Screaming Circuits: March 2010

DB25 - Survey Says...

DB25 with 
issuesI've gotten a number of answers to my question: "What's wrong with this picture?" All good responses.

  • Via between the connector pins
  • Soldermask covering the pads
  • Tight zig zag on some traces
  • Traces real close to some of the pads
  • Uncovered/partially covered via in between the pads
  • No clear pin one markation
  • Tear dropped pads might be helpful

The biggest problem that I see in terms of buildability is the solder mask on the pads. Yes, we can make it work. In a prototype world, we make a lot of stuff work that by all rights should not, but once into production, or for best reliability, that mask is a problem.

The tight spacing between some of the traces and the pads is not inherently a problem, but based on the mask issues and the poor drill alignment in the via, I'm lead to question the workmanship on this PCB to the point that I would be concerned about anything too close to anything else on this board.

DB25 with issues del The third concern that I noted, I'm no longer quite sure of. I just have the photo in hand, not the actual PCB. When I first looked at this photo, I interpreted the lighter area around some of the pads as delamination. We see that sometimes and it's a bad problem. With a second look though, I'm not so sure that I'm not just seeing an inner copper layer with openings around most of the pads. Hard to say just by the photo.

I'm not terribly concerned by the lack of pin one designator. This is the solder-side of the board and as long as this connector is on the correct side of the board, it's not reversible. The via doesn't bother me much either, as long as the connector doesn't have any exposed metal over that side. Most of these would have plastic down there now.

Duane Benson

8 bit vs. 32 bit Microcontrollers

There's a lot of talk these days about the new generation of 32-bit microcontrollers and the demise of the 8-bit controller. I'm a big fan of the Beagleboard and mbed boards (ARM Cortex A8 and ARM Cortex M3). And the Cortex M0 processor looks to be a very promising low-end 32-bit ARM. By the looks of it, ARM could end up ruling the below-X86 world soon.

SP16-1_layout But, one consideration to the 8 vs. 32 discussion that I haven't heard much about is the start-up effort required and the barriers to entry for non-experts. The new ARM Cortex-M processors look to be a great move toward addressing the low-cost and low-power end of the microcontroller market, but they don't really address the buildability issue and the category-entry issue.

At Screaming Circuits, we run into quite a few designs in industries that are just now beginning to automate. In many of these cases, mechanical engineers, not software or electrical engineers, are tasked with putting the brains into the product. These mech folks have to learn, design, layout, build and code. The PIC and Atmel processors, with their thru-hole or big SMT packages, easy 5V power, low clock-speeds and huge base of community support make an impossible job possible for the new entrants into the embedded world. If a thru-hole part with a 20MHz clock can do the job, novice designers can greatly increase their chances for a successful design than if they have to deal with fine pitch parts and 100MHZ clocks.

In a perfect world, this wouldn't be a concern, but as it is, a lot of companies need these parts that are easy to implement for a new designer. M0's may be priced in the sub-$1.00 range, but piece price is not the only component of "cost".

Duane Benson
"Apple II forever"

DB25 With Issues

DB25 with issues

Take a look at this footprint. We're looking at the solder-side of a DB-25 connector footprint. Given all of the smt and HDI (High Density Interconnect) we see these days, this looks pretty primitive; like it could be straight out of 1979 or something. But it is a recent PCB design. Some of those good old-standard parts still have a place in today's world.

However, old or new, issues can still pop up. So, just what do you see wrong with this one? There are at least two pretty obvious, and a third issue that may be a little open to debate.

Duane Benson
What's wrong with this picture?

Holiday Shutdown, April 2nd

Screaming Circuits will be shut-down on Friday, April 2nd for the Good Friday holiday. That day will not be counted toward turn times and we will not be shipping on that day.

This means, for example, that if you place a 24 hour turn assembly order and we receive your files and parts by Thursday morning, the 1st, your job will ship out on Monday, the 5th.

Who's Resonsible For The Footprint?

I recently wrote a bit about CAD library parts for QFNs and a reader posed an interesting question in response to it:

"Okay, that makes sense.. But why don't manufacturers of the part put out their footprints and schematic symbols for their parts in some common format?"

