24
\$\begingroup\$

These days misleading commercial practices are not uncommon despite the presence of laws and regulations designed to protect consumers.

It should not be impossible for electronic component manufacturers to provide misinformation about their products in order to create a competitive edge against their rivals.

In fact, it might be more likely for such illegitimate acts to take place undetected in technical fields than in the general consumer market, by relying on the fact that few people have the resources, expertise and equipment required to verify the esoteric technical specifications.

For example, it would be nearly impossible to find out the noise performance of a high-end auto-zero precision op amp without equipment that is more precise and noise-free; or to produce a statistically significant assessment of the endurance and data retention of a flash memory chip without a long test period and a large sample quantity.

  • How common is it for major component manufacturers to lie about their components?
  • What regulations are there to prevent such acts from taking place?
  • Where and when should an electronic engineer be cautious about the trustworthiness of the component specifications? e.g. which areas of electronic engineering, which characteristics in the specifications, etc.
\$\endgroup\$

closed as primarily opinion-based by The Photon, Bence Kaulics, PeterJ, Daniel Grillo, Dave Tweed Aug 22 '16 at 13:14

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.

  • 3
    \$\begingroup\$ I admire your cynicism! I would add to your list: How often do component manufacturers not mention well-known shortcomings of a device in their specifications and data sheets? A related area of interest is counterfeit parts, which are coming into this country from China. Many are rejects from the original name brand manufacturer, but have found their way to repackaging operations, and thence into the mainstream electronic components supply chain. This is one of the downsides of having components manufactured in the Far East - there's a lot of "leakage". \$\endgroup\$ – FiddyOhm Aug 21 '16 at 15:07
  • 3
    \$\begingroup\$ Keep in mind that the vast majority of the volume of the components used are done by technical teams. At the company I work for, we have uncovered performance problems in big-name companies' components sometimes due to system-level interaction. Just about every part will have its expected vs. real performance considered and documented by experienced engineers. We certainly expect performance no worse than what's in the public datasheet. \$\endgroup\$ – user2943160 Aug 21 '16 at 15:13
  • 4
    \$\begingroup\$ Reputable 'big name'manufacturers are usually honest enough [tm] in the specs they write. Complex digital parts will usually start life with substantial errata and it can take years to eliminate some. And a die shrink may add new ones. | I discovered a fault in a microcontroller made by a manufacturer that used to be reputable but whose brand quality had been eroded by spin offs of various divisions. The part failed data sheet spec BUT this was due to a lack of detail - they failed to specify a maximum allowed power supply t urn on rise. Initially they accepted my reports and ... \$\endgroup\$ – Russell McMahon Aug 21 '16 at 17:06
  • 4
    \$\begingroup\$ ... started to to me about the issue but suddenly went all formal and tricksy - the us office realised they did not want to be liable for the problem and stared asking for detailed reports and conditions where the fault occurred - while and competent fool could duplicate the fault wit ease following my simple instructions. In the end they were useless, but did not change the data sheet. I had to build a work around add on PCB to salvage boards that had been made. Manufacturing was in taiwan. I was surprised to subsequently have a moderately senior and helpful company rep turn up... \$\endgroup\$ – Russell McMahon Aug 21 '16 at 17:16
  • 4
    \$\begingroup\$ ... at the factory and ask re problems. BUT by then the damage was done and bypassed. [[The device was a Zilog Z8 microcontroller. Very slow Vdd rise time caused the clock to not start and it had to be powered down to Vdd <= ~ 0.2V to clear the lockup. This was an exercise machine and power came from user driven alternator so could not be controlled simply. So yes, in that case the manufacturer were shonky and refused to acknowledge their data sheet inadequacy. I'd not use their products again - should they still exist. \$\endgroup\$ – Russell McMahon Aug 21 '16 at 17:20
22
\$\begingroup\$

In the general sense, I have never known a reputable electronics device manufacturer to deliberately lie about the performance of their products, although the datasheet may be updated to reflect more complete test results.

A preliminary datasheet should be viewed with a bit of caution as the device is unlikely to be fully characterised at that point with some graphs and entries calculated rather than measured.

A bigger issue is understanding component specifications (typical subjects linked).

The time when someone has deliberately lied is where a counterfeit component is sold to you. As noted in the comments, many (but by no means all) of these are sourced from China.

This is a major problem in the high reliability arena and although convictions do occur, the problem persists.

The reasons are numerous; the issue in avionics (for instance) is that boxes may be 40 years old and sourcing components for repair is incredibly difficult (and expensive); the financial gain for a successful counterfeiter are large.

Other industries face the same problems.

