Statistically Significant vs. Practically Significant in SMT Data Collection

Folks,

Let’s assume your company has decided that transfer efficiency (TE) is the key metric in determining solder paste quality. Transfer efficiency is the ratio of the volume of the solder paste deposit divided by the volume of the stencil aperture. While you agree that TE is an important metric, you are a little troubled with the recent results in a solder paste evaluation. Two out of 10 pastes are fighting for the top spot and it looks like TE will be the deciding metric. Paste A had a TE of 99.5% and Paste B had a TE of 99%. So management wants to go with paste A. You are troubled because paste A has a poor response-to-pause. If it is left on the stencil for 15 minutes or more the first print must be discarded. This weakness may result in 30 minutes or so of lost production time in a 3-shift operation.

However, the TE test results showed that the TE of paste A was statistically significantly better than paste B. You think about this situation and something doesn’t make sense — 5% and 99% are quite close.

You dust off your statistics textbook and review hypothesis testing. Then it hits you, with very large sample sizes, means that are closer and closer together can be statistically significantly different.

The data show that paste A has a mean of 99.5% and a standard deviation of 10%, whereas paste B has a mean of 99% and also a standard deviation of 10%. The sample sizes were 10,000 samples each. These large sample sizes are important in the analysis. The standard error of the mean (SEM) is used to compare means in a hypothesis test. SEM is defined as the standard deviation (s) divided by the square root of the sample size (n):

So as the sample size increases, the SEM becomes smaller or in statistics lingo “tighter.” With very large sample sizes, this tightness enables the ability to distinguish statistically between means that are closer and closer together. This situation was not a concern with sample sizes of less than 100, however with the modern solder paste volume scanning systems of today, sample sizes greater than 1000 are common.

Figure 1 shows the expected sampling distribution of the mean for samples with a TE of 99.5% and 99.0% and a sample size of 100, both have a standard deviation of 10%. Note that to your eye you do not see much difference. However, with the means and standard deviations the same and sample sizes of 10,000 the sampling distributions of the mean are clearly different in Figure 2.

The reality though, is that there is no difference in the results in Figure 1 and 2. The tiny difference in the means (0.5%) may be statistically significant with a sample size of 10,000, but is it practically significant? Would this small difference really matter in a production environment? Almost certainly not.

Figure 1. Sampling distribution of the mean for a sample size of 100.
Figure 2 Sampling distribution of the mean for a sample size of 10,000.

So, with large sample sizes, we need to ask ourselves if the difference is practical. For TE, I think we can be confident that a difference of 0.5% is not practically significant. But, what if the difference was 2% or 5%? Clearly, experiments should be performed to determine at what level a difference is significant.

With the case discussed above, I would much prefer the paste that has a 99.0% TE and a good response-to-pause.

Cheers,

Dr. Ron

ODMs-OEM Gap Narrows Further

OEMs have long been told that, as their suppliers, EMS companies would not compete with them.

Then EMS companies added design services. “We won’t compete with our customers.”

Then EMS companies added full turnkey services. “We won’t compete with our customers.”

Then EMS companies added logistics services. “We won’t compete with our customers.”

Then EMS companies started calling themselves ODMs.
We won’t compete with our customers.”

I think all pretenses are over. Check out this latest report from DigiTimes from the Embedded World trade show, where Pegatron exhibited an all-in-one dashboard for smart driving vehicles.

Still not competing? Think again.

Is the Supply Chain Holding On, or Holding Out?

Foxconn, for once, was probably the first company in the electronics manufacturing sector to acknowledge the looming financial hit from the coronavirus. Apple, which relies on the Taiwanese ODM for the majority (?) of its production, was naturally forced to follow.

Speak no evil?

Jabil and Plexus have now lowered near-term estimates, and market research firms are piling on, with IDC downgrading its outlook for smartphones and PCs and DigiTimes Research slashing its notebook shipment forecast by a third or more.

In a timely column on CIRCUITS ASSEMBLY, EMS expert Sue Mucha lays out a strategy for handling sharing bad news with suppliers and customers. “Transparency matters,” she says. “The goal shouldn’t be to paint a rosier picture than the situation dictatesThe goal is to fill the communications void and establish trust that your company will provide news as the situation evolves.”

That begs the question, why haven’t more firms come forth with sales or profit warnings? Are Apple, Foxconn, Jabil and Plexus the only ones that will be affected? Or are they simply the vanguard?

Coronavirus: Curse, or Blessing in Disguise?

I’m not one to make light of health epidemics, especially given that my college major was initially epidemiology. But the global slowdown in electronics demand — underscored by the earnings reports over the past couple months — is poised to worsen in the grip of the coronavirus outbreak, which started in China and has now reached more than a dozen more countries.

