Hacker News Reader

Comment by olalonde | Hacker News Reader

Why is it so expensive though? Also, why not fine or ban products that cause problems rather than requiring certification. It seems that would be a lot more efficient.

View on HN

Certification requires testing in a specialized lab that has a special anechoic chamber along with equipment that runs tens of thousands to hundreds of thousands of dollars. You also need an engineer or highly trained technician to run the test and/or interpret the results. You also have to wire up and instrument every device that comes in and they do vary a little. That's why the cost is in the thousands of dollars. There is a profit margin for the labs but it's also not a cheap service.

It is not efficient policing bad devices after the fact. You can have a bad device interfere with RF communications for miles. It then requires dispatching a team of humans with RF probing equipment and escalating to the Feds if the person refuses to comply. It does happen but it is time consuming.

Leaky RF can cause mass events such as https://www.cbc.ca/news/canada/calgary/carstairs-westview-co...

Here's a IEEE Video on the process of interference hunting: https://www.youtube.com/watch?v=elUDfDmIHLs

A short article on how some equipment on a tug boat was interfering with a town: https://www.verizon.com/about/news/verizon-test-force-braves...

And this is the interference hunting that has to happen now with devices certified. It would be an absolute hell if devices could just be introduced willynilly.

> Also, why not fine or ban products that cause problems rather than requiring certification.

Because certifying upfront is cheaper than trying to find products that cause problems in the field, do lots of testing and find somebody and recall all the other products that are in the field

It seems like fining or banning problems found problems 'in the wild' would require customer troubleshooting to find the cause, manufacturers would lose the product development cost for banned products, and the regulator(s) would have to staff for discovery in the field rather than in the lab. IMO, this seems like it could be more expensive than finding potential problems early.

> Also, why not fine or ban products that cause problems rather than requiring certification

Because people could get hurt or killed in the meantime. See the 737max for example.

Now it is a dramatic example, but think about any device unwillingly emitting too much electromagnetic radiation potentially being harmful to kids or messing with people pacemakers.

You could make similar arguments for basically any product, doesn't seem worth it to get every product certified though.

Also, didn't the 737max pass certifications?

Almost every product is certified in the EU (CE marking), although if you look closely it's self-certification and it's not always clear which sets of certification might be required for a product.

(I'm on the fence about this; personally I think there need to be far more small company / small production run exemptions from these requirements, but on the other hand I don't want a loophole for recklessly dangerous products)

> Almost every product is certified in the EU (CE marking), although if you look closely it's self-certification and it's not always clear which sets of certification might be required for a product.

Not to be confused with the Chinese Export marking: https://www.kimuagroup.com/news/differences-between-ce-and-c...

The amount of misinformation on the internet is worryingly high but not unexpected. No such thing as Chinese Export logo exist. It cannot officially exist in the EU since the CE sign is protected. Whatever this website is sharing is FUD and misinformation.

It doesn't mean that nobody fraudulently puts those markings on devices without a testing certificate backing it. However anybody who puts CE mark on a product without complying is risking being punished by EU member states. As an importer and distributer of such goods you'll be punished as well. CE certification is indeed self disclosed but it doesn't mean that you would get away with noncompliance.

Source: https://www.europarl.europa.eu/doceo/document/P-6-2007-5938-...

The "China Export" logo is extremely funny as it's basically a Chinese response to "you need a label? OK, fine, we'll give you a label" with none of the underlying bureaucracy.

The "UKCA" one, on the other hand, is real but tragic.

Worst of all: if you package certified CE product... you need to certify the whole package too!!! Think about using some certified CE PC in a cabinet for example...

On the other side: if you forget ventilation in the cabinet, the certified PC might burn... so packaging CE certified components doesn't mean that the whole is risk-free

> Worst of all: if you package certified CE product... you need to certify the whole package too!!!

Yes. This is a significant difference from the US, where FCC compliance testing for modules is much more reasonable.

The US also appears to contract out basic electrical safety to the insurance industry (Underwriter's Laboratories).

The NFPA writes the national electric code. They also came out of the insurance industry, and like UL and FM, predate adoption of their standards in regulation. FM is an actual insurance company but it is not OSHA. The history is similar to European organizations like the German TUeV, which came out of the boiler industry. All of them have their roots in disasters of the industrial revolution.

Governments today generally don't operate test laboratories, standards organizations, or certification bodies. DIN, ISO, IEC, BSI etc are not government organizations.

You're right, though, that there's more history of insurance industry involvement in developing safety standards and testing in the US.

There used to be. But they've been almost completely stripped away because of rampant abuse.

Using fraud as a counter example, where the self certifying manufacturer admitted to the crime before Congress, kinda seems like missing the point.

Ditto Dieselgate.

Doesn't negate the need for testing. Rather, it shows the need for effective oversight.

It's not expensive. What does a week of a good hardware engineer's time cost?

It's not even necessarily that expensive: $3K-$5K is a good budgetary range if you don't have any pre-compliance test and engineering capability because you might need 2-3 trips to a certified test lab to pass. If you have experience in your product space and good pre-compliance testing, you can definitely be one-and-done for less than that.

"We've just bought 50,000 new whatevers for our worldwide chain of stores... Oh, oops, they cause EM interference, throw them away I guess."

(More realistically, you'd just see extremely slow adoption in any important use-case.)

Reply on news.ycombinator.com