QFN Freescale eagle copper layer That is a very good question. When you purchase or download a CAD package, it will typically come equipped with an established library. Those libraries vary greatly in coverage and quality though, and, of course, they only cover the parts available at the time of release - and only a subset of the most popular, at that.

Who's job is it to make workable library components and or new ones? Ultimately, it usually falls to someone in the organization doing the circuit design. Some people will pay a third party to make the parts. NXP has made the complete library for their chips for the PCB123 CAD package. Some independent companies make it their business to create and maintain libraries. But, really - who should be responsible?

A components manufacturer needs to document a new chip anyway, but they'd have to make a library part for a dozen or so CAD packages for each part variation, depending on how much coverage they want. The CAD companies would need to create maybe a hundred thousand or a million library parts, depending on how much coverage they want, but the CAD package is useless without the library. The designer would need to create just the library components needed for the specific design, but that could easily double to design time. I guess that's why we have what we have - a combination of the above.

Still, the big question is: why isn't their a standard format for the libraries? That would make everyone's life easier. So, CAD folks; why no standard?

Duane Benson

Many Faces of the QFN

When I first started using a CAD package for circuit design, I couldn't understand why the software didn't come with all of the standard footprints and why that wouldn't be good enough. That level of ignorance was hopefully pretty short-lived. But back then, things were a bit simpler. Most of what I dealt with were in DIP packages - 0.1" lead to lead spacing and 0.3" or 0.6" width.

QFN copper layer QFN solder mask layer Even if the reality was never that simple, it's a lot more complex now. Take the simple QFN. Your CAD package probably has a decent variety of QFN footprints, but most of them probably look like these first two here.

On the left is the generic footprint with a wide open flag pad in the middle. On the right, the cross-hatched area is likely representative of what the solder paste layer looks like for that standard library part. Looks simple, but there's a whole lot more to the QFN (and DFN).

Take these next two images. The image on the left here shows what should be a pretty common standard solder paste layer for a QFN center pad. You want to keep the paste coverage down in the 50 - 74% range QFN solder paste stencil layer good QFN Freescale eagle copper layer to prevent the QFN from floating up during reflow.

Some QFNs, especially high-frequency and RF parts require a special copper pattern to ensure proper grounding and clean signal conditions. This one on the right shows the recommended copper pattern for a particular Freescale ZigBee radio chip.

Even if your CAD package seems to have the QFN covered, check the component data sheet for any special land pattern requirements and check the paste layer to make sure it has you covered to prevent float or voids.

Duane Benson
One face, two face
Red face, blue face

Screaming Circuits is going to ESC

ESC sm logo Screaming Circuits will be exhibiting at the Embedded Systems Conference in San Jose this April. The ESC exhibition runs April 27, 28 and 29 and Screaming Circuits will be in booth 827.

As usual, the show will be held at the McEnery Convention Center in downtown San Jose, California. More information can be found at the show website.

Last year, the mood was pretty subdued due to the economic uncertainties. I'm hoping for more optimism and excitement rolling through the show this year. And maybe even some strudel. Myself, I'll be keeping an eye out for new developments with the Beagleboard and mBed.

Duane Benson
Embedded in my head.

Mysteries of Engineering

I (and many, many of us, presumably) have been reading more about all of the Toyota woes and the to-date unanswerable questions. Still, so much of the material written about the issues seems to be coming from the untrained. Certainly, human behavior suggests that some of these problems could be the result of operator error. But, I'm not an expert in human behavior, so I can't really say. And, certainly, problems do crop up in complex machinery, like cars. I don't know if that supposition falls within my area of expertise, but a few decades of operating motor vehicles gives me some personal empirical data on that one.

The area that does bother me the most is probably those that speculate that since the problem hasn't been found, it doesn't exist. This is an area where I can claim some level of expertise as well as plenty of personal empirical data.

