Longetivety and long Term Reliability of Today´s Integrated Circuits

[sergeauckland]

Good luck fixing a broken bit on a core memory... Try running Windows 10 on a Sinclair ZX80...

Integration and non-replaceability are price you (don't) pay for small, cheap, low-power, reliable electronics.

Exponential growth in performance of digital electronics over the last 50 years means continual obsolescence. That silicon is sand; the only thing worth recycling is the precious metals, and that's what is recycled.

But what if I prefer to have larger, more expensive and possibly less reliable but repairable electronics? That's why I still use a 30 year old CD player and pre-amp, 36 year old turntables and loudspeakers. My amplifiers are more recent, but at least mostly discrete and repairable. I think there's far too much emphasis these days on price and fashion, much less on sustainability and keeping stuff until it wears out, not just until the next new version comes out.

S.

 

[captain paranoia]

But what if I prefer to have larger, more expensive and possibly less reliable but repairable electronics?

Then you're SooL, I'm afraid...

Most modern, consumer, digital technology would be impossible to implement in a more discrete form. It would be huge, and take a huge amount of power, even if the discrete digital logic circuits were available. Think of the environmental impact of that power.

Analogue electronics hasn't changed much, if you discount digital filtering/EQ, so you can easily buy an amp that should last you years. My main system is still the one I bought in 1988. You can still buy all-analogue amps.

DAC technology has been pretty much nailed for years, so that shouldn't need changing, except to adopt the 'latest and greatest' [sic] medium pushed by the record companies in an attempt to get us to buy the same music again.

 

[sergeauckland]

Then you're SooL, I'm afraid...

Most modern, consumer, digital technology would be impossible to implement in a more discrete form. It would be huge, and take a huge amount of power, even if the discrete digital logic circuits were available. Think of the environmental impact of that power.

Analogue electronics hasn't changed much, if you discount digital filtering/EQ, so you can easily buy an amp that should last you years. My main system is still the one I bought in 1988. You can still buy all-analogue amps.

DAC technology has been pretty much nailed for years, so that shouldn't need changing, except to adopt the 'latest and greatest' [sic] medium pushed by the record companies in an attempt to get us to buy the same music again.

I had to Google SooL. I assume you mean So Cool, in which case,thanks.

I just choose to use stuff until it's beyond sensible repair. I don't have a smart phone both for reasons of privacy and obsolescence. My mobile 'phone is now 11 years old and stays switched off except when making or expecting a call. My camera is 18 years old, so as long as it keeps working, I'll keep it. I just hate buying new stuff.

I know that if everyone was like me, tech companies wouldn't have a business, but that's not my problem. Maybe the world would be a better place with less consumption and fewer tech companies anyway.

S

 

[restorer-john]

Good luck fixing a broken bit on a core memory... Try running Windows 10 on a Sinclair ZX80...

To be fair, the Sinclair ZX-80 (early versions) were completely socketed and ICs could be replaced individually by the end user/hobbyist. Even on the ZX-81 (four main chips) two were socketed and all were available as spare parts. I remember as a kid replacing my RF modulator (astec IIRC) as a spare part when it failed.

And, we can run a ZX Spectrum emulator on Windows10...

https://www.microsoft.com/en-au/p/zxspectrum/9wzdncrcs8tt?activetab=pivot:overviewtab

That silicon is sand; the only thing worth recycling is the precious metals, and that's what is recycled.

Think of the environmental impact of that power.

Let's be honest, only one tiny part of the product is 'sand'. All the rest is petrochemicals, heavy metals, huge amounts of production and waste chemicals and an enormous amount of wasted fossil-fuel based energy. All of which ends up doing wonders to our atmosphere-not.

Premature disposal of vast quantities of entire products due to component part failures cannot be discounted as a "price to pay" for progress. It is a poisonous legacy, and one that will be derided as irresponsible by generations to come.

Manufacturers the world over must be held accountable for premature failures, ultimate disposal and recycling. Sure consumers will pay more in the short term, but efficiencies in manufacture to aid disposal/recycling will come thick and fast.

 

[captain paranoia]

Let's be honest, only one tiny part of the product is 'sand

Yes. I know how much energy is required to make that bit of sand. My point was that, once used and made obsolete, there's no intrinsic value in old silicon.

