Component reliability

Is it me, or has component quality gone rapidly downhill over the last few years?

I remember "Back in the day" (:rolleyes:) when it was shockingly rare to get a faulty component. It was that rare that cross shipping with some of the big retailers of the time was common place if you asked for it.

I suppose its a case of components constantly getting more complex and QA not being able to keep up with it? or perhaps its worth letting the few that slip through based on the economics of faulty# vs cost of RMA ratio?

PSU's and graphics cards are the prime ones for this. Although the number of cheapo PSUs on the market really makes the situation worse.

I pretty much accept that for every 3/4 machines I build, something will have to go back (and I don't buy cheap parts, is just false economy).

Anyone else think its getting a bit silly? I'd happily pay a few quid extra for higher quality components / manufacturing.

well when i built my PC in december, the motherboard didn't work, the graphics card didn't work and the PSU had a faulty PCI-E rail.

was nearly january before i got the motherboard back becuase City-Link bolloxed up with the return of the motherboard.

rather annoying, but i'm with you, i'd definately rather pay an extra few quid to know it's gonna work when i plug it all in!
instead of having to stick old parts in and test them 1 by 1..

edit: and there's also all the delivery and everything, if you RMA something you're tripling the CO2 emmisions on the delivery.. may not sound like much, but add everything up that's been RMA'd that could've been helped..

I totally agree with you both, with my current new build I've had 4 faulty mobos, 2 faulty PSUs and 1 faulty GFX card! This kinda thing rarely happend in the past, I just think companies put themselves under such pressure to be the 1st to market and forgoe (sp?) alot of QA and testing and hope for the best on returns (lack of funding to QA vs RMA costs).

I'd happily pay a few pennies more knowing that its all top quality stuff and will work 1st time and last a long while too.

Thats another thing I guess things arent designed to last too long either as they want you to upgrade often and buy more of their 'wares!

Just my 2 cents! :)

I get very suspicious about this Agent, certain suppliers have a very high incidence of sending me dodgy components, while others always send me working ones. I have to wonder if it's the etailers sending out stock that has been RMA'd to them, to save money on having to return it to their suppliers themselves, in the hope that either: a) we might keep it, or b) we might not notice in time before the 30 day RTB period and have to RMA it to the manufacturer ourselves...... it's just uncanny, I can predict with a fair amount of certainty the chance of getting working components it seems depending on the company.

it maybe has to do with the exponential increase on numbers of transistors... there more there is the more likely it is that one of them will go wrong and create a faulty device... just a thought

It's called relocating your factories to China. QA just isn't as good.

The dept I work in spends most of it's time giving concessions to manufacturers who make random changes to the materials they make components from. They get away with murder but because they are in India it costs 2p to make something and accountants only see the bottom line they don't see the additional costs involved in getting compliance. Scares me really because I don't think we get the same quality of aircraft engine parts.

I remember doing some research for my 28" flat screen CRT telly and you could guarantee that Sony, Panasonic and Phillips would be very reliable. 2 years later and the prices in general had halved and the quality across the board had gone through the floor. What happens is a company pioneers a process, the quality is high, then they flog the process to someone else who cut corners and the quality drops.

Quote:

---

Originally Posted by D.M

...with my current new build I've had 4 faulty mobos, 2 faulty PSUs and 1 faulty GFX card!

---

Bloody ell !!

Can you name them, or at least give us a clue?

Quote:

---

Originally Posted by DWhitley

Bloody ell !!

Can you name them, or at least give us a clue?

---

lol, I can laugh now! :crazy:

2x fautly abit IX38 Quad GT's
3x faulty Gigabyte X38-DQ6's (Ive just remember its 5 faulty mobos!)

2x faulty Tagan BZ900w PSUs (the new one with LED lights on the locking connectors, 3rd one is being shipped to me for Wednesday delivery, so it may be 3? just have to wait and see :rolleyes:)

1x faulty BFG 8800GTS!

Phew!

Oh and I forgot my 1st delivery of my Cosmos case was also damaged so I guess that counts as another fault! (ebuyer were fantastic on the RMA though so no biggy) :)

Its a good thing we have e-tailers such as scan and ebuyer to make up for the suppliers lacking standards :)

I havn't had a single problem with 5 new different builds and 3 new laptops in the last year..

Mostly from scan, a bit from OCUK and LOW and the laptops from Dell. Guess I'm just lucky.. although I am very picky about parts I buy of course.

Quote:

---

Originally Posted by Biscuit

Its a good thing we have e-tailers such as scan and ebuyer to make up for the suppliers lacking standards :)

---

True, True!

