EN
Quick Links
Last Updated:08/19/2026 Author: Zeon
The X-ray picture told the whole story, yet no one intended to review it. A consumer's board-- a thick module developed around a 0.5 mm pitch BGA-- had actually come back from reflow withh a 9% first-pass failing rate. The stopping working joints weren't missing out on. They were thin. The X-ray showed solder joints that looked like they 'd gotten on a diet: partly filled up, deprived of solder, with a couple of that were clearly borderline opens up.
We ran the usual suspects. The stencil was determined-- fine. The solder paste volume was within specification-- penalty. The reflow profile matched the paste producer's referral-- penalty.
Then an engineer revolved the board and noticed something on the bottom side. Right under the BGA area, the vias were full of solder. Not a little bit. Chock-full, like tiny metal straws.
The style had placed vias straight in the pads-- via-in-pad, the format method that makes dense BGA directing feasible-- and no one had actually defined by means of plugging. During reflow, the molten solder did what molten solder always does when it locates an open hole: it streamed in. The joints ahead spent for it.
Here's the mechanism, removed to its basics. When solder paste melts in the reflow oven, it becomes liquid. Fluid solder moistens copper-- that's what makes soldering possible at all. An open through in the middle of a pad is a copper-lined tube directing straight down right into the board. The liquified solder does not require to be pushed. Capillary activity pulls it right into the through barrel, the same way water climbs up a paper towel.
The amount of solder a through can ingest is not unimportant. Think about a common setup: a 0.25 mm by means of in a 0.4 mm pitch BGA pad. A 0.25 mm hole with a 1.6 mm board has an interior volume of roughly 0.08 cubic millimeters. A regular stencil for that pad deposits somewhere in between 0.1 and 0.15 cubic millimeters of paste-- and paste is only around half steel by volume after the change burn. In other words, a solitary open through can soak up most of the solder that was meant to develop the joint. In some cases all of it.
That's the mathematics behind the issue. It's not a low result. It's not a "often you get a somewhat weak joint." An unplugged via in a BGA pad can take in the entire solder supply for that round, leaving a joint that's either open, approximately starved that it falls short within months under thermal biking.
Nobody draws an open using in a pad intentionally. The via-in-pad strategy exists since it's the only way to run away the pins of a fine-pitch BGA. Listed below about 0.5 mm pitch, there's no space in between pads for dog-bone routing-- the typical option of running a brief trace from the pad to an using positioned to the side. The pad itself has to carry the via, or the design can not course at all.
So the technique is legitimate, and significantly necessary. What journeys people up is the goal: an using in a pad is only acceptable if it's processed appropriately. The design data claims "through below." The construction order requires to say "fill and cap this via"-- and if it does not, the fab builds the via as an easy plated-through hole, exactly as attracted, and the board ships with numerous little solder-thirsty straws under the BGA.
I have actually seen the series play out three times in the last two years. A developer includes via-in-pad to break out a dense device. The fabulous notes discuss the stack-up, the product, the copper weight-- every little thing other than by means of treatment. The board home quotes and builds it without asking, since their system treats an unspecified by means of as an open through. The first setting up run discovers the issue. The blame video game starts. No one wins.
The failure modes from solder wicking aren't limited to slim joints. There are 4, and they pile:
Solder hunger and opens. The joint types with much less solder than it needs. In the worst cases, the BGA ball never ever moistens to the pad whatsoever, and you obtain an open that only appears at functional examination-- or even worse, in the field.
Gaps. Even when the joint looks appropriate, the solder that didn't move right into the via frequently leaves voids behind. Spaces in BGA joints elevate electric resistance and develop tension concentration points. Under thermal biking, they expand. An invalidated joint passes the X-ray assessment at the factory and falls short at month 14.
Outgassing. Vias that aren't connected can trap flux residue, wetness, and handling chemicals inside the barrel. During reflow, that contamination vaporizes and and presses bubbles up via the molten solder. In the most awful cases, it causes solder to spit out of the by means of onto bordering pads-- a flaw that can short 2 unconnected internet and generate a failing that's very tough to trace.
