Tedium - Socket To Me 🖥

The Intel Socket 1, used for 486 chips. The socket has a zero insertion force (ZIF) mechanism. (KJanos1396/Wikimedia Commons)

Five interesting facts about processor sockets

1. The placement of the pins is often very specific. To ensure that a chip doesn’t get placed into a socket the wrong way, or to keep incompatible processors out of the sockets, the pin grid array (PGA) is often set up in an almost—but not quite—symmetrical way. This can be seen in one of the earliest CPU sockets to use a pin grid array, the Intel Socket 1, which dates to the days of the 486. The socket type has three perfectly square rows of pins on either side, minus one extra pin in the bottom-left corner, ensuring alignment is perfect. PGA was spearheaded by Intel but is still used heavily by AMD today.

2. You often don’t force chips into place. You know how, when you put an NES cartridge into the console, you have to press down to fully insert it—but not too hard, lest you bend the pins inside? (But you bend them anyway, because you were too aggressive?) Many replaceable chip adapters work like that, using something called a “zero insertion force” socket to hold things in place. With a ZIF chip socket, you basically get it lined up, and in many cases, use something to clamp it down into place—say, like a metal bar. The approach came into use with the goal of making it easier for end users to change out processors without a lot of hassle.

A Cyrix QFP chip. (Geoff Martin/Wikimedia Commons)

3. Some processors have pins that stick out to the side. Over the years, there have been lots of different kinds of chip layouts, some of which are more efficient in terms of the number of connections than others. One technique widely used in the early ’90s on many chips is the quad flat package (QFP), which attached to a piece of silicon using dozens or even hundreds of little pins that jutted out to either side. This technology, which can be soldered to a board by hand, eventually fell out of favor for top-tier CPUs, in part because newer technologies used space more efficiently, but you can still find examples around, such as this very inexpensive ARM chip.

An Intel LGA 775 socket, which features pins on the motherboard itself, rather than on the chips. (stnc06/Flickr)

4. Major chip companies have been known to change up their techniques. When people think of processors, ten to one, they might be thinking about the pin grid used on many of Intel’s early chips, but it’s far from the only technique in use. For example, the Land Grid Array (LGA), used on many modern Intel chips and even some recent AMD chips, basically flips the model of the PGA chips, putting the contacts on the processor itself, but the pins on the motherboard. This allows for higher pin densities than PGAs, and when needed, a way to hold the chips into place using solder paste—a technique commonly used on laptops, where space is a luxury. Speaking of …

(conifer/Flickr)

5. Many smartphone chips are held into place with tiny balls of solder. There obviously isn’t a ton of room for a processor on a socket in a smartphone or an ultra-thin laptop, so techniques for putting processors into place have evolved. One of these techniques is something called ball grid array, which involves the exact placement of balls of solder in a layout not dissimilar to that used in a pin grid array or a land grid array. Why solder? For one thing, it takes up less space than a mechanism that allows a device to be easily removed; for another, it’s more efficient, as the entire chip can be held down by balls of solder, an approach that allows for the shrinking of components, such as solid-state drives, to ever-smaller sizes. The downside for consumers, of course, is that it makes devices far less upgradeable, because it requires a level of precision that a machine can easily handle most humans can’t, and it’s way more complicated to desolder a chip than it is to solder one in, to the point where you’re better off buying a new phone than trying yourself. And, of course, sometimes things come loose, leading to things like the iPhone 6’s infamous “touch disease.”

It’s weird to think that, a quarter-century ago, our CPUs looked like video game cartridges, a move that weirdly feels like a bad memory today, in an era when computers take up less and less room, where many highly-integrated modern computers would never allow a CPU to take up anywhere near that much space.

But what’s great about those slotted chips is that, for all their faults, they could be removed and replaced by an end user.

Our move towards highly integrated, ever-smaller devices is, of course, a double-edged sword. On the one hand, moving to technologies that solder chips directly onto the board has been a real kick in the pants for speed and size. On the other hand, as chip sizes shrink as new techniques get used, the upgrade paths that we once had, both on mobile and in some desktop forms, become less clear.

The rise of the mini PC, as highlighted by the Intel NUC and the Mac Mini, has made it so that people are much more likely to replace an entire machine rather than go through the trouble of upgrading a processor, as an example.

I mean, that’s what people already do with smartphones.

This high level of integration means that our computers need to do the job a little bit longer, because once we buy ’em, we’re stuck with them a few years, and once you buy it, it’s what you own. Not even the RAM chips are easy to replace in many models. It’s a dead end, one that left a formerly buzzy idea—upgradability—on the cutting room floor.

More than a decade ago, YouTuber Dave Hakkens came up with a clever concept that effectively brought the modularity concept of desktop computers to the world of smartphones. PhoneBloks, as he called it, was a hugely inspiring kick in the pants for tech at the time it was announced, but after some missteps in the form of Google’s shuttered Project Ara, the idea had lost buzz and momentum as it seemed that the world had decided to go another direction entirely. In a retrospective from about four years ago, Hakkens remained optimistic, but the idea (when it has emerged) has largely leaned on the idea of repairability, not upgradeability.

Still, it’s hard not to watch the clip and feel an immense sense of disappointment that this big idea couldn’t keep up with the demands of ever-shrinking components. Hakkens’ instinct is right, but the hard part is that there’s no money, or at least not as much money, to be found in upgradeable components, e-waste be damned.

All that said, one company in the laptop market is helping to give postmarket upgraders some hope. That company, Framework, has released strategies to replace RAM, storage, batteries, even mainboards. The company uses ball-grid array chips, so the CPU won’t leave the board, but you can take it out, replace it with a new one, and even reuse the old one.

The processor socket, for all its complexities and technical quirks, represented a step in the favor of the consumer, even if just those technically minded enough to know how to upgrade a CPU on a motherboard or to know the difference between a pin grid array and a ball grid array.

Now, if we could only shrink our upgradability as fast as our processors.

--

Find this one an fascinating read? Share it with a pal!

Also, be sure to give our quick-hit newsletter Lesser Tedium a look—and get a two-minute look at one of the many elements in Tedium’s deep archives.

Is now the time to upgrade your news diet? Our sponsor 1440 can help.