Explaining solder joint defects, one interruption at a time.
“Thank you all for promptly joining this morning’s webmail conference call. Today we are going to review the x-ray images our outside lab FTP’d us yesterday. Please set your phones on mute by pressing *6 if you are not speaking, and set your computers to the link provided by the meeting invite. You can unmute your phones in the same way, by pressing *6, if you wish to contribute to the conversation. You may listen via phone link, or by plugging earbuds into your laptops and activating sound. If you feel compelled to speak, unmute your phone and speak one at a time. We’d like to give equal time to as many members of our several participating engineering groups as possible. We have a big agenda, and many participants, so let’s get started.”
Here we go. Amateur hour.
I’m the x-ray lab representative (and president). Lucky me.
“First, a roll call of participants. Please respond by saying ‘here’ when I call your site name.”
“… BEEP! Denver here. Good morning.”
“Good morning, Denver. As I was starting to say, Palmdale.”
“… BEEP! Good morning, colleagues. Boulder here.”
“Nice to hear from you, Boulder. It would have been nicer to hear from you at the top of the hour, as our Outlook invitation instruct ...”
“… BEEP! Washington here. Time to get started.”
In a federal tone of voice. Rome has spoken.
Six seconds of silence ensued. Enough time to reload.
“Colleagues, thank you for joining us this morning. Our x-ray lab has graciously made themselves available to answer any questions you may have regarding the 362-page PowerPoint report, with slides, they submitted last week, concerning our little matter of deep concern. As you are well aware, we have experienced intermittent failures in this life-support system. Troubleshooting has narrowed the failures to two printed circuit boards. We suspect solder joint defects, and we’re hoping our friends at the x-ray lab confirmed our suspicions with their diligent work. Much depends on what they’ve discovered. I don’t need to remind you that intermittent failures in a life-support system are a contradiction in terms. With that, I will turn the floor over to our audiovisual expert, who will put our x-ray lab’s report on display.”
Blank screen. For six minutes. Then an error message: A/V link failed.
“Don’t you guys in San Jose have an HDMI cable?”
(Muffled sound of mumbling. I think. At one point, a sharp, single-syllable retort, resembling an expletive.)
“BEEP. Phoenix here. Sorry we’re late.”
“Hello, Phoenix. Thank you for gracing us with your presence.”
We, the soon-to-be interrogated, endure by sitting in silence. No comment necessary. The silent imagination soars.
Sixty-nine bottles of beer on the wall …
“As I was saying, 25 minutes ago, let’s take a look at our lab partner’s presentation.”
“Uh, Wichita here. We’re having trouble logging into the video uplink. What was the password again?”
More silence, this time church-like. Funereal even.
“Got it now. Thanks!”
“Ladies and gentlemen, all technical difficulties having been successfully overcome, shall we finally begin?” intones our host.
“Consider slides 1 and 2, with a statement of the problem. Intermittent field failures of several units, with images serialized to those units. Slides 3 through 30 map the areas of interest, specifically the FPGA at U1 and the CPLD at U13. Both have been known to fail on certain units after several days, occasionally weeks, in service.
Always inconvenient situations, and always unacceptable. We suspect the problem lies with …”
The monologue is interrupted by a sound that can only be described as the repeated shuffling of a deck of cards.
“May I remind all participants to please mute their phones?!”
Thus shamed, the Shuffler mutes and silence resumes. Wonder if it was Texas Hold’em. No doubt more fun there than here.
Water torture resumes.
“Our suspicion lies with marginal soldering of BGA balls at either U1, U13, or both. Slides 3 through 30 show various views of the pin array for both devices on four suspect serial numbered units. Views are top-down 2-D, oblique 2-D (top view), oblique 2-D (bottom view), and 3-D (partial CT scan). I would ask our laboratory partners to explain to our assembled colleagues exactly what we are seeing with these images.”
“Well, the best way to describe our job is one of pattern recognition. With the FPGA at U1, you see on the screen a 2000-pin device, which we examine with ...”