In response, businesses in Wuhan, the epicenter for the disease, have closed and Chinese government has effectively quarantined the entire city of 11 million. Wuhan is the capital of Hubei province, and between the two entities there are roughly 1,500 factories and related facilities.

Among the electronics manufacturers affected at ground zero are Avnet, Siemens, and Foxconn. After Hubei issued a mandatory shutdown notice, Foxconn furloughed more than 13,000 workers at its campus there. (The plant generates an estimated $300 million in revenue per year.)

As the disease spreads, so too have the shutdowns. IMI announced last night it is suspending operations at plants in Shenzhen, Jiaxing, Chengdu and Suzhou. The furloughs will last up to several days.

While electronics assembly plants can be brought back up to speed fairly quickly, the impact missing even a few days of revenues will be felt in the quarterly results. And here’s the possible silver lining.

Many companies have been reporting weaker results in the most recent quarter. For instance, Celestica’s revenues were down 14%, Amphenol’s dropped 3.3%, and Sanmina’s were off 16%. TTM went so far as to sell its entire mobility unit. Plexus saw sales rise, but is cutting an entire design unit. (UPDATE: Flex reported sales down 7%, adding to our picture of how widespread the weakness is.) Often the US-based firms see a slowdown in presidential election years as OEMs take conservative approaches to ordering ahead of potential administration changes. The outbreak, deadly and unwished for as it is, gives cover to management for any recurring revenue drops, at least for a quarter or two, and perhaps longer if the coronavirus gains a wider foothold. A cynical view, to be sure, but hardly an unrealistic one.

We need only look back to 2003, when the SARS outbreak came on the heels of a worldwide business slowdown and according to one analyst “accelerated that downturn and spread it to many other countries in Asia.”

Already, airlines are bracing for lower capacity utilization and Asian firms are fearing the worst as businesses enact travel restrictions. Again as with SARS, the timing comes as Asia (and most of the world’s major economies) is experiencing tepid growth, and the best way to stop a recovery in its tracks is to sever the flow of goods and services.

For everyone’s sake, let’s hope this virus burns out fast.

For Tom Hicks, Like Father Like Son?

Those who have been in the PCB industry since at least 2001 will likely remember the New York buyout firm known as Hicks, Muse, Tate & Furst. Better known as Hicks, Muse, they laid waste to the industry, buying up the largest fabrication shops like AT&T’s 400,000 sq. ft. plant in Richmond for dollars on the penny (you read that right), then attempting the same with EMS, before watching it all fall to pieces with multiple bankruptcies and revaluations. (The entity known as Viasystems eventually landed under the control of Tim Conlon, a highly capable executive who right-sized the company and made it profitable enough to sell to TTM in 2015.)

One of the beneficiaries of the merger mania in the late 1990s and early 2000s was Bob Herring.

Not once but twice, Herring built up and sold printed circuit board fabricators to larger bidders. The latter time, Herring sold Herco Technology and a second firm, Synthane Taylor, to Teradyne for about $122 million in stock.

You know how this ends. Teradyne’s stock collapsed, Herco was shuttered, and then came the lawsuits.

Herring later turned his sights to media, launching a news network that is popular in certain segments of the American public.

Tom Hicks of Hicks, Muse turned his sights on sports, buying — and bankrupting — the Dallas Stars hockey team and Texas Rangers baseball team, among others.

Tom Hicks’s son is now in the game. Under the eponymous billing of Hicks Equity Partners, Tom Hicks Jr. is apparently attempting to corral fellow money-men to pony up about $250 million to a certain news network.

And who owns One America? Bob Herring.

As the safe harbor warning every equities firm is obligated to announce says, past performance does not guarantee future results. But if I were one of Junior’s friends, I’d keep wallet in my pocket.

AR/VR: What’s New?

CES 2020 offers the latest augmented and virtual reality technology, creating immersive experiences across industries from health care to agriculture and manufacturing to retail.

The latest in AR/VR technology from CES 2020.

Dispelling The ‘Five Ball Rule’

Michel writes:

Dr. Ron, when if comes to SMT printing of solder paste, why do some people use the five-ball rule for rectangular apertures and the eight-ball rule for circular apertures?

Michel:

The “Five Ball Rule” is another metric that SMT assembly industry leaders believe, but it is difficult to find its origin. It states that when selecting a solder paste, five of the largest solder balls should be able to fit across the width of the smallest rectangular stencil aperture. See Figure 1a for a 0.2mm wide rectangular aperture.