Quote:

---

Originally Posted by staffsMike

I havn't had a single problem with 5 new different builds and 3 new laptops in the last year..

Mostly from scan, a bit from OCUK and LOW and the laptops from Dell. Guess I'm just lucky.. although I am very picky about parts I buy of course.

---

God knows how many builds i have done over the past year but my record has been pretty decent also. 1 faulty motherboard 2 faulty hard drives and thats it!

Quote:

---

Originally Posted by staffsMike

I havn't had a single problem with 5 new different builds and 3 new laptops in the last year..

Guess I'm just lucky.. although I am very picky about parts I buy of course.

---

It does seem to be very much the case for myself too, touch wood. It would seem that being picky about the components you choose for a build can in fact save you a lot of headache.

As Biscuit says, with more and more transistors being added, the likelihood of failure is that more likely but as a result you could think that QA might be falling as the level of complexity in component build increases.

Things have changed in the last 5 years.
The reliability of the components hasnt but the QA/QC procedures have.

I have been in the electronics industry on/off for the best part of my civilian working life (around 10 years) and here's why things have changed.

The 'fall-out' in production has nearly always been at around 9%. Not the PCB's or the seperate parts. Its the finished product boxed and ready.

Thing is, with the Far Eastern markets, they only have a AQL (Acceptable Quality Level) test. This means, depending on the standard followed, that maybe only 1 out of every 30 units is tested in a 500 units batch. This keeps manufacturing costs down but inevitably leads to higher failures in the field.

THIS is the price you pay for buying abroad.

In the UK manufacturing the AQL would be, on a 100 batch, usually 15-20 per 100 tested. So, although not fool proof, it will usually spot errors more frequently.
Unfortunately though, to have any chance of competing with the foreign markets, we too have had to drop our batch testing levels to bring the costs down.

------------------------------------------------------------------------------------------------

I went to South Korea in 2002 to a VERY well known electronics manufacturer which is based there.
When i was on the plant tour, the Manufacturing Director told me that they have a 100% pass rate leaving their factory.
We all know that its impossible to get 1st time right every time.

This is how they achieved a 100% success rate upon finished goods leaving.........they had a manufacturing floor about 4 times the size of an aircraft hangar and there was faulty units/parts EVERYWHERE.
The other side of the plant they had another building that was at least 5 times the size of the manufacturing unit.
Know what it was??????

The rework station.....all the scrap (mustve been 30-40% of all manufactured product), ended up there to be reworked and shipped.

So please, how do SCAN rate for sending out reliable parts, and how good is their RMA system?

regards

acro666

I'd regard scan very highly at both personally although the RMA process is only being judged by what I have heard because i've never had to use it.. such is the quality of parts that get sent :D

It's very difficult to rate RMAs for any company unless you've had recent experiences. A lot of places the first problem is getting an RMA number as they don't answer the phone (often 1 person working 9-5 which is stark contrast to their selling staff).

The company with the best RMA I've ever seen is Crucial. It's bloody brilliant.

Crucial are good for RMA but were a pain with my address last time. Despite updating my details on their site, responding to their email with my new address and including a note in the box they still sent the replacement ram to my old address. Luckily I got on well with my old land lord. Aside from that they were very efficient.

I have had RMAs with Scan before where they haven't picked up the fault that I sent the product in about. Having said that it's pretty easy to get them to retest stuff if you think there's something that they've missed. I would rate their service as very good.

Crucial, Viewsonic and Logitech have the BEST RMA service i have ever come across.

Overclockers.co.uk and Dabs have the worst i have come across.

I wonder though, with regards to the OP, isn't it likely that we're simply seeing the result of having to stretch the same amount of resources further and further? After all, companies have to make more profit each year, by selling more product to more customers than ever before. What doesn't pick up so quickly though is the available resources (time, money and energy) to do the work, so things like quality and control testing fall by the wayside.

I think you only have to look at modern components to see why failure rates are so high. I have an old Gigabyte motherboard (AMD K6-2) from ~10 years ago that's still going strong. The tracks on the board are all about 1-2mm wide with good separation, and it's probably a 2 or 3 layer board. Yet many more recent boards have failed in less than 3 years and they all have tracks the thickness of a human hair that are packed so closely together the tolerances are beyond belief. Modern kit is just so much more advanced, I'm not at all surprised the failure rates are so much higher. That, combined with reduced costs and building down to a price point and I think you have the answer.