And, actually, much modern silicon is simply that; silicon; there's no plastic in a flip-chip package. There's a bit of metal for the solder balls, but that's it. This is true of large memories.

I'm not sure what we should do about the consequences of Moore's Law. It has given us enormous benefits, but has consequences for waste.

I hate waste. I'm not a 'must have latest iThing' sucker. My main PC is rescued from a skip. My bedroom PC is an old secondhand all-in-one. My PC monitor is a TV, rescued from the side of the road and repaired. Apart from my main, 1988 hi-fi system, all my other audio equipment is secondhand. I repair stuff for friends. I still use my 1993 Psion Organiser, which I have repaired on a number of occasions. I hate not being able to repair stuff.

BUT I was merely pointing out that there are many benefits that come from integration. The fact that I can type this post on a tiny handheld computer that cost me £30, and has lasted me three years so far, is one of those benefits.

Our profligate use of fossil fuels, and single-use plastic is a far greater environmental threat.

ps. I too built a ZX80. And later, a Jupiter Ace (and made plenty of mods to it; it still works, as does my QL...). I also created a VHDL implementation of the Ace some years ago...

 

[restorer-john]

BUT I was merely pointing out that there are many benefits that come from integration. The fact that I can type this post on a tiny handheld computer that cost me £30, and has lasted me three years so far, is one of those benefits.

Absolutely. Those roadside-pickup computers and rescued and repaired monitors, along with many decades worth of HiFi and electronic repairs are our contributions to a better environment.

It's hilarious that a few hipsters are now getting TV airtime for running "repair workshops" where, shock of shocks, people can be shown how to do basic repairs themselves. Fixing things themselves- who would have thought?

Do you remember how hot the 7805 got in the Sinclair? Typical English design for subzero temperatures. In Australia, my the ZX-81 got so hot it almost burnt my hand through the plastic case. I put a large heatsink on the poor regulator. And the RAM-pack. Let's not talk about intermittent contacts in that blue edge connector socket/pcb...

We used it with an I/O interface I built, to run a Tasman Turtle tethered robot and came second place in a school science competition in around 1982. I built the turtle, the interface and ironed out all the electronic bugs, my best mate was the programmer. He had the turtle map the entire science laboratory and plot out a mimic on butchers paper. Here's a pic of a turtle from the internet.

Back when this was brand new:

Februrary 1982. Time flies, and I'm feeling really old right now...

 

[sergeauckland]

I still have my Sinclair Scientific calculator from 1974, with reverse polish notation. Still use it (very) occasionally for calculating filters. It cost about a week's salary then.

 

[restorer-john]

I still have my Sinclair Scientific calculator from 1974, with reverse polish notation.

You'd like this one of mine:

I can't resist an old calculator, especially the Nixie, VFD and early 7 segment LED ones.

 

[captain paranoia]

Back when this was brand new:

I picked up one of those for £5 a couple of years ago...

 

[Soniclife]

In Australia, my the ZX-81 got so hot it almost burnt my hand through the plastic case.

It was no different in the freezing UK, they got hot, the ram packs wobbled, and keyboard was crap. They were great.

 

[Soniclife]

My mobile 'phone is now 11 years old and stays switched off except when making or expecting a call.

This behaviour drives me bonkers. The number of times I've been told of by an old person for not letting them know something important, when they had their phone off, because they didn't know they needed to turn it on, because no-one called them to tell them to turn it on.

 

[sergeauckland]

This behaviour drives me bonkers. The number of times I've been told of by an old person for not letting them know something important, when they had their phone off, because they didn't know they needed to turn it on, because no-one called them to tell them to turn it on.

If somebody wants to tell me something important, they can send me an email or call my landline. My mobile 'phone is for my convenience not theirs.

S

 

[Wombat]

This behaviour drives me bonkers. The number of times I've been told of by an old person for not letting them know something important, when they had their phone off, because they didn't know they needed to turn it on, because no-one called them to tell them to turn it on.