\$\endgroup\$
  • 11
    \$\begingroup\$ "never known a reputable electronics device manufacturer to deliberately lie". Obviously, then they're no longer reputable. :) \$\endgroup\$ – pipe Aug 21 '16 at 15:29
  • 1
    \$\begingroup\$ @pipe: indeed. The usual suspects are pretty well reputable; if I have not heard of the outfit before that can be a red flag (no pun deliberately intended). :) \$\endgroup\$ – Peter Smith Aug 21 '16 at 15:41
  • 1
    \$\begingroup\$ Second the notion of understanding the specs. For example, it can be dangerous to design for "typical" or other statistical specs on a data sheet, as many individual parts may not meet them. Or that tolerances and temperature ranges tend to come in bins; that is, a 20% part may never be within 10%, because 10% parts may be taken out to be sold as 10%. Finally, most specifications are measured under a very specific set of conditions (supply voltage, temperature, input range) and performance may change significantly under other conditions. \$\endgroup\$ – Urausgeruhtkin Aug 21 '16 at 16:02
  • 2
    \$\begingroup\$ "A preliminary datasheet should be viewed with a bit of caution..." - especially, I might add, when the component it describes has been on the market for some number of years and the data sheet is still denoted as "PRELIMINARY". Dozens of examples of this out there. \$\endgroup\$ – FiddyOhm Aug 21 '16 at 17:00
  • 1
    \$\begingroup\$ @FiddyOhm: I have recently had a component where the datasheet never got beyond preliminary be discontinued; the manufacturer clearly could not get the yield required to the datasheet parameters. \$\endgroup\$ – Peter Smith Aug 21 '16 at 17:03
13
\$\begingroup\$

There are many experiences I have had with components not meeting specifications. MFG process changes will correct the problem sooner or later. Often from customer feedback as in my case.

examples; Fan MTBF. We had a product using 8 large HDD's and 8 fans per cabinet and needed to verify system MTBF was > 25khr. Thus we ran extended life tests on 30 fan units with start stop every minute and same with HDD's rated for 10k min start stops. Eventually we found Nidac fans, a reputable source, to fail on certain positions and I determined it was a process failure with Hall sensor alignment at the magnetic commutation points were off. I sent our simple start stop circuit design and demanded 100% testing for 100 cycles. They fixed the problem and our yield went to 100%. 15 yrs later different company product and fan, product reported a few dead fans. I found the same symptom and tested 100 fans with 10% failure rate and told supplier ( big fan OEM ) the same and sent circuit and got the same results.

Back in 1977 when Burr Brown made fast Hybrid ADC's mil-std 883B qualified for nuclear inspection robot system I designed, I found 2 chips had the same problem with missing codes on a linear sweep of input voltage usually occurring near the xxxxxxx01111 to xxxxxxx10000 boundaries. So I surmised it is was an internal VRef shift due to digital currents on internal wire bonds and asked BB for a solution. They had none at the time. So I ordered Industrial quality parts instead and found they did not have this problem and chalked it up to a BB process error or change not implemented yet on the Hi-Rel part process with X-Ray inspection.

I have had hundreds of similar experiences like caps leaching after reflow and changing values only from certain vendors or flash memory reliability issues, but for the most part in my years as Eng Mgr for contract MFG. 1% of the failures are bad parts and 95% are solder process related in a good design, while design margin faults made up the rest. The solder process could be argued whether it was PCB design or process design/ materials, I suppose.

As a Test Engineer I took my job very seriously and the quality of the datasheets detailed specs and test conditions is critical to the faith I put into suppliers. ... which says a lot about EBay items with no specs. ..... that's a crap shoot and the seller's reputation is at stake.

As far as Independant Brokers are concerned, having seen the operations of many from the inside, they have no clue about detailed specs or traceability and fake parts and rely on theIR supplier relationship to block bad/fake/clone parts. If you do large sales business, you must establish trust and consequences. For military orders, Traceability Certs are required, but can also affect your credibility if fake. Just ask any broker if they have 10ns rise time and see what response you get ;)

As far as litigation and liability all of the Big OEM's have a legal staff to handle law suits on issues like performance and patents.. So read the Fine Print.

\$\endgroup\$
  • 6
    \$\begingroup\$ Your examples sound more like issues the manufacturer really didn’t know about, not deliberate lies. \$\endgroup\$ – Michael Aug 21 '16 at 17:02
  • 1
    \$\begingroup\$ true, I trust all major brand OEM's but verify. so I think diligence is due to both the supplier and customer to validate all assumptions before product is ready. I would call it "escapes in quality" , not lies. In some cases Quality Mgr's did not validate the work of their process engineers like PowerOne did moving mfg. from SD to Mexico and was told emphatically I was wrong, then I corrected their test process to HIPOT with secondary grounded and 30% units failed due to stray coupling stress from process control faults on gaps, adhesives etc. They agreed. \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Aug 21 '16 at 17:13
  • 4
    \$\begingroup\$ In most countries , Engineers are ethical and marketing plays games with specs. If an error exists, I give the benefit of the doubt until verified, and chalk it up to Murphy's Law and the Peter Principle. \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Aug 21 '16 at 17:21
12
\$\begingroup\$

What regulations are there to prevent such acts from taking place?