Quote:

---

Originally Posted by acro666

So please, how do SCAN rate for sending out reliable parts, and how good is their RMA system?

regards

acro666

---

I had no problems with Scan. Last summer, I received an expensive PSU on which it appeared the warranty sticker had been cut although nothing else was physically wrong with the unit (it tested fine). I contacted Scan and they immediately arranged for a collect and return at no cost. A new unit was shipped (I checked the serials so it wasn't the same unit sent back out), and it was identical with a cut warranty sticker. I contacted Scan again, who checked their stock and found all units were the same. They contacted the supplier who informed them that the seal had been cut at the factory to make a modification. Scan issued me a written warranty by letter acknowledging that the seal was cut and that the warranty would be honoured should it ever be needed (I'm sure they would have offered me a credit or refund had I wanted that instead). Throughout the process I was kept informed by email at all times.

I really can't fault that level of customer service.

Oh, and I wouldn't buy any memory other than Crucial - again their returns policy is "no questions asked" and first rate (I've had 2 returns with them out of many purchases).

One final one that WAS a surprise to me was Microsoft. I had a faulty webcam, and rather than fork out for return shipping to Amazon, I just phoned up Microsoft and after taking the serial no they just sent me a replacement free of charge, plus I must have had at least half a dozen emails from them asking if I was satisfied with the service I received. Like I said, a real surprise and again first rate!

Quote:

---

Originally Posted by Blitzen

Things have changed in the last 5 years.
The reliability of the components hasnt but the QA/QC procedures have.

I have been in the electronics industry on/off for the best part of my civilian working life (around 10 years) and here's why things have changed.

The 'fall-out' in production has nearly always been at around 9%. Not the PCB's or the seperate parts. Its the finished product boxed and ready.

Thing is, with the Far Eastern markets, they only have a AQL (Acceptable Quality Level) test. This means, depending on the standard followed, that maybe only 1 out of every 30 units is tested in a 500 units batch. This keeps manufacturing costs down but inevitably leads to higher failures in the field.

THIS is the price you pay for buying abroad.

In the UK manufacturing the AQL would be, on a 100 batch, usually 15-20 per 100 tested. So, although not fool proof, it will usually spot errors more frequently.
Unfortunately though, to have any chance of competing with the foreign markets, we too have had to drop our batch testing levels to bring the costs down.

------------------------------------------------------------------------------------------------

I went to South Korea in 2002 to a VERY well known electronics manufacturer which is based there.
When i was on the plant tour, the Manufacturing Director told me that they have a 100% pass rate leaving their factory.
We all know that its impossible to get 1st time right every time.

This is how they achieved a 100% success rate upon finished goods leaving.........they had a manufacturing floor about 4 times the size of an aircraft hangar and there was faulty units/parts EVERYWHERE.
The other side of the plant they had another building that was at least 5 times the size of the manufacturing unit.
Know what it was??????

The rework station.....all the scrap (mustve been 30-40% of all manufactured product), ended up there to be reworked and shipped.

---

Rework is damn expensive because it's labour intensive, however the far east pay peanuts so I guess they can afford to do it. Getting the process right first time is extremely important in the UK because any rework can severely reduce profits. I've seen cases whereby the firm has made a loss on a contract due to requiring rework on a load of cast aluminium gearbox casing because of the huge penalty for being late with the delivery. Damn bean counters chose the cheapest casters and the company paid the price.

Your quote of 9% "fall-out" reminds me of the issues RR has with their single crystal turbine blade foundry in Bristol. They were getting 9% yield per batch! Yep 91% scrap.

The Op (Agent) referred to components, but really they are sub-assemblies - but the QC process starts before then at the 'true' component level. The transistors/ICs?resistors/capacitors are all made in their millions, and generally (at that level) the performance of each component in a batch is repeatable - so those components undergo some form of batch testing to ensure the batch has conformaty to the device's specification. Nothing wrong with that, except that there will usually be some devices that fall outside that or will fail prematurely, which batch testing probably won't pick up.

Those components then go to a sub-assembly - like a computer PSU. These 'commodity items' are generally built down to a price. The design should take into account component tolerances, but anything outside that may cause latent problems later on. Electrolytic capacitors are a relatively common failure item. A component filtering a 12 volt rail must be adequately rated to carry the ripple current and have a voltage rating commensurate with the steady state voltager. For a 12V rail, I'd probably specify a (say) 20V component - but a builder building down to a price might use a 15V rated item. Result - risk of premature failure.