Solder balls. Solder that that arises from an unplugged through on the second side can develop loose rounds under of the board. Loose conductive debris inside an enclosure is a periodic brief waiting to take place.
Each of these failing settings is avoidable. Every one of them is unseen right now it's developed-- the damage takes place inside the via barrel throughout a 30-second window in the reflow stove.
Fortunately: this issue has a well-defined, standard option. The problem: it has to be specified, due to the fact that it sets you back cash and it won't take place on its own.
Via connecting is the minimum demand-- filling up the using barrel with a product so solder can not enter it. For via-in-pad applications, the common practice is via filling plus cap plating: the using is full of a conductive or non-conductive resin, the resin is cured and planed flat, and after that a copper cap is layered over the top. The result is a pad surface that's tantamount from a pad without via in all. The solder paste sits on solid copper, exactly as the stencil design meant.
The sector categorizes through security under IPC-4761, which specifies seven kinds of using security, from easy tenting approximately filled-and-capped constructions. For via-in-pad, the pertinent kinds are the loaded ones-- Kind IV (filled up and covered) and Kind VII (filled up and capped with added mask). Below's the difference that trips people up: tenting-- where the using is just covered by solder mask on both sides-- is NOT via-in-pad suitable. Tenting maintains solder out of the hole on the surface, but the mask can split under thermal anxiety, and it not does anything for outgassing or contamination caught in the barrel. Tenting is for vias between pads, not in them.
2 even more information that matter in production:
First, fill product choice. Conductive material filling (usually copper-filled epoxy) is chosen where the via lugs current-- it maintains the electrical course via the pad. Non-conductive resin is acceptable where the through only requires to be mechanically shut. The option impacts cost and electrical performance, and it should be made in the style testimonial, not discovered at the fab.
Second, the fill quality. A partially filled using is as poor as an open one-- if the material stops halfway down the barrel, there's still a space under the pad that can catch change and outgas during reflow. IPC advice requires the fill to be essentially void-free in the BGA region, and a proficient fab will certainly cross-section a coupon from every manufacturing panel to prove it. If your supplier can't reveal you that cross-section data, ask why.
If the transmitting enables it, the cleanest remedy is to stay clear of via-in-pad completely. Dog-bone routing-- a brief trace from the pad to an using put close to it-- is the classic option, and it removes the trouble at the source. The tradeoff is room: the dog-bone footprint is bigger, and below 0.4 mm pitch, it commonly does not fit. That's why via-in-pad exists in the first place.
There's also the option of via-in-pad just where needed. A board can course most of its webs with dog-bones and get via-in-pad for the rows of BGA pins that really can't get away differently. This hybrid method decreases the variety of vias that need filling-- which is where the construction price really sits-- without compromising routability.
The solution for this entire group of issue comes down to one line in the manufacture notes: fill and cap all vias in part pads per IPC-4761 Kind IV/VII, with void-free fill confirmed by cross-section.
That sentence sets you back practically nothing to write. It saves the rework, the scrap, the area failings, and the disagreement concerning whose fault it is.
Here's an examination you can run on your next board prior to it goes to the fab: open your design data, count the vias that touch or drop inside solder pads, and ask what's going to quit liquified solder from moving right into them. If the response is "nothing," you have actually found the issue before it cost you a manufacturing run. That's the most inexpensive repair in this sector-- and it's readily available to every person that takes thirty seconds to look.
Article written by Zeon
Hi, I'm Zeon — 20 years in PCB and electronics manufacturing. Front-end design and R&D, components sourcing, precision SMT, DIP through-hole assembly, and complete unit assembly. That's the full path from concept to finished product, and it's the path I've walked for two decades.
Hot News2026-09-17
2026-09-09
2026-09-07
2026-09-04
2026-09-03
2026-09-01
2026-08-31
2026-08-27
