3
\$\begingroup\$

When I was little, the general advice for electronics hobbyists was to socket all their ICs. Indeed you would find the word must referring to the need for sockets. This was to prevent thermal damage to the chips. There were even sockets for transistors.

Then surface mount technology came along. It's harder (perhaps well nigh impossible) to hand solder some of the dust like surface mount components. Yet we automatically try to.

I realise that high frequency circuitry can suffer from parasitic impedances due to the socket. But I'm still thinking of the super simple hobbyist like me. I don't want to dwell on any particular chips, but consider the unremarkable LM358M op amp in SMT form. It has a humble bandwidth of only 1 MHz so impedance is probably not of concern. It runs on a proper 15V. Yet if used tonight, it would be soldered directly to a PCB. No one would flinch at a recent Elektor Magazine article telling them to do so.

Similarly I have a 6502 processor. Advice in the day would have been absolutely to socket it. Many people now directly solder on SMT Amtel controllers with no consideration of sockets. And hands up, who used a transistor socket last year?

What changed? Why were hobbyists advised to socket ICs before, but not now?

(I'm focusing on hobby level PCB fabrication, not bread boarding. I'm not really looking for a treatise on the benefits of SMT per se.)

\$\endgroup\$

closed as primarily opinion-based by Dave Tweed Jan 7 '17 at 13:30

Many good questions generate some degree of opinion based on expert experience, but answers to this question will tend to be almost entirely based on opinions, rather than facts, references, or specific expertise. If this question can be reworded to fit the rules in the help center, please edit the question.

  • 1
    \$\begingroup\$ I would say there is no technical reason to, only practicality and its a sign of the times perhaps. ICs are so cheap these days. \$\endgroup\$ – crowie Jan 5 '17 at 0:55
  • 1
    \$\begingroup\$ Hand soldering 0603 sized components is pretty easy. 0402 is a bit tricky but still doable. \$\endgroup\$ – whatsisname Jan 5 '17 at 1:20
  • 4
    \$\begingroup\$ @AliChen -- industrial gear has this nasty habit of outliving its manufacturer board level spares support, and you better take that into account in your design lest you wind up with pissed customers down the road because your no-serviceable-parts-inside box took their production line out for a week. Besides, there's really not a whole lot of reason to superminiaturize industrial gear that's going into control panels and 19" racks, anyhow. ;) \$\endgroup\$ – ThreePhaseEel Jan 5 '17 at 2:44
  • 1
    \$\begingroup\$ Somebody help ! \$\endgroup\$ – Paul Uszak Jan 5 '17 at 22:08
  • 1
    \$\begingroup\$ I do a bit of work with GE Challenger alarm panels and even the recent models are chock full of through hole and socketed dip chips. Contrary to what Ali may or may not have read, many industrial electronics systems have design lifetimes exceeding a decade so repairability is key. When a whole new panel can cost you $2000, what idiot would design it so you couldn't just swap out the $5 micro? Only consumer gear is intended as "throw away", that's how they make money. Besides, good luck to any budding new hobbyist trying to breadboard in 5 mins with a QFN... DIP is quick 'n' dirty but easy as. \$\endgroup\$ – Sam Jan 5 '17 at 23:50
7
\$\begingroup\$

Paul I'm going to go out on a limb here and try to answer this.

In my experience, the 80's era and prior did use sockets quite often for everything from IC's to transistors. There may have been several good reasons for this:

  • If the part failed, a socket simplified repair.
  • Most IC's were DIP package (relatively low pin-count) so could be socketed easily.
  • Older designs were engineered for serviceability, not lowest cost.
  • If you botched one pin, replacing a socket was inexpensive compared to the IC.
  • Newer versions of the IC could easily be "upgraded."
  • Early-production parts may have been slightly less reliable (opinion.)
  • Yes, soldering the socket prevents thermal damage to the component.
  • Soldering irons of that era were larger - 35W or more - hence a socket.
  • The IC could be removed and swapped out to test (or put in your other gadget.)
  • Full-fledged circuit simulators of the time were not very robust, if available at all. If any design calculation error crept into the PCB, it meant troubleshooting hardware.

At the onset of the electronic revolution the costs were high so focus was more on quality workmanship and serviceability. Nobody really knew how long a fancy new transistor would work in service - so they often built them way over-spec. Transistors rated for 10A could routinely tolerate 15A and survive.

Today the situation is exactly the opposite - costs are low, production is lean, and all the corners that can be cut, are. We have great data now on exactly how well devices work in the field, and have trimmed much of the "headroom" off component specification, both figuratively and literally. If you try to pass 15A through a 10A device today, it'll surely be damaged. Not only that, but we have a good idea where the technology is going today, and can make good predictions on how long a device will remain in service. For those components which might be prone to failure, many DIP and other larger packages are still available, should the engineer choose to use them.

