Advertisement
Articles
Advertisement

Guest Blogger: Design Challenges Are Getting Smaller (at Least In Terms Of Packaging)

Tue, 03/03/2009 - 7:53am

LISTED UNDER:

Small packaging is becoming a necessity for every day design

By Duane Benson, Marketing Manager, Screaming Circuits

It seems like everyone’s talking about the economy these days, but I don’t want to. As far as I’m concerned, the best thing to do is the same thing we did the last half dozen times the world was coming to an end: just keep doing stuff. For an electrical engineer, that means building hot new designs. Well, hopefully, not too hot.

"The best thing to do is the same thing we did the last half dozen times the world was coming to an end: just keep doing stuff."

It seems like everyone’s talking about the economy these days, but I don’t want to. As far as I’m concerned, the best thing to do is the same thing we did the last half dozen times the world was coming to an end: just keep doing stuff. For an electrical engineer, that means building hot new designs. Well, hopefully, not too hot.

On my Screaming Circuits blog, I spend a lot of time writing about new advanced packages such as the QFN (quad flat pack, no leads) and CSP BGA (chip scale package, ball grid array).

These tiny new packages are no longer just a curiosity or the province of obscure applications. They’re becoming a necessity for every day design. In today’s column, I’d like to look at how demands on engineering teams are changing in terms of design trends.

Back in the 1980s, power supply sections usually included a couple of LM7805s or LM7812s in TO-220 packages. Most designs seemed to require somewhere around half an amp or just a bit more, so that worked out well.

With the 7805, if the draw got up closer to a full amp, the solution was the addition of a heat sink and then a bigger heat sink and then maybe two parts in parallel. Linear supplies and heat removal were easy that way.

Twenty years later in 2005, designs still used a TO-220 packaged LM7805T, or a low dropout (LDO) equivalent linear regulator. Same story. Add a heat sink if necessary, then a bigger heat sink and then start paralleling them. Not much difference. Linear supplies were still pretty easy and the rules hadn’t changed much.

Move just a bit forward to 2009 and enter the QFN. Today, we see a little MCU board. The venerable LM7805T has been ditched for a low drop out MCP1726 regulator. The old TO-220 package was about 10mm x 20mm plus the area needed for a heat sink to deliver a full amp.

This new part comes in an 8-lead DFN (same thing as a QFN, but with leads on two sides instead of four). It’s 3mm x 3mm and, with proper PCB design, can supply the same 1-amp that the TO-220 packaged part could deliver with something like 50 times the surface area, before heat sinking.

It’s still a linear supply, but things are different now. You can almost put the little 1726 between two of the legs of the 7805. The pin pitch on the MCP1726 is 0.65mm and it has a 2.5mm x 1.5mm exposed metal pad on the bottom side of the part between the eight leads.

All of the leads are under the part. There pretty much is no hand soldering this part. And the land pad on the PCB has to have properly designed vias to extract the heat from the center pad and move it to the other side of the PCB.

There is also a TPS62600 half-amp step down switching regulator in a 1.9mm x .9mm chip scale BGA package. The motor driver attached to this MCU is an A3901 – in a 3mm x 3mm DFN package with 0.5mm lead pitch. It seems like all the new parts are coming out in these super small form factors. Cost containment and cell phone demand seems to be driving a lot of the package size reduction. It’s cheaper and one part can service both markets that are size sensitive and those that are not.

There’s still a place for big old parts, but if you want to be on the leading edge, you have to know how to use the new packages. With the leads so much closer to (or even right on) the die, performance can be greatly enhanced with QFN/DFN and CSP BGA packaged parts too.

But you have to use them properly. You can’t just slap a heat sink on these things and call it good. PCB design is critical when you are trying to drive current through something that small. Thermal management and manufacturability require very specific techniques in order to get these packages on the board reliably and have them perform up to specification.

I received a comment on one of my QFN posts on my Screaming Circuits blog recently:

“I must admit that as a designer I am totally paralyzed in the face of the QFN. I avoid them despite their superior high frequency and thermal performance vs. the QFP.”

I hear things like this very frequently, and at Screaming Circuits, we see the results of this confusion frequently on the prototype PCBs we assemble.

So here’s what you have to consider

  • The mechanical connection.
  • Thermal management.
  • Signal integrity for high-frequency applications.

The mechanical connection largely comes down to proper design of the solder paste layer in the CAD symbol parts library. You need to segment the stencil opening over the thermal pad in the middle of the QFN/DFN. Check with the part manufacturer’s data sheet first, but if they don’t tell you how to do it, shoot for 50 percent paste coverage. Too much and the part will float up and the leads on the side may not connect.

Too little and you’ll get too much voiding and may not get good ground or heat conduction. And make sure any vias are capped or better yet, filled and plated over.

Thermal management is about proper use of vias and a heat sink pad on the backside of the board. Again, the manufacturer’s data sheet and app notes are the place to look for the best information. But make sure that you don’t leave open vias in any of the pads. Doing so can cause solder to wick through to the other side of the PCB and leave too much voiding under the part.

Signal integrity issues will vary from part type to part type. Some don’t care. High frequency components however may care a whole lot. Often they will require a specialized land pattern on the PCB. Again, make sure you dig deep in the data sheet and app notes to get the application specific requirements here.

With such new components, you’ll most likely need to create your own CAD symbol library part. When you do so with QFN/DFN and CSP BGAs:

  • Check the data sheet to see if you need SMD (solder mask defined) or NSMD (non solder mask defined) pads.
  • Make sure you put vias where they need to be.
  • Make sure the vias aren’t left open. Make a special land pattern for the center thermal pad if necessary.
  • And shoot for about 50 percent solder paste coverage in the thermal pad.

Don’t fear the QFN. Embrace it. Learn how to use it properly and you’ll have an advantage over those more timid about it and similar tiny packages. With a little study and some extra care, you can effectively use the components and can keep your skills sharp and in demand. And that is something quite important, even more so in an economy like this.

Visit The Screaming Circuits blog: blog.screamingcircuits.com

Advertisement

Share this Story

X
You may login with either your assigned username or your e-mail address.
The password field is case sensitive.
Loading

Curated By Logo