Microwave Tech - Why is the dielectric constant DK value of FR4 of the rigid-flex board factory erratic?

Recently, the customer of the rigid-flex board factory specified that the dielectric constant of FR4 must be 4.5. However, the manufacturer cannot guarantee such an accurate DK value for FR4. Today, I will explain to you why the dielectric constant (DK) value of FR4 (glass fiber epoxy copper clad laminate) is usually marked between 4.2-4.8?

Adding woven glass to the printed circuit board (PCB) material can increase the structural strength of the material. This helps to increase the mechanical stability of the laminate material, but does this have any effect on the electrical behavior of the material? One of the classic problems with woven glass-reinforced laminated PCB boards is that the "glass weaving effect" may negatively affect the electrical performance of the high-speed or high-frequency circuits processed on these laminates.

Depending on the specific resin system of the fiberglass laminate, the dielectric constant (Dk) of this material actually varies with position in a very small periodic manner. These small areas with different Dk values may be caused by the unique physical weaving structure of glass fibers, in which the glass fiber woven fabric is woven from glass fiber bundles, and there are small open areas between the glass bundles. Among them, the Dk of the glass fiber bundle is usually about 6, and the Dk of the laminate in the open area between the bundles is much lower than that of the glass fiber bundle, usually about 3. Since the impedance of high-speed/high-frequency transmission lines is highly dependent on Dk, the change of Dk value has always been a problem for circuit design engineers using woven glass laminates.

The prepregs and copper clad laminates are what we often call PP and core boards. The medium is a mixture of epoxy resin and glass fiber cloth. (Rigid printed circuit board)

There are many types of glass cloth, each of which has different thickness and weaving size. Below is a list of some commonly used glass fiber models in the rigid-flex board factory. You can see the grid-like fiberglass cloth very intuitively. Some models have large empty windows, and some models have very small empty windows.

For the specific impact of the glass fiber effect, please see the legend below, which are the impact on impedance, the impact on delay and the impact on loss.

It should be noted that the glass fiber effect has the greatest impact on high-speed long traces, and low-speed systems or very short traces can be ignored.

Here is an example to illustrate how glass fiber affects the microstrip transmission circuit: Consider a double-sided copper clad laminate with top and bottom (signal transmission and microstrip ground plane), and its Z axis (thickness) direction at 10GHz The dielectric constant DK is 3.0. Usually at higher frequencies, such as millimeter wave frequencies (30GHz and above), changes in DK will affect the performance of the material. For example, a quarter-wavelength of a signal propagating through a circuit at 77 GHz is about 0.024 inches, which means that one-eighth wavelength is about 0.012 inches. Theoretically speaking, when electromagnetic (EM) waves encounter any type of DK change in their propagation medium at a wavelength greater than a quarter of the frequency of interest, the propagation of electromagnetic waves will be disrupted and resonance may occur.

Practical experience shows that even anomalies as small as one-eighth of the wavelength can cause electromagnetic wave propagation problems. Circuit laminates with one-eighth wavelength or higher in the glass or glass beam may be subject to the distribution of the glass beam (and the corresponding Dk change) resulting in irregular performance. Given the types of glass that can be used to reinforce different circuit laminates, it is not uncommon for some of these glass types to have gaps of one-eighth wavelength or greater at 77 GHz (0.012 inches).

It probably means: the millimeter wave has a small wavelength, and when its size is equivalent to the "gap" of FR4 glass fiber, the fluctuation of its DK will change greatly. This is one of the reasons why FR4 is not suitable for millimeter wave circuits.

Regarding countermeasures, it is mainly material selection, design avoidance and production avoidance. Material selection circumvention:

1: Use glass fiber cloth with small empty windows. Also known as flat glass cloth, open fiber cloth, etc. Avoid from the source the existence of fluctuations in the effective dielectric constant of the glass fiber cloth empty window. For example: 1067/1078/2116, etc.

2: Use multiple PP overlays to reduce the probability of window exposure. The method is feasible. Unless the medium is thicker and requires multiple PP overlays, my personal opinion is not as good as the first one. One is the cost, and the other is that it is easy to slip when more than three PP sheets are manufactured.

3: Use low dielectric constant glass fiber cloth to reduce the difference in dielectric constant between glass fiber and epoxy resin, and reduce the difference in effective dielectric constant inside and outside the empty window. Note: Low dielectric constant glass fiber cloth is usually only equipped with ultra-low loss sheets. In other words, the high-speed sheet is often said to have a high cost.

Design circumvention:

1: Important signals go with lines with a certain angle, 3°, 7°, 11°, etc., basically do not increase the cost, but the layout is more difficult to do, I think all the layout partners have already held this hatred in the small It's on the books.

2: The rotation angle layout of important signals increases the design difficulty. The small trick is to rotate the chip as a whole after FANOUT.

3: After the normal design, rotate the jigsaw by 7° during the jigsaw puzzle. This is equivalent to a full-page 7° wiring operation.

Production avoidance:

The normal design allows the rigid-flex board factory to rotate the material during production. The core board we use is cut from the big material. The big material is square. Rotary cutting will inevitably reduce the utilization rate of the board, and PP must also use larger sheets. Will increase manufacturing costs.