Today we'd use no socket/SMT for the following reasons:

  • No need to drill the PCB.
  • No need to plate any through-hole connection (optional at home.)
  • "Jellybean" parts are incredibly inexpensive today - if you damage one, scrap it.
  • Modern IC's can have hundreds of pins and dozens of form factors - making DIP obsolete and impractical for them all. (Let alone the increased speeds and all that entails.)
  • Modern IC's are much smaller. This may be perceived as a hinderance to the hobbyist, but is the whole reason our cellphones are even possible. Luckily, many devices still exist in DIP packages. For those that do not, expander boards are often available.
  • New designs are less-engineered for serviceability, and more as "throw-away" devices. (Good point @dim, added.)
  • Modern simulators can offload much of the "did I get all my maths right?" into the click of a button - oops that resistor needs to be 10x larger, oops that op-amp exhibits saturation on it's negative rail - before even building a PCB. (Thanks @Ali Chen.)
  • Instead of 35W irons which can easily lift traces, temperature-controlled irons and/or hot-air is used, which makes soldering safer, faster, and easier.
  • SMT lends itself better to mass-production. A socket is just another step someone has to take in handling, which costs money.

Less work, faster results - the mantra of the new millennium. I believe that's why we've moved away from most through-hole and socketed designs today.

I used to make my own PCB's; it was fun for small projects. Perchlorate, ferric chloride, developer, toner, even cupric chloride - tried them all. Satisfying? I guess so, but today I'd much rather send my design files along with less money than I would have spent to do it myself, and still end up with a better product. Efficiency has become paramount.

While components are getting smaller and smaller, the PCB details must match. Sure, one could get pretty good results with the laser-toner transfer method at home, but it would be risky for something with a very fine pitch such as a VQFN. (Yes, even those can be soldered by a hobbyist - only need a hot-air gun, solder paste and some flux. (And likely some kind of magnifying lens.) There are many videos out there on the subject, some even using traditional irons. (Try a hot-air station; you'll love it.)

Edited to reflect changes to question.

\$\endgroup\$
  • 3
    \$\begingroup\$ One more factor: use of SPICE simulators and IC models before even plugging a soldering iron. It makes making PCBs with SMT circuits with higher chance of success. \$\endgroup\$ – Ale..chenski Jan 5 '17 at 1:20
  • 2
    \$\begingroup\$ And planned obsolescence, which is the norm now, makes the "easy to repair" factor unnecessary, and even unwanted. \$\endgroup\$ – dim Jan 5 '17 at 8:26
  • \$\begingroup\$ Hmm. Pick & Place is not really on topic vis a vis my question. So reading between the lines what's changed is: EDA software PCB fab houses, hot air soldering guns and reflow ovens? \$\endgroup\$ – Paul Uszak Jan 5 '17 at 12:45
  • 2
    \$\begingroup\$ Well such a poll would be inconclusive at EE.SE because there is such a great divide between "hobbyist", "veteran", and "maker" today, both financially and intellectually. The hobbyist may have been around to see sockets, but are just "not interested enough to keep up with the times." The seasoned vet scoffs at this discussion and goes back to measuring eye-diagrams on equipment more expensive than our salary. And the maker cares little about any of this and just wants us to get their light blinking for them. "What? You mean I can't plug this VQFN into my arduino?" :) \$\endgroup\$ – rdtsc Jan 5 '17 at 22:35
  • 1
    \$\begingroup\$ @rdtsc, precisely captured! BTW, I do have the el-cheapo 858D hot air handle, works just fine. The 862D station ($45) looks even better for a hobbyst, having a temp-controlled iron as a bonus. And yes, I started 50+ years ago with transistor sockets, now ending up with soldering 0402 caps and 0.5mm pitch VQFNs under 30X stereo microscope. Unfortunately, the $500 kbucks equipment (~3x early salary!) is a luxury of the past, and it is really difficult to make something up-to-date in retirement. \$\endgroup\$ – Ale..chenski Jan 5 '17 at 23:26
2
\$\begingroup\$

I haven't seen the days when transistors used to be socketed, but as a young hobbyist I did socket my ICs. What changed? Well for one, ICs were expensive. If you soldered it onto the board and something else got botched, you'd have a hell of a time removing the IC. (Ever tried desoldering a 32pin DIP?). The other is re-use - I once managed to get (free sample) a digital voice recording chip which could record/playback an impressive (at the time) 30 seconds of voice. I used the single chip for multiple projects - just unplug from old board and plug into new. Even though SMD components cannot be socketed, many components can be connected to a breakout board which can then be socketed. Finally, the reason for socketing transistors is that if you blow one you can quickly replace it without having to use the soldering station. As a power electronics student that saved me a lot of time during lab assignments.

Since you asked who used a transistor socket: I did, but I used something like this: http://p.globalsources.com/IMAGES/PDT/BIG/921/B1052247921.jpg

\$\endgroup\$

Not the answer you're looking for? Browse other questions tagged or ask your own question.