I find most people don't answer their phone because most calls are trivial or inconvenient. Eventually they get around to checking their message bank and respond at inconvenient times. This can become a merry-go-round. The mobile phone has become a lifestyle toy for bored individuals and an unwanted marketing sewer.

I leave mine at home(I use it like a landline phone) unless I am travelling. It has never been connected to the internet - I have a computer for that.
Of course, I grew up in a time when we didn't have one hand disabled by a bottle of water or a smartphone.

 

[ajawamnet]

...there also seem to be severe problems with lead-free solder which seem to not have been resolved yet, and which may reduce durability drastically ...

...and I'm not sure if, at least, the consumer industries' desire to resolve them is that strong, while, on the other side, many 'consumers' have lost the sense for durability and don't demand for it anymore...

https://nepp.nasa.gov/whisker/background/

https://www.maximintegrated.com/en/app-notes/index.mvp/id/5250

Yep - we are forbidden to use lead free in any of the Mil stuff I design. In fact, there's a facility that measures lead content at one of the army depots:

http://www.army.mil/-news/2010/06/11/40712-army-works-to-decrease-lead-free-electronic-components/

In fact, the whole Toyota accelerator thing - check this out:
http://nepp.nasa.gov/WHISKER/reference/tech_papers/2011-NASA-GSFC-whisker-failure-app-sensor.pdf
There was a whole sh## storm running about that at EE Times...
http://www.eetimes.com/messages.asp?piddl_msgthreadid=44469&piddl_msgpage=1#msgs
http://www.eetimes.com/author.asp?section_id=36&doc_id=1321694&

When the government agency that had NASA do this got the report, they stated that there was no more funding for it. NASA said OK cool, we don't care we; we got satellites to worry about... so the released the report. Evidently some at DOJ were pissed...

CALCE at U of MD is a great resource for subject like this:
https://calce.umd.edu/
https://calce.umd.edu/research

The counterfeit part thing.... for instance we can't buy parts from certain places (Newark for one) due to them not adhering to our requests for counterfeit part mitigation and tractability.

There's a great story over at one of the "maker" suppliers - Sparkfun (I have my opinions on this - see the marketturd link in my sig) of fake Atmels for the maker's darling Arduino boards:

https://www.sparkfun.com/news/364

https://www.sparkfun.com/news/384

Turns out they're actually regs for Intel's
https://www.sparkfun.com/news/395

 

[boXem]

Yep - we are forbidden to use lead free in any of the Mil stuff I design. In fact, there's a facility that measures lead content at one of the army depots:

http://www.army.mil/-news/2010/06/11/40712-army-works-to-decrease-lead-free-electronic-components/

In fact, the whole Toyota accelerator thing - check this out:
http://nepp.nasa.gov/WHISKER/reference/tech_papers/2011-NASA-GSFC-whisker-failure-app-sensor.pdf
There was a whole sh## storm running about that at EE Times...
http://www.eetimes.com/messages.asp?piddl_msgthreadid=44469&piddl_msgpage=1#msgs
http://www.eetimes.com/author.asp?section_id=36&doc_id=1321694&

When the government agency that had NASA do this got the report, they stated that there was no more funding for it. NASA said OK cool, we don't care we; we got satellites to worry about... so the released the report. Evidently some at DOJ were pissed...

CALCE at U of MD is a great resource for subject like this:
https://calce.umd.edu/
https://calce.umd.edu/research

The counterfeit part thing.... for instance we can't buy parts from certain places (Newark for one) due to them not adhering to our requests for counterfeit part mitigation and tractability.

There's a great story over at one of the "maker" suppliers - Sparkfun (I have my opinions on this - see the marketturd link in my sig) of fake Atmels for the maker's darling Arduino boards:

https://www.sparkfun.com/news/364

https://www.sparkfun.com/news/384

Turns out they're actually regs for Intel's
https://www.sparkfun.com/news/395

Not sure to follow the links between lead free, tin whiskers and counterfeit parts. Could you develop?

 

[ajawamnet]

Not sure to follow the links between lead free, tin whiskers and counterfeit parts. Could you develop?

As to lead free one of the issues with long term reliability is tin whiskers. One of the reasons a few others have mentioned. We've seen whisker growth across tin-plated pins on IC SMT leads. If you look at the NEPP/NASA thing that Bob-23 mentioned, you'll see pics of SMT 0803 caps begining to exhibit dendrite growth.