I'm an engineer, not a lawyer, but what's been explained to me is that the datasheet essentially forms part of a contract between me and my supplier. I agree to pay for the parts, they agree to supply parts that meet the specifications laid out in the datasheet.

That means if they deliver parts that don't perform as specified, I (my company) would at least in theory have the ability to take them to court for failing to meet their obligations.

In practice, the potential loss of sales if a company gets a reputation for selling poor-quality goods is a more effective motivator to provide accurate datasheets.

\$\endgroup\$
  • 3
    \$\begingroup\$ Many years ago my boss the genius did a circuit design with a number of poor design choices. The circuit work well enough however, because a quad opamp worked far better than spec sheet said (offset, open loop gain, etc). Then one day the manufacturer lost the original recipe and the new parts only worked a little better than specified. The circuit stopped working until I could redesign \$\endgroup\$ – Robert Endl Aug 21 '16 at 16:22
  • 2
    \$\begingroup\$ "the manufacturer lost the original recipe" haha, love it. It's a shame how the manufacturing process is still so wildly varying that this can even happen \$\endgroup\$ – KyranF Aug 21 '16 at 17:42
  • 2
    \$\begingroup\$ @KyranF, that typically happens when you try to move the manufacturing site. \$\endgroup\$ – The Photon Aug 21 '16 at 23:56
  • 4
    \$\begingroup\$ @KyranF: The manufacturing process is so bad that you can't even be sure that cutting LEDs from the exact same wafer in nearby areas will be sufficiently similar that you can use them in the same display on a piece of aircraft equipment. I made a fair bit of money just making equipment to bin (sort) LEDs by intensity and apparent wavelength (as measured by a human vision CIE model) so they could be sure when they sell a box of display pieces to a customer, that their customer could use ones from the same box on the same instrument. \$\endgroup\$ – jonk Aug 22 '16 at 2:27
  • \$\begingroup\$ @jonk cool, quality control is a lucrative business \$\endgroup\$ – KyranF Aug 22 '16 at 16:07
11
\$\begingroup\$

Datasheets are legally binding, so lies would open a manufacturer up to lawsuits. Big customers can and will return shipments that don't meet the specs, and when quality guarantees are required (as in automotive), they'll insist you track down and fix the root cause of the defect.

What's far more common is "specmanship" -- providing and advertising specifications only for the most favorable operating conditions. For example, a frequency-dependent parameter like total harmonic distortion or PSRR could be given at only one frequency. Other parameters could be given only at the nominal temperature and voltage. It's also possible to specify an external test circuit that gives better results than what you'd see in common applications.

Another thing to watch out for is typical values, which aren't guaranteed by production testing.

I think you're overestimating the amount of equipment and knowledge that's required to test datasheet parameters. Semiconductor companies use the same kind of lab equipment as everyone else -- Agilent, Keithley, etc. And if customers didn't understand the parameters, we wouldn't bother advertising them, much less paying to test them in production.

\$\endgroup\$
4
\$\begingroup\$

Lie? no, it really isn't in their best interest. For the little designers a datasheet being wrong is a case of deal with it. But for the big companies...

Without going into too much detail a component manufacture moved their fab and the "leakage" changed and it caused a number of major in-field issues. A request to match their datasheet was met with a simple "NO" which was met with a simple "your company is blacklisted, every single component. Will will accelerate obsolescence activities to actively remove your parts from usage". An amicable deal was reach to offset the penalties from the end customer as well as redesign cost as the part could not be made to meet the original datasheet.

That's not to say they don't make mistakes (equally a recent project was met with a factor of 5 difference in stray capacitance) and an update to the datasheet corrected this.

Likewise there is sales pitch with regards to their datasheet & equally failing to deliver what it intends. Here is a link to a discussion I had with TI over one of their digital isolators.

ISO7420M failure to transmit data reliably

\$\endgroup\$
4
\$\begingroup\$

I highly doubt reputable US electronic component manufactures will deliberately have inaccurate infomation in datasheets. But I will not put it past some off the offshore electronic component vendors.

True Story: I worked with a off shore electronic component manufacture who promised a let say a 1uF +/- 5% capacitor far cheaper that the US based vendor. When they sent engineering samples for validation they sent 10% more then the required amount. When queried about the generosity, the response was 10% components fail, thus the extras. When suggested that the components be tested prior to shipping, component price was increased, thus making the US base vendor competitive.

Morale of the story buyer be aware

Sometime way back when I remember reading an article from legendary Bob Pease who started that datasheet is a just a marketing document. Only way to understand the true specifications is to test them yourself. So my advise to you is if a certain parameter is critical to your design then tested it yourself and make yourself comfortable with the part.

Reference:

\$\endgroup\$
  • \$\begingroup\$ Bob Pease was a very wise man. Datasheets are good but testing is better. \$\endgroup\$ – Robert Endl Aug 22 '16 at 1:47

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