nice routing software and want to learn more. Seems to be a subject that's not often discussed with regard to amplifier design.

Any help would be great

[nrgkits.nz]

High power requires thick wire or coax of a specific impedance and length for impendace matching. I try to avoid PCB etched inductors if I can avoid it.

[s2000]

Yes, on this circuit, its mainly coax impedance transformers and atc capacitors. No etched coils. I will need some hand wound coils for the filter though.

I've been looking around and seen a few people using Arlon Tc350, might have a go on that.

[thewisepranker]

Most of those reference designs use that particular substrate just because they can, and it makes their product look better. It's not essential, especially at VHF and even UHF.
There is nothing particularly spectacular about the power handling of PCBs with PTFE or ceramic substrates. You get less parasitic loss, and less heating as a result, but that is absolutely minimal - the effect of which is ever decreasing as the operating frequency decreases. You also get better creepage performance at fixed distances relative to FR4, but not really much since we've got to match it all to about 25 to 100 Ohms anyway so the clearances are massive relative to the min. requirement, and are inherently about the same relative to those on FR4.

You shouldn't need any of the exotic types of PCB material to mitigate against high temperatures, not that they will actually help you anyway. If you decide that you need it for this reason, you are likely going to suffer from self-desoldering of certain components like series inductors and output device pins. Fixing the power dissipation problem is always preferable over trying to "go up one level" in PCB material.
PTFE itself is almost as bad a thermal conductor as FR4, so there are no gains to even be had.

There really shouldn't be a reason to need anything better than FR4 up to a few GHz. I have seen a picture of an absorptive filter floating around using something more exotic than FR4 but that is probably because the elements weren't in the right order, resulting in nodes that are particularly fussy when it comes to stray capacitance.

Most of your problems will come from lack of copper thickness rather than anything to do with dielectric material. Use 2oz copper if you need it, it's way cheaper than Rogers material. Also prefer squares rather than rectangles on your pads/tracks if you can, because copper resistance is in Ohms per square. That's an actual unit.

[nrgkits.nz]

Another thing to watch out for with FR4 is it will start turning to glass under high temperatures, this could start to happen if there’s a mismatch at high power or no or insufficient cooling. Output low pass filters should be swept on an analyser to confirm their shape, taking note of return loss at the input and insertion loss, ideally you want return loss to be atleast 25db, preferably 30db at the input from 88 to 108 and insertion loss no more than 0.3db. I usually build my LPF’s with 9 poles and 27p ATC100b or 100c or e series caps depending on the power requirement, the inductors are fine tuned while doing a sweep on the analyser with a directional coupler and one end of the filter terminated into a load.

[s2000]

Thanks for the help guys, I've decided to do a dummy run on homemade fr4 board to see if its ok first. Will let you know what happens