This is why the designs I do for MIL and critical infrastructural are exempt from RoHS.

As to counterfeit parts, unlike the obvious silly that occurred at Sparkfun, we've seen issues with IC's that are coming in that are NOT from the OEM semi mfg, but appear to be form-fit-function the same as the original. Problem is that they suffer from infant mortality and premature failure. When de-encapsulated they do not match the die's from a known OEM part.

When we send out builds to contract manufacturers for audio designs for stuff like critical comm gear, we see these failures after the assembled boards arrive back at out integration facility. This has led to use being very picky with things like AS9100 cert of those facilities.

With the commercial/consumer gear, we have even less control over what they hell they end up buying from god-knows-who, since a lot of this happens off-shore. All we know is when I get a buncha complaints for the testers on our lines I see a lot of CF parts. The scary part for the biz guys is when they send out stuff that appears to pass IIR and other inspections as well as functional tests, and then fails in the field it can ugly real quick.

So a lot of what we see when trending MTBF (mean time between failure) is skewed due to the counterfeit's that make it thru all this. At first everyone was blaming reduced package sizes, and things like the newer DFN/QFN IC packages (which really do have issues like center pad float due to improper SMT paste mask design), but in some cases it turns out to be CF parts.

I'd hate to be a CM right now...

There's a great vid - part of our MIL guys training - that was a 60 Minutes special years ago:

At 7:08 you'll see scads of IC's sitting on the ground - these end up being re-labeled and sold as look-a-like parts.

As I mentioned there are few distributors that the MIL branch of our work is allowed to specify/buy from. Since it's also ITAR we have a bit of control...

With the consumer gear I design - eh, not so much.

 

[boXem]

As to lead free one of the issues with long term reliability is tin whiskers. One of the reasons a few others have mentioned. We've seen whisker growth across tin-plated pins on IC SMT leads. If you look at the NEPP/NASA thing that Bob-23 mentioned, you'll see pics of SMT 0803 caps begining to exhibit dendrite growth.

This is why the designs I do for MIL and critical infrastructural are exempt from RoHS.

As to counterfeit parts, unlike the obvious silly that occurred at Sparkfun, we've seen issues with IC's that are coming in that are NOT from the OEM semi mfg, but appear to be form-fit-function the same as the original. Problem is that they suffer from infant mortality and premature failure. When de-encapsulated they do not match the die's from a known OEM part.

When we send out builds to contract manufacturers for audio designs for stuff like critical comm gear, we see these failures after the assembled boards arrive back at out integration facility. This has led to use being very picky with things like AS9100 cert of those facilities.

With the commercial/consumer gear, we have even less control over what they hell they end up buying from god-knows-who, since a lot of this happens off-shore. All we know is when I get a buncha complaints for the testers on our lines I see a lot of CF parts. The scary part for the biz guys is when they send out stuff that appears to pass IIR and other inspections as well as functional tests, and then fails in the field it can ugly real quick.

So a lot of what we see when trending MTBF (mean time between failure) is skewed due to the counterfeit's that make it thru all this. At first everyone was blaming reduced package sizes, and things like the newer DFN/QFN IC packages (which really due have issues like center pad float due to improper SMT paste mask design), but in some cases it turns out to be CF parts.

I'd hate to be a CM right now...

There's a great vid - part of our MIL guys training - that was a 60 Minutes special years ago:

At 7:08 you'll see scads of IC's sitting on the ground - these end up being re-labeled and sold as look-a-like parts.

As I mentioned there are few distributors that the MIL branch of our work is allowed to specify/buy from. Since it's also ITAR we have a bit of control...

With the consumer gear I design - eh, not so much.

From the NEPP/NASA thing: "It is widely believed (though reasons remain somewhat of a mystery) that Pb when alloyed with tin imparts whisker-inhibiting attributes to the final finish."
Not really a proof.
The main issue with lead free solder is its not so good handling of mechanical stress.

