Can we have impedance control and contain costs at the same time?

We receive a fair number of questions regarding impedance, and not the kind of questions you might think. In general, the industry is full of smart people who understand the theory and practice of good signal integrity. No, the usual questions are more on the order of, “Why is my flex as stiff as a ruler?”

There is one basic tenant in flex circuit design: If it’s good for mechanical performance, then it’s going to be bad for me electrically. And of course the opposite is true: If it’s good for electrical performance, then it has to be bad for me mechanically. This is especially true when it comes to impedance control.

While impedance theory could (and does) fill a textbook, we want to take a more “practical” approach with a bend toward low-cost manufacturability. Theory is important; flex circuit designers and engineers need to know what will actually happen to the physical flex circuit when asked to meet specific impedance requirements.

Dielectric materials for flexible circuits are a somewhat limited palette; only so many solutions are available. Let’s look at a real-world design, and how a simple request for impedance control can lead to some real issues with flexing and assembly. We start by reviewing which physical properties impact impedance, in their order of importance.

Dielectric thickness is enemy No. 1 in our desire for a flexible circuit with good signal integrity. Because of the relatively poor dielectric constant of common flex materials, the most direct way to drive impedance values higher (90-120µm) is to increase our dielectric core thickness.

One of the most common films used in flex manufacture is polyimide, with a dielectric constant (Σr) of 3.4. (Measured at 1 kHz, remember this “constant” changes at higher frequencies.) A “standard” flex uses a 1 or 2 mil core and has an impedance of roughly 35µm. To drive that impedance up to 100µm requires boosting the core to 5 mils. It is not too difficult to see why a “small” change in electrical properties can impact overall flexibility. Same circuit construction, different flexibility. Figures 1 to 3 show a “simple” microstrip. The “problem” really begins to compound itself when we move to more complex signal arrangements such as a stripline (Figure 3).

As you can see, the additional reference plane required for a stripline construction adds a second layer of dielectric. Compounding the problem is the requirement to add an adhesive layer. Common adhesives have higher Σr values than polyimide, making our stackup thicker in order to get greater separation from signal to return.
Using a PTFE (Σr 2.0) can be a real advantage here. It is possible to reduce the core dielectric from 5 mils to 3. Like all “advantages” however, these materials can come with a cost penalty.

After dielectric thickness, conductor width is likely tied for the biggest impact on impedance. Simply put, the narrower the conductor, the higher the impedance value. This is where we can gain the biggest advantage in flexibility with the lowest cost. Reducing conductor width from 5 mils to 2 mils can increase impedance values by roughly 20%.

The downside: Finer line widths come with a higher cost due to process losses from lower etch yields. Finer line widths are less able to carry current, which brings us to our next player in impedance control: copper weight.

Copper weight (thickness) does have an impact, though a bit smaller than the others. The rule here is the lighter the weight, the higher the impedance. This is primarily because thicker copper requires more adhesive/dielectric to maintain distance from the signal trace to the reference plane. Make a real effort to avoid running signals on a plated layer. Plating adds thickness to the conductor, driving impedance down. If you must run signals on an outer layer, consider “button plating” to avoid increased copper weight. Ideally signal layers should have 9Ωm or even 5Ωm copper.

The biggest impact on flexibility is perhaps not so much with the conductor but the shields. Heavy copper shields are going to create a stiff circuit. The “easy” answer is to cross-hatch our shields, but it has been my experience that the amount of cross-hatch required for significant increases in flexibility can make for poor shielding.
Silver epoxies can be an excellent alternative for shielding, but do come at a higher cost in both materials and processing.

One of the newest tools in our bag of tricks is the relatively new silvered/metalized films. These films have several advantages over both silver epoxy and copper planes. They have excellent dynamic flexing ability with cycle numbers higher than both copper and epoxy. They also commonly come with anisotropic pressure-sensitive adhesive, which can reduce or eliminate the need for plating, thereby cutting cost and increasing flexibility even more. One potential downside: Abrasion resistance is not quite as good as a polyimide.

Remember to review the application. Dynamic flex applications require careful consideration of the entire stackup and desired electrical properties to provide a long service life. Impedance control comes at a price, both in flexibility and procurement costs. Careful review upfront will return a design that will please not only the mechanical and electrical engineering team, but the purchasing agent as well.

Mark Verbrugge is a field applications engineer at PICA Manufacturing Solutions (picasales.com); This e-mail address is being protected from spambots. You need JavaScript enabled to view it . He and co-“Flexpert” Mark Finstad ( This e-mail address is being protected from spambots. You need JavaScript enabled to view it ) welcome your questions. Mark and Mark will be speaking in September at PCB West 2012 at the Santa Clara (CA) Convention Center.