It is possible to spend uncountable hours testing various possible conditions and still never uncover the one scenario that will cause a systems failure in the hands of the general public. Many years ago, I worked for a company that designed, built and sold projectors. In that day, these were big things with short-life, very hot, incandescent lamps. We thought that we had done a very through job of testing under various conditions and had been selling the product for a little while when reports started filing in of bulbs exploding. It wasn't just a simple break. The bulbs were exploding with such force that the bulb area was filled with a fine grained, razor sharp glass dust. Nasty.

ExplosionDuring a weekend burn in session with a couple dozen projectors, including some returned from the field, the engineer monitoring the process thought he heard a gunshot and dove to the floor. It wasn't a gunshot, but it was the first clue in a long investigative process that did end up finding the problem. It seemed that if a bulb was too deeply seated in the socket by a couple of millimeters, the reflection of the filament in the mirror would exactly line up with the actual filament, causing it to melt and arc. The arc would run in one direction, down the filament leg to the base and stop.

One filament leg had a few coils of small diameter tungsten wire wrapped around it. The other leg did not. Depending on the orientation of the supposedly non-polar bulb, the arc would either run down the leg with no coil or the leg with the coil.

If the arc ran down the leg without the coil, nothing happened other then the bulb needed to be replaced. If it ran down the leg with the coil, that small amount of additional vaporized tungsten increased the internal pressure sufficiently to explode the quartz bulb in a very catastrophic manner. Okay, now that's weird and obscure. Technically, you could call it operator error. If the customer had just inserted the replacement bulb the exact same way we inserted the bulbs during production, the problem would never have happened. But, realistically, it was a design flaw that set the customers up for a failure.

Duane Benson
Duck and cover

Thermal Mass Follow Up

RoHS has been with the electronics manufacturing world for quite a while now but there is still a lot of issues and uncertainty associated with it. As I wrote not long ago, even parts that are supposedly compliant can in some cases not cut it.

Taylor asked in the comments section of that post: "Have you noticed any pattern in capacitor manufacturersClose caps 3 exhibiting this problem? How can make sure to spec a capacitor that is more robust?"

Close caps 1 I can't say that I've seen a real consistent pattern with components from different manufacturers here. It's a case where the design engineer may have to compare the exact thermal specs from different components' data sheets and throw in a good measure of intuition and judgment as well.

In some cases, you might be able to replace a couple of capacitors with a single of a larger value, but in general, if you need multiples, combining them won't do. There are certainly good reasons to parallel up capacitors. You may need a few of different values to cover different frequencies. You may have a clearance issue and not have enough height for a taller cap. Or you may need to keep the ESR (Effective Series Resistance) down. Whatever the reason, if you need a number of caps close together, and they are big SMT electrolytics, you could be setting yourself up for this problem.

Close caps 2 Image A illustrates the issue found in that earlier post. The thermal mass of all of those big metal can caps can slow the solder melt. The most vulnerable pads are the two inside pads for C3 and C4. Keep the heat up long enough to fully melt the solder on all pads and you may destroy the caps, or other components.

You could just spread the two rows apart a bit like in illustration B. This might be enough to allow all pads to solder well or, if nothing else, it would give you enough room to touch up with a soldering iron.

Probably the most common solution though is to take the approach used in illustration C. Just put all the caps in a row so none of the pads are vulnerable.

If you need a compact layout like A, you'll just need to spend some extra time with datasheets to find a specific cap with a bit of extra RoHS temperature margin. Look at the maximum solder temp, the maximum dwell time and the profile curve if available. Don't forget to check your other components too to make sure that the extra reflow time wont harm them either.

Duane Benson

Beagleboard Innovation

Bb-part Open source hardware makes me happy. Open source has been around in the software world for a long time, but it's still fairly new in the hardware dimension. I think 2009 might just be the year that OSHW reached critical mass. Certainly, the Arduino is now everywhere, but there are other great opens source hardware projects getting some coverage too. My favorite is still the Beagleboard - a super powerful ARM Cortext-A8 based computer in an open source 3" x 3" form factor.

Gerald Coley, of Texas Instruments wrote a great article in EDN Magazine last year. His work is a finalist for the best contributed article category in EDNs Innovation Awards. Check it out. It's a great read.

Duane Benson
Curse you, Red Baron!

« February 2010 | Main | April 2010 »