Typically, the largest solder ball diameter is assumed at the 90th percentile. See Figure 2. So, in this example, a type 4 solder paste would fit the five ball rule as the largest solder ball is 0.038mm. Five times 0.038 is 0.190mm, just a little less than the aperture width of 0.2mm. It should be remembered that this is a “rule.” not a “law.” So let’s say you had 4.5 balls across the aperture with instead of 5, it would most likely be OK. 

Figure 1. A comparison of the Five and Eight Ball Rules

Figure 2. Solder Powder Sizes

A generation ago, the advent of circular apertures to support BGA and CSP packages necessitated a new “rule.” Figure 1b shows why the five-ball rule is inadequate for circular apertures. Although five type 3 solder balls fit along the 0.275 diameter, off the diameter, there is not enough room for many solder balls.  Hence, an insufficient amount of solder paste would be printed.

For the same aperture, if a type 4 paste is used, 7 or 8 solder balls span the diameter and the amount of paste printed would be much closer to the volume of the aperture.

For a little more on this topic, see a past post.

Cheers,

Dr. Ron

At AltiumLive, Lots of Sun (and One Big Cloud)

by Brooke Anglin

One of the first things I heard when I arrive at AltiumLive last week was was how much they loved the recap of the event my colleague Chelsey Drysdale wrote last year. They added that Chelsey truly captured what AltiumLive is about. Talk about pressure!

Let’s start at the beginning.

The Piano Arc contains a total of 12 PCBs.

Drinks are on Them. The week (well, three days, but they really pack a lot in!) kicked off with a great reception on an outside terrace at the Loews Coronado Bay Resort.  Dave Starkey, the engineer who designed and built the Piano Arc — the circular piano used by Lady Gaga’s pianist, among others — was there. So was the pianist. They displayed the unit, a 360-degree custom keyboard, with one of the panels removed, exposing the PCBs.  Altium director of community engagement Judy Warner met Dave at a user group meeting in San Diego, heard his story and recruited him for the event.  In case anyone wants to know the piano has 288 keys, a total of 12 circuit boards and only cost $57,000.

I saw our customary PCB West speakers Suzy Webb, Rick Hartley, Gary Ferrari and Eric Bogatin and met a lot of engineers.

Getting Younger. One of the things you note about AltiumLive is the demographics. It’s a younger crowd than you typically see at industry events. Eric Bogatin noted this in the first keynote (AltiumLive had not one, not two but three keynotes). He brought some of his students, had them stand up for recognition and encouraged hiring companies to talk to them.  That drew a good laugh from the crowd.  In his talk, he pointed out that while many designers will be retiring in the next 10 years, AltiumLive gives us reason to think a new generation is coming on strong.

Keynote #2 featured a celebrity from outside our industry. Jor Grand of Grand Idea Studio has a short-lived TV show called Prototype This! on the Discovery channel. The concept, which was similar to Mythbusters, was about the viability of gadgets and technology and how engineers feel a since of winning when a product doesn’t work the first time, but they don’t give up until their creation finally works successfully.  While the show lasted only one season, Grand shared several clips and generated lots of laughs.

Finally, the main event: Altium rolled out its much-anticipated Altium Designer 20 ECAD. VP of marketing Lawrence Romine detailed the new “push and shove” capabilities that enable routing of complex HDI boards and ramp up design times over 20%. (For more on the tool, click here.)

Cloud in their Eyes. This was the third year of AltiumLive, and the multi-city event (another is being held in Frankfurt this week) draws 500 designers. The company grow is equally impressive. Last year 2018 Altium saw over 8,000 new users for its Designer PCB CAD tool, and on average Altium users do over 10 designs a year.

Looking ahead, Altium wants to move toward the cloud, something the industry has been avoiding for years. Its has added real-time collaboration, and hopes to bridge the gap between manufacturers and designers.

Not surprisingly then, AltiumLive focused as much on tomorrow as on today. The highlight of the event was the Robot Build and Battle. Attendees split into teams of six or more. Ours had nine; eight engineers, from all of the country, and me. I was cheerleader, motivator and general handyperson. (They said I sounded like each of their bosses … I beg to differ!) Team Sierra beat last year’s record, but we weren’t far behind.  Amit Bahl of Sierra Circuits told me he didn’t realize just how geeky our industry was until the robot build. Such a fun and educational event!

Brooke Anglin is senior sales associate with PCD&F/CIRCUITS ASSEMBLY.

What a Waste

I like articles like this one from Forbes — even if they are out of date, because they cast attention on the intrinsic complexities of manufacturing.

But, while Forbes may think manufacturing inefficiencies cost the industry $8 trillion a year (no idea where that figure comes from, by the way), it points to the need for tools like the digital twin.