PCB Tech - How to put through holes on the PCB?

I can't find an explanation why people want to place as many vias (~50) as possible along the copper traces (or anywhere on the PCB), these vias transmit high frequency RF (100 MHz to GHz) signals.

In my case, there are two ground planes (dumping) on both sides of my circuit board. My intuition is that, anyway, through the via, you create a mutual ground between the two ground planes, so that no signal passes from the top ground plane to the bottom ground plane along the edge of the PCB.

This is because the periphery of the PCB is like a loop of RF leakage current. In order to minimize the induced current loop, a "shortcut" must be provided so that the two planes can be in direct contact? Am i right?

It is related to impedance. For high-frequency signals, if you have to walk a long distance from one piece of GND copper to another piece of GND copper, it will experience high impedance, so voltage may be generated. In other words, for high frequencies, unless there is a short path, GND is not a true GND. Therefore, sometimes you need to add vias to stitch different GND copper wires together.

I think the first-class answer is too much for me. I have work and family. The basic idea is that if the return current has to follow a long path or bypass something, there is inductance, and even if the DC resistance of the path is low, the impedance of the path will be high at high frequencies.

You are actually talking about two very different things. One of them is through stitching, which is a grid-like pattern of vias. You may see two ground planes connected.

The other is through the fence, which is a via that completely surrounds the RF traces on all sides, except for the end that can terminate at the antenna or the like.

Now, on any PCB, whether it is carrying a DC or 5GHz signal, the ground plane can be spliced together with a certain conventional pattern (within a reasonable range, you can do this according to your own preferences). It ensures that any copper islands that might go unnoticed are actually grounded (it happens in our best place), that everything has the shortest possible ground path, and that the ground is generally kept as ground.

Now, the ground is not a particularly useful concept at high frequencies. Even at DC, there is ground return current flowing through your ground plane, and copper does have a little resistance. The ground is just a fantasy, there is no magical copper plate that has the same potential at every point. Grounding simply means that we are trying to maintain close to the same potential, with varying degrees of success, and choosing this potential as the 0V reference voltage for the rest of our circuit.

However, as long as any current starts to flow, it will generate a voltage on the current flowing through the copper, which can either spread the current out or make our ground "bounce", depending on which part of the PCB you are looking at, and ...Nothing really has the same potential. Even the ground. Sewing is considered to be a "best practice", and to ensure that the ground is more tightly coupled in terms of voltage and potential between one part of the ground and another part of the ground through a low-cost way.

Like they said, you will never have too many ground pins. It also applies to ground vias.

Another important use of vias is thermal performance. Vias, as a very good thermal conductor, is definitely much better than FR4. Whenever vias are used for thermal performance, you will usually see that their packaging is as reasonable as possible, covering a copper plate, which will burn out when the circuit board is energized. Even in more modest requirements, it is almost always preferable to have a tighter PCB with less thermal coupling. If the temperature on the PCB is more similar, then everything that drifts with temperature (almost everything on the circuit board) will drift together.

Now, for the RF board, the situation is different. In other words, the return current no longer really leaves trouble for your ground plane. At low frequencies, our ground loop current spreads a bit and takes the shortest geometric path to reach the lowest potential (things that ground our ground plane, like the ground of the power supply, the battery).

At a uniform and moderate frequency, the ground return current is controlled by the reactive component of the impedance. The reactance (imaginary part) of a complex impedance is a measure of impedance. This is because the various elements in the circuit store energy at a given rate. In contrast to the resistance (actual) component, the resistance (actual) component is simply the impedance caused by energy consumption. . A certain ratio.

Reactance is frequency dependent, because the stored energy will not only disappear, it will return to the circuit, and the speed at which something swings will determine how much time (and therefore how much energy is needed) stored before it arrives to be released with the next swing.

Reactance is always due to two kinds of energy stored in an electric field or a magnetic field. And capacitance and inductance are only measures of the ability to store energy in an electric or magnetic field. Now everything is starting to come together, don't you?

The current will follow the path of least impedance. As the frequency increases, our return current needs to minimize the inductance and capacitance formed between the positive current and the ground return current. It hopes to minimize how much energy can be stored in the parasite.

Our ground current will flow directly under the path of the original current as much as possible.

As you can see, 100MHz is not interested in the beautiful short ground path we provide. In fact, it completely ignores them.

This is why the through-hole stitching and fences on the RF board are completely different from those related to the ground or maintaining a good ground potential. Yes, I finally have to answer your question!

Electromagnetic waves in the sub-300 GHz range, which we usually call radio waves, are the result of acceleration of charge carriers. Whenever any charge carrier is accelerated, electromagnetic waves will be emitted. Due to some serious physics beyond this range, it will contain a little energy, momentum and angular momentum, and the radiation will be just right to preserve them. Of course, it can interact with long-distance charge carriers, and this momentum, angular momentum, and energy can be transferred back to other charge carriers, thereby accelerating them. Of course, this is the physical basis of all radio technologies.

For the charge carrier to be accelerated, it must be mobile. In other words, we need command.

The scary fact here is that everything that conducts electricity is an antenna, and will soon radiate and pick up almost all frequencies high enough to make the wavelength small enough to fit the conductor.

Our only real defense is to make all our conductive paths too short to be effective heat sinks at the frequencies of interest.

Therefore, the best practice is to switch any copper pouring on the RF board, where the via pitch is at least λ/10 of the target highest frequency λ/10, that is, λ/10. The smallest. If possible, you really want to aim at the λ/20 pitch in the via in a grid pattern.

This brings us to the most terrifying part, arguably the most important stimulus and the main driving force behind the passing fence: nothing is directed by the charge carrier...

...Is a great electromagnetic wave guide.

This is correct, everything we call insulators, dielectrics, including vacuum or beautiful PTFE wire insulation layer or our PCB FR4 laminate-they are all conductors for current, but for electromagnetic waves. They are conductors of electromagnetic waves. On the other hand, a conductor is an insulator of electromagnetic waves (reflectors may be a better analogy).

If you have cable TV or the Internet, you are familiar with those 75Ω RG6 or RG59 coaxial cables that carry it and make it work. Looking at the cross section, you will see the white material between the shield braid and the single center conductor. That is a dielectric foam. The signals that travel along the cable are not carried by copper conductors, they are carried by white foam. Coaxial cables are not conventional old conductive cables. Coaxial cables are waveguides.

When the frequency becomes high enough that the wavelength is similar to the copper feature size on the PCB, you must fight a long-lasting battle to bottle up all these electromagnetic waves and move them where you want them to go instead of where you don’t. place. And they will happily pass through the delicious dielectric core of your PCB, made of FR4, all the way to the side of the circuit board, exuding like hell-like little swinging bats.

Your two ground planes will be excellent waveguides! They will bounce between them on the way off the side of the circuit board, and may directly enter the RF measurement equipment used in the FCC certification, and you will fail.

Therefore, the grid spacing of the vias we lay is tighter than the shortest wavelength we need to worry about. Not less than λ/10, but better λ/20. Like the grid on the microwave door, these vias are packed too tightly so that these waves will not leak out.

Passing fencing is for the same reason, but usually because we are actually trying to radiate some waves, but we want to bottle them until they can escape through some kind of antenna feature or any way we want. Under normal circumstances, the fence can also be used as the external part of the waveguide, if you want, it can also be like a flat coaxial cable. In addition to the carefully calculated stripping size of the microstrip, the gap is also important.

In any case, the final answer to your question: all these filters are to keep shaking.