Fimnally testing - again one of the more expensive parts of the mfr process. A cheap builder might test 10% of a production line output - and that will probably only be a cursory test. The cost of replaceing the odd failure is more than offset by the saving in testing. A more reputable mfr might test every PSU - with some going through an extended test process. Result - they cost more but the product will probably be more reliable. In these case - cheap commodoty items, any failures probably won't be re-worked - as Iranu says, it is very expensive.

Compare that with an industry where safety is critical - nuclear, perhaps, or aerospace.

In that case the components that go to make up the PSU will have tighter specifications, and be conservatively rated. They will be more expensive. The final sub assembly - say a PSU - will be individually inspected, and probably soak tested and 'burnt in'. That is they will be run at full load for a period of time - perhaps 24 hours. This will weed out the premature failures on the 'bath tub' curve. Consequently the assembly can be put into service with a high degree of assurance. If goes into store, it may have a shelf life, after which it has to be retested. (Those electrolytic capacitors again!) That again pushes up costs - which is why (in this example) a switched mode PSU for use in an aircraft might cost a several hundred punds, while its ostensibly identical specification counterpart slots into the back of your PC for about £30.

But which one would you fly with? (and which one would you buy for your PC?)

One other thing is that computers have moved into the consumer market now, with ever increaseing speeds and newer versions to sustain a steady consumer demand.
where as a few years ago a product could have an expected running life of 5-10 years, now within 3 years it'll be obsolete and would of probably been replaced anyway.
Also as it moves into the consumer market prices are dropped to sell more units and corners are cut to reduce costs.

How often do we shop around or sugest alternative suppliers because they are cheaper?

Quote:

---

Fimnally testing - again one of the more expensive parts of the mfr process. A cheap builder might test 10% of a production line output - and that will probably only be a cursory test.

---

Not always...and not as a rule either.

All parts manufactured, especiall here, have to be CE marked for approval.
In order to gain CE Marking it has to be rigorously tested (golden sample usually i grant you).
THis will also entail a specified AQL for the end manufacturer.

If the manufacturer changes ANY component this must also be part of a submission to the accreditation body and the part may not be changed until it has been certified.


I have been involved with this for quite a long time. Not just through CE Approval but also FCC Approval (USA version) and ATEX Approval.
Although all 3 are slightly different and obviously ATEX is more rigorous.
(If something is FCC Approved though it is automatically qualified for CE Approval)

Quote:

---

Originally Posted by acro666

So please, how do SCAN rate for sending out reliable parts, and how good is their RMA system?

regards

acro666

---

I have been sent quite a lot of dodgy parts by them, but to be fair their RMA service is good, although refunds can sometimes be very slow.

Quote:

---

Originally Posted by Blitzen

In order to gain CE Marking it has to be rigorously tested (golden sample usually i grant you).

---

Problem with the CE mark is that anyone can issue it them self, with no high / governmental body checking up on them :(
It's only when people start to complain that issues arise.

In reality this means that a company (like Q-tec) can CE mark their products, blow up a few systems, and disappear overnight :(

Really good write up: HEXUS.net - Review :: HEXUS PSU (Power Supply Unit) Roundup - Taoyuan 2005 : Page - 23/26

I wish they would do more spot checking on people with CE marks.

Please allow me to explain just how reliable SCAN equipment is. I bought my first video-recorder from Scan many years ago, it is the SC-2500. It is still set up in my lounge and is in permanent use, it plays ALL regions and so far has never let me down.

I hope that I will be renewing my present computer set-up sometime in the near future and will certainly buy from Scan.

My question in this forum of two days ago was asking about scan was to ensure they are still utterly reliable!

One further question please, I note that my user name on my previous (two days ago) is now a very bright red. Is this significant, or has my machinery gone loopy?

regards

acre666

I lucked out with my last build in December (well, my only self-build, actually); everything worked, or worked well enough (I can't get my RAM timings down to the advertised 4-4-4-12, but instead have them at 5-5-5-16), and sometimes Windows doesn't recognise my optical drives properly (or do, but certain functions will be flaky - Vista especially does this) but even many of my games in XP only see the DVD drive, not the CD one. But the components do 'work' (though for the past few days I thought i had a problem with my sound card, but it somehow fixed itself).

But judging by reviews on NewEgg and such, it does seem like product QA has gone to the crapper...

Can we keep retailer comments to the retailers section please, it really has nothing to do with QA of parts :)

Quote:

---

Originally Posted by Agent

Problem with the CE mark is that anyone can issue it them self, with no high / governmental body checking up on them :(
It's only when people start to complain that issues arise.

In reality this means that a company (like Q-tec) can CE mark their products, blow up a few systems, and disappear overnight :(

Really good write up: HEXUS.net - Review :: HEXUS PSU (Power Supply Unit) Roundup - Taoyuan 2005 : Page - 23/26

I wish they would do more spot checking on people with CE marks.