“Is that the sound of a dog barking?”
The dog did not sound happy. Neither did our host.
“Mute your dog, er, phone, please.”
“Sorry. Dog slipped into the home office and burped his oats on the settee. A mess like you wouldn’t believe.”
Thanks for sharing.
“X-ray partners, please resume.”
“Yes, er, right. As I was saying, our job is pattern recognition. We carefully examine each pin of the device for uniformity of shape, placement, and reflow quality. Each pin is compared with its immediate neighbor. Deviations from the norm draw a heightened level of scrutiny. We begin in the conventional manner, with top-down 2-D x-ray analysis. Unfortunately, this method usually fails to reveal open pins, head-in-pillow situations, and the like. We usually move on to employ oblique 2-D analysis, which many confuse with 3-D analysis. Oblique 2-D allows us to look at a BGA pin at up to 70° from vertical, and orbit around the pin of interest. That circumnavigation often yields the culprit. If for some reason it doesn’t, we proceed one step further to …”
“… Tell the customer to go to hell.”
“… And tell them the order is nonrefundable, ECO or no ECO. If I’ve told you once, I’ve told you a thousand times. If they make a change midway through an order, they’ll need to pay for it. Either cancel the existing order, with appropriate cancellation charges, or cut in an ECO, with appropriate ECO charges. No freebies! If you can’t deliver the news straight, then most assuredly I will.”
“But it’s our best customer!”
“I don’t care if it’s Queen Elizabeth II and the whole damned House of Windsor on the purchase order. No fr- ... wait a minute. Is this thing on?”
Now it’s really silent. Except for that lonely train whistle off in the distance. And a dog barking.
“Well then, as I was saying, if oblique 2-D x-ray inspection proves inconclusive, yet we continue to have strong electrical reason to believe the defect lies in this device, on that row, in this cluster of pins, we can up our game and do a CT scan of the suspect pins, slicing them into as many as 6400 slices, should that become necessary. Each slice is a fraction of a 360° revolution around that particular BGA ball or balls, thus yielding information from 0.06° for each discrete slice. Software then recompiles those 6400 discrete slices into a three-dimensional whole, which can then be manipulated and examined from all angles, on three axes, in order to find the defect. We can even use that same software to drive clean through the BGA ball from front to back or from ...”
The sound of elevator hold music comes wafting over the phone. Henry Mancini and the Orchestra playing “Love Story,” with an idyllic string section to rival the best Nat King Cole recordings. Captivating.
This time our host’s voice has a decided edge to it. He is decidedly not captivated.
“May I remind one and all for the last time to mute their phones if they do not wish to speak.”
I thought love meant never having to say you’re sorry. Another youthful icon crashes and burns.
“X-ray lab people, please resume.”
I was waiting for the “dammit” at the end. It was implied, not expressed.
“Uh, yes, we can use modeling and rendering software to acquire a highly detailed image of the object of interest, looking at it from all available angles, until we either find the suspect defect or we’ve exhausted our search and move elsewhere. If we find a suspect defect, i.e., a violation of IPC-A-610, we take multiple images, usually from several angles, and catalogue them in our report, as you see in some of the later pages of the presentation before you on your screens. Any questions?”
“Yes, can you tell me what I’m seeing on slide 54, serial number 666?”
“It’s an open pin at location U13-A4.”
“Thank you. Can you tell me which clause of IPC-A- did you say 600, or 610, that it violates?”
Easy. Show decorum.
“Yes, I can.”
“Would you mind annotating your report by citing the pertinent clauses of IPC-A-610 that have been violated in each image?”
Actually, I would mind: 362 pages of findings. But I will. I charge by the hour.
“No problem. Give me a week. To whom should I address this when finished?”
“Thomas F. Joad, Ph.D. I will send you my contact information.”
“Your title, sir?”
“Senior supplier quality engineer.”
email@example.com. His column runs bimonthly.is president of Datest Corp. (datest.com);