FR4 or do I nead better?

[FriedMule]

I have looked at board prizes at such places as JLpcb and PCBway, the first is way way cheaper but do only offer FR4 while the last offers several solutions. Last do unfortunately not write prizes for special boards but have 5-10 x prize for standard FR4.

The question is if FR4 are gotten so good that I can run 2GHz on FR4 or do I have to go up a notch? It's for a 1.4 inch (31 mm) x 2 inch (51 mm) differential probe.

[vk6zgo]

I have looked at board prizes at such places as JLpcb and PCBway, the first is way way cheaper but do only offer FR4 while the last offers several solutions. Last do unfortunately not write prizes for special boards but have 5-10 x prize for standard FR4.

The question is if FR4 are gotten so good that I can run 2GHz on FR4 or do I have to go up a notch? It's for a 1.4 inch (31 mm) x 2 inch (51 mm) differential probe.

Don't let the fanbois of other materials beat you into using something more expensive, & almost always, less durable.
FR4 has been used at much higher frequencies than 2GHz for decades.
Studio to transmitter & OB links are commonly around the 7 to 13GHz range & use FR4.

[FriedMule]

Fantastic, that's what I thought. :-)
It may have been "in the good old days" that FR4 was made in a far less quality.
What about the trick with angling the board 45 deg to avoid going only across or lengthwise with the fibers, is that even necessary?

[vk6zgo]

Fantastic, that's what I thought. :-)
It may have been "in the good old days" that FR4 was made in a far less quality.
What about the trick with angling the board 45 deg to avoid going only across or lengthwise with the fibers, is that even necessary?

I doubt that it is, but I must confess, I have only been concerned with such devices after they were in service.
The original design EE would have made such choices if necessary.

I have seen some shoddy FR4, but it is usually pretty good quality.
I have seen some very poor (counterfeit?) Rogers material, too, unfortunately, in finished equipment.

[FriedMule]

Thanks a lot, I have calculated the trace-length from input to component (not counting components) and further to the next component and so on. The total trace length is 2 × 1.45 inch (36,5 mm) so in total 2.9 inch (73 mm). If I understand correctly, should that not be any problem below 2 GHz?

The FR4 has different heating-ratings, do a better heating give a better board for the signal?

[FriedMule]

Fantastic, thanks a lot!! :-)

[Mechatrommer]

If I understand correctly, should that not be any problem below 2 GHz?

you must gather knowledge as much as you can. from reading, signal loss will not be an issue even beyond 2GHz, but if trace length is not impedance controlled AND more than like quarter wave length, you'll run into reflection and signal integrity issue if that matters. currently discussed 3GHz NanoVNA is an example of very cheap FR4 board. but layout need to be number 1 in mind. about the FR4 pcb price, if you just climb one stair up the ladder to like low end Roger, you'll run away if manufacturer quote the price. go check seeedstudio just for FR4 with impedance controlled, the difference is near $100. pcbway Roger is near $300 addition just for basic 100x100mm board. make sure your prototype board is right in the first place, otherwise you'll need investors. ymmv.

[exmadscientist]

about the FR4 pcb price, if you just climb one stair up the ladder to like low end Roger

Rogers material is not just one step above "FR-4", because "FR-4" doesn't really mean much. It's become quite a generic term.

At its most basic level, FR-4 is a garbage-grade material one step above FR-2/CEM. But no one actually buys that anymore (at least at our level; it is still used in extremely cost optimized designs). Rather the term "FR-4" has come to apply to any material that can be processed like FR-4: it follows the same manufacturing flow, and is the same sort of stuff (E-glass–reinforced epoxy), so it's acquired the same name. These materials are more properly called "FR-4 compatible", but the nomenclature is what it is.

High-quality lead-free FR-4–compatible materials are very good indeed. Some of them are optimized for signal integrity, one example of such being Isola's FR-408. This class of material is where one should look if one wants the best signal integrity per dollar.

Rogers-class materials do not use traditional FR-4 processing, so their extremely high cost comes both from the high base material cost and from requiring a different fabrication flow. But they are, nonetheless, where you have to go for deep RF designs.

[FriedMule]

I have the possibility to choose a board type JLpcb calls FR4 standard 130-140 deg C or the FR4-4 TG155. To me do they sound alike, except for a tiny higher heat resistance. They offer impedance option (JLC7628) whatever that means.
I have tried to impedance mach the traces to 50 Ohm by making 11.55 mil (0,3 mm) wide traces.

About reflections, since the probe shall function from DC to 2 GHz, do I not know how to avoid reflection, what is the wavelength?

[halexa]

Hi

FR4 will do fine, its more important to look into the glass fiber weave used.
https://www.isola-group.com/wp-content/uploads/Understanding-Glass-Fabric.pdf


I would not recomend a standard 1080 weave for high speed signals.

[Mechatrommer]

About reflections, since the probe shall function from DC to 2 GHz, do I not know how to avoid reflection, what is the wavelength?

https://searchnetworking.techtarget.com/definition/wavelength

(v is speed of light) with 0.97 speed of light factor, 2GHz quarter wavelength is 3.6cm, be careful when you have traces longer than that.

[Mechatrommer]

(v is speed of light) with 0.97 speed of light factor, 2GHz quarter wavelength is 3.6cm, be careful when you have traces longer than that.

vp=0.97c means dielectric constant is 1.015. That has to be air.
Normal FR4 has a dielectric constant of around 4, so vp is about 0.5c in FR4.

this was a surprise to me when tuning a 433MHz dipole antenna, measuring the length and back calculating, it turned out 0.97c must be brainfart or the formula/idea i was using is obviously wrong.

[Mechatrommer]

Is it a free standing air dielectric antenna? If it is a PCB antenna with vp=0.97c, I'd be very surprised.

well, dipole with 135 degrees apart, yeah free standing in the air. maybe that make sense. for 0.5c on pcb, 2GHz quarter wavelength should be somewhere 1.87cm.