Current Issue

Bill Hargin

The "founder" of TMI asks, How much is more analysis worth?

Just because something can be done doesn’t mean that it should be done.

A few customer encounters this past month caused an issue to ricochet around in my mind like a 1970s pinball machine. I’m referring to a trap we’ve all fallen into: analysis paralysis.

Three interrelated definitions I have for analysis paralysis are worth enumerating:

  1. The condition of being indecisive while overanalyzing alternatives. (Classic analysis paralysis.)
  2. Allowing a project to mushroom into something bigger than it needs to be to get the job done. (This column is a good example.)
  3. Using data from the most expensive tools you own just because you have the tools or the data (e.g., it’s expensive and took a lot of time, so it must be good).

It’s not that analysis or expensive tools aren’t good, but their employment is an optimization process.

Relative to the above, I can’t and won’t lecture on trying to rely on overanalyzing things or using “too much information” as if I have a solid handle on it. When non-engineers say to me, “That’s TMI,” I say, “I invented TMI.”

To continue reading, please log in or register using the link in the upper right corner of the page.

Read more: Winning the War against Analysis Paralysis

Bill Hargin

A methodology for selecting the right material and the right price point.

When I started writing this column a couple years ago, I wondered how much I’d have to say. An experienced media guy told me to watch my inbox for topics and questions that may be of general interest. That turned out to be excellent advice. Here’s one such example.

“What is the best laminate for a loss budget of x dB for y inches? I was thinking in terms of Panasonic Megtron 6 or something like it.”

Megtron 6 is an excellent material, but it’s not cheap and it’s not the only horse in the race. My response was to focus on a loss and material-planning methodology rather than making a firm material recommendation.

Why we care. Everything that improves material performance – in particular, reductions in loss – comes at a price. Loss versus cost is a classic optimization problem. Designers want to pay just enough to meet loss requirements, but not more than they need to.

In the past, speeds were slow, layer counts were low, dielectric constants (aka Dk or Er) and loss tangents (aka dissipation factor, or Df) were high, design margins were wide, copper roughness didn’t matter, and glass-weave styles didn’t matter. We called dielectrics “FR-4,” and their properties didn’t matter much.

To continue reading, please log in or register using the link in the upper right corner of the page.

Read more: Cutting Your Losses Upfront in PCB Design

Bill Hargin

The impedance implications of the trapezoidal trace.

Until recently I thought those who believed in rectangular traces were about as common as those who believe in square waves and a flat earth. Recently, though, I came to realize it’s not as clear as I thought, not only for newbies but in general. Over the past 25 years, I’ve acquired a good number of books on PCB design and signal integrity, and you wouldn’t know from reading most of the industry literature that traces were anything but rectangular. Interesting, right?

If you’ve read previous “Material Matters” columns, you may recognize the following cross-section from our Z-solver software. Among other things, it shows that the base of a trace, facing the core dielectric, is wider than the side of the trace that faces the prepreg. As such, the trace trapezoids face both up and down in a multilayer stackup. There’s no relationship to the layer number or whether the trace is on the top or bottom half of the board. For this reason, some including me – but not everyone – avoid using terms like “top” or “bottom” with regard to trapezoidal traces.

To continue reading, please log in or register using the link in the upper right corner of the page.

Read more: Etch Effects Explained

Bill Hargin

Weight is still used as a determinant for copper thickness. Why?

Sometimes my columns tie to issues or stackups that appear in my inbox each week. I’m occasionally asked why 0.6 mils (15µm) is often used for the thickness of 0.5-oz. copper, rather than 0.7 mils (18µm), and similarly why 1.2 mils (30µm) is often used for 1-oz. copper instead of 1.4 mils (36µm). If you’re curious about the details, or if none of these numbers seems familiar, here’s a quick primer. The thickness parameter “t” in FIGURE 1 shows the thickness we’re interested in here.

 

 

 

To continue reading, please log in or register using the link in the upper right corner of the page.

Read more: Actual Copper Thicknesses (As Opposed to What You’ve Assumed)

Page 151 of 211