 

[ajawamnet]

From the NEPP/NASA thing: "It is widely believed (though reasons remain somewhat of a mystery) that Pb when alloyed with tin imparts whisker-inhibiting attributes to the final finish."
Not really a proof.
The main issue with lead free solder is its not so good handling of mechanical stress.

It's a combination of creep and tin whiskers. Since the time that Clech published his paper in 2004 ( http://www.jpclech.com/Clech_APEX2004_Paper.pdf ) comparing SnAgCu (SAC) to SnPb (typical tin-lead), companies like AIM and Indium have developed solders that exhibit stress characteristics that are quite good over time and temperature:

https://en.wikipedia.org/wiki/Solder_alloys
Note the mechanical characteristics of Sn77.2In20Ag2.8 are very similar to Sn63Pb37...

https://www.researchgate.net/public...ad-Free_Solder_Joints_in_Electronic_Packaging

https://aimsolder.com/technical-art...ering-properties-alternative-lead-free-alloys

But the problem for what we do in the MIL arena still lies in the worry that whisker growth can occur - esp. with the addition of alloying materials for increasing mechanical strength -

https://www.nature.com/articles/s41598-019-44758-3.pdf?origin=ppub

From : https://www.mdpi.com/2076-3417/9/10/2044/pdf

"Currently, researchers mainly improve the performance of lead-free solders by microalloying and
nanoparticles strengthening. Microalloying is a method of improving the properties of the solder by
adding trace alloying elements to change the solder composition. The alloying elements can be mainly
divided into rare earth (RE) elements (Ce, La, Pr, Nd, etc.) [6,7] and other metal elements (Ga, In, Mg,
Ni, Ag, etc.) [8–11]. However, there are still some shortcomings in this method, which limits the widely
application of the lead-free solders containing minor alloying elements. For example, adding excessive
RE elements will result in the formation of Sn whisker on the surface of solder joints, causing premature
failure of electronic device [12,13]."


Another issue occurs due to embrittlement. (https://en.wikipedia.org/wiki/Embrittlement) Some of the failures that were initially attributed to SAC solder mechanical failures were exacerbated - and possibly caused by - ENIG ( Electroless nickel immersion gold ) of the bare PCB's.

ENIG is one of many methods to prevent oxidation of PCB lands and pads. For years, SMOBC ( Soldermask Over Bare Copper) with HASL (Hot Air Surface Level) was used. In the early 90's some PCB fab houses started using OSP (Organic Surface Protectants). For example Rad Shacks protoboards used this. Solderability sucks as compared to other methods.

Typically, ENIG/ENEPIG (nickel electroless palladium immersion gold ) are preferred by consumer Contract Manufactures (CM) since it provides a much flatter surface for automated paste and pick and place/reflow. The gold dissolves into the solder; to prevent embrittlement., the percentage of gold in the joint should be < 2.5-3% /wt ( http://www.circuitnet.com/experts/87487.html )

The problem is that if not controlled it can build up to a point that long term reliability is affected, since embrittlement can occur months after assembly.

Although gold of a thickness under/equal to 2.54 um was and still is considered OK under 4.5 of the J-STD_001 the F revision of the specification from , section 4.5.1 added:

"Note: Gold embrittled solder connections can occur regardless of gold thickness when solder volume is low or the soldering process
dwell time is not sufficient to allow the gold to dissolve throughout the entire solder joint. "

When we were dealing with what we thought was SAC mechanical failure for some consumer gear, the lab in Oregon found that cross-sections under electron microscopy were actually embrittled with a very obvious non-metallic structure that had formed.

Some great papers here:
https://semlab.com/papers/

So another thing our MIL guys will not allow in addition to lead-free is ENIG/ENEPIG - all our fab and assembly drawings now state this.

All in all, it's difficult to ascertain, in postmortem of a failed assembly, that the cause is due to the solder or the PCB surfacing...

 

[Sal1950]

But what if I prefer to have larger, more expensive and possibly less reliable but repairable electronics?

Same way I feel about my old pickup truck. Mostly manual everything, but it just keeps on trunkin.

 

[rwortman]

I have no data but personal experience outside of audio. I was in the business of repairing equipment in automated high tech factories for a long time. In the 80's component failures on circuit boards were pretty common. In the last 10 years much less so.