---

True.....so very true.

There was one manufacturer (which shall remain nameless in case they sue me) in 2003 that put 150,000 security devices onto the European market without CE Approval.

They were fined a 7 figure sum. They actually had the nerve to plead ignorance in court even though they have been manufacturing in the UK (and Europe) for over 24 years.

As much of a git it may make me sound, I have little sympathy for them :)

24 years.....ignorance....classic :D

Peterb makes an excellent case in point with the PSU example. It's worth saving that post for when someone queries whether spending £50 rather than £25 on a 500W PSU.

One of the reasons I went for the IP35 Pro over the IP35-E is it's solid state capacitors. I'm sure I wouldn't have issues with the cheaper board but I think it's worth spending the extra for the quality (within reason) especially as I am overclocking.

It's a while since I was actively involved in QA and batch testing of components for the satellite industry but iirc there are certain specifications that we would hold the manufacturer to with regard the number per 100 to be tested, either by us, them, or a 3rd party. If the components had passed this continually for a given period then the AQL is dropped. I'm sure British Standards or

There will be various standards required by various industries and as has been mentioned, critical parts for things like aircraft require higher standards. The thing is it's not just batch testing that is required but a whole Quality Management System that has to be in place. Some companies ask suppliers to be ISO 9000 compliant etc. Big companies will have produced their own internal documentation that will then be audited to comply with EU/US legislation.

QA can be a minefield and I've got a story or two. Anyway off to watch England v France.

I noticed the surge in faulty electrical goods seems to have come after the ROHS legislation was proposed and later made into law. Apparently lead free solder just isn't as good.
If this is a factor, then failure rates will drop again as manufacturers learn better how to make reliable parts with the new restrictions.

acutally, i think things have got a fair bit better lately.

go back to the 2000-2003 erea, i was RMA'ing all sorts. Leadteck wrote on graphics cards resolutions they simply couldn't achive.

Some people 'borrowed' ideas on how to make capacitors only to find out they didn't work well for long. etc.

Seriously, things are much better now, the last 3 builds i've done all worked perfectly, with no hickups what so ever.

Quote:

---

Originally Posted by badass

I noticed the surge in faulty electrical goods seems to have come after the ROHS legislation was proposed and later made into law. Apparently lead free solder just isn't as good.
If this is a factor, then failure rates will drop again as manufacturers learn better how to make reliable parts with the new restrictions.

---

Bang on.
But im not sure things will improve alot from here on in.
Lead free solder and the cross over from Lead solder a few years back, and it didnt have the transition that most wanted.

Once a PCB is populated, the solder goes through an oven at a high temperature to 'cure' it off.
Unfortunately, everyone had to 're-profile' their ovens at the end of SMT lines as the lead free solder simply didnt act like anyone expected it too.

Not only ROHS was involved either.
On the back of that there is the ELV Directive and the WEEE Directive.

(Bloody hell :p ...... I thought i got away from all this stuff in September when i moved to the Chemical Industry)

Oh christ I remember when Matra Marconi Space (Now Astruim) were looking into lead free solder for satellite applications when I was there in '97 and I've no doubt that it cost them a bundle in development even though they routinely had solder development programs. The main issue was thermal cycling and not knowing how any of these new alloys would perform during a -55 to +125°C (2 degrees a minute iirc) cycle for 100+ hours. Then there were the metallurgical considerations with regard to alloy phases and there stability in the cycle and what effect this had on strength along with the formation of tin whiskers.

It was bad enough getting all the garb on for a class 2 cleanroom and then spending the next 4 hours staring down a microscope whilst lining up tiny capacitors under a machine that would then break the join and measure the force applied for a normal project let alone the amount of work for a whole new series of solders.

Whilst I've been out of that game for a while it seems that the Mil-Aerospace industry still has problems and at the time I remember our lead PCB designer who had 40+ years experience tell us how much nonsense it was for the satellite industry. She was most annoyed.

Military & Aerospace Electronics - Electronics designers grapple with lead-free solder guidelines

I meant to post about the ROHS in the opening post.

I've fixed *2* ATI graphics cards that were artifacting by using the heat gun method and a MSI 939 board which is next to me. The MSI is still flackey and randomly decides not to work, but every time I give it another heat gun blast, it'll come to life again for a bit. The graphics cards have both been fine.

BGA fitted components seem to have been hit really hard by ROHS.

Its also one of the main causes of the Xbox 360 failures!