Bring a product to market without certification?

[Mr. Scram]

Prototypes and custom units might be covered, although depending on the scenario might also be covered by “professional indemnity” insurance.

Of course doing your best is no guarantee - but I don’t consider doing mandated testing/certification as ‘gilding the lily’, and it’s probably a good idea to get a second opinion on anything that might sail close to the wind.

Certifying one offs or prototypes seems to qualify as gilding the lily. There's a whole class of aircraft who get a lot of leeway due to being experimental or one offs. Something similar seems absent when it comes to electronics. I don't see how prototypes and custom units would be covered if the law doesn't provide exceptions for them.

[fcb]

Prototypes and custom units might be covered, although depending on the scenario might also be covered by “professional indemnity” insurance.

Of course doing your best is no guarantee - but I don’t consider doing mandated testing/certification as ‘gilding the lily’, and it’s probably a good idea to get a second opinion on anything that might sail close to the wind.

Certifying one offs or prototypes seems to qualify as gilding the lily. There's a whole class of aircraft who get a lot of leeway due to being experimental or one offs. Something similar seems absent when it comes to electronics. I don't see how prototypes and custom units would be covered if the law doesn't provide exceptions for them.

I was referring to the liability element, not the certification of prototypes.

If I make a prototype for a client, and it burns down their facility - it would likely be covered by “product liability” - in that case I would have to show that the unit was in-compliance with any rules/regulations just as much as any series produced unit. You get round this to some extent on prototypes by making them sign a waiver.

If you crash your experimental aircraft into someone’s property you’d assume that your insurance would cover it - that’s why they have low weight limit.

Technically if I build a prototype device and it starts emitting RF out of ISM band limits - I am immediately in breach of the Wireless Telegraphy Act 1949. No ifs - no buts.

EDIT: Or it’s covered by my radio amateur license..

[skyjumper]

This is a great thread, I'm in the same position as OP, getting ready to launch a small run niche product and trying to understand the certification requirements. From reading, am I to understand that there are not in fact criminal penalties, at least in the US, for selling a device that lacks FCC certification?   Or in Canada?

I got a quote for a certification service, it was relatively low at about $10,000 US. IIRC, it was $1,800 for FCC, $600 for whatever Canada calls it and a whopping $7,500 for two different CE type certifications for Europe. I can swing North America, but wow, CE, wow! Not only is it a lot of money out of pocket for a label, These costs really add to the final price of the product.

[fcb]

With some exceptions (radio).

CE is just a declaration that your product meets the applicable standards.

You can ask (&pay!) for the 'opinion' of testing lab, include their report (more$$) in your TCF (technical construction file - might be called something else these days, but we still call it that).  But it is just that, it is their opinion (based on tests), that the one sample of your device has been tested to a certain standard and found to be in compliance.

So if your device was a battery flashlight for adults (as in bulb and alkaline cell) - you could easily skip EMC tests and argue (if it ever came to it) that it was incapable of generating EMI outside the limits.  You could even wave a near field RF probe over it and capture screen grab the spectrum analyser for your TCF.

But what happens if..
So lets say your competetior gets pissy with you and your flashlight, they can PAY for an EMC lab to test your product and write a report.  Then submit that report to who-knows as a reason to get your product banned from sale, repeat x26 countries and your flashlight might just end up on a list for customs to seize.

As far as I know this only used to happen between laptop maufacturers.

Customs are far more bothered about imported toys without secure battery compartments, or leaded paint, or toxic sofas.

https://www.poison.org/articles/button-batteries
https://www.cdc.gov/features/leadintoys/index.html
https://www.theguardian.com/commentisfree/2020/feb/26/toxic-sofa-eu-red-tape-flame-retardants

It's not unreasonable to use your FCC testing (with some maths) as evidence for your TCF.

[EEEnthusiast]

Many products without an intentional radiator (non wireless devices) can be released to the market using a SDoC (Supplier's  Declaration of Conformance). This process does not require the product to be certified in a lab or by a 3rd party. It requires the manufacturers to ensure that the product meets the requirement by means of measurement. If the manufacturer can ensure that the product is safe for EMC and other specifications, they may release the product to the market using the SDoC. The SDoC is valid for FCC, CE. I am not sure for other markets.

Some good sources
https://www.tele.soumu.go.jp/resource/j/equ/mra/pdf/30/e/03.pdf

[splin]

One aspect that I've never seen discussed is the impact of software/firmware on EMI emissions from a product. Specifically the need to revalidate emissions with any firmware updates. How often do you suppose that ever happens, even for the big players?

A firmware change could have a big, easily predictable impact if it added or modified a feature such as changing a USB I/F from lo-speed to full-speed, or slightly less obviously to increase GPIO port drive strength to improve reliability of a signal. In the first case you would almost certainly have to repeat the EMC testing; in the latter it would be a matter of judgement/risk tradeoffs.

But what about a 'minor' change such as a security or bug fix? A debugging/hacking technique that was ocaisionally used in the past was to place an AM radio close to the device. You could deduce certain patterns of behaviour from the interference caused - eg. a busy wait loop would be very different from complex code execution (so you could tell, for example, when a program was running out of control).

In most cases varying code execution patterns wouldn't  be an issue (unless the original certified product had very trivial code). But emissions can be particularly sensitive to peripheral usage on microcontrollers, especially clock related. Doubling an ADC clock rate for example, to improve analog resolution or for higher bandwidth could cause emission limits to be breached.

Problem is the programmer may have little understanding off these consequences so it would require a suitably skilled engineer to review the code changes to make the decision to re-test or take the risk of not doing so.

Another example; a purely software implementation of some security code is updated to something more secure using the microcontroller's inbuilt encryption unit. The engineer responsible for certification might not even be aware that the implementation had changed substantially, perhaps assuming that the hardware encryption peripheral had always been used.

Be honest - how often have you gone through an expensive EMC re-test following a firmware update?

[coppice]

Be honest - how often have you gone through an expensive EMC re-test following a firmware update?

Some firmware updates are specifically dishonest. Run firmware that turns on spectral spreading. Test. Change firmware to turn off spectral spreading. Ship.

[fcb]

Sometimes (the opposite of cheating) you have to run special FW for EMC testing.

Had to do this in the past for things like full 150KHz-30MHz sweep on a product (takes over 30 mins per line/neutral, conducted emissions with quasi-peak) - product shut off way before sweeps ended.

[NiHaoMike]

One aspect that I've never seen discussed is the impact of software/firmware on EMI emissions from a product. Specifically the need to revalidate emissions with any firmware updates. How often do you suppose that ever happens, even for the big players?

What about if the product is designed to run software or firmware developed by a third party or end user and the manufacturer has little or no control over it? For example, the Raspberry Pi is FCC approved, but it's pretty easy for the end user to program it to be a FM jammer.

[splin]

Be honest - how often have you gone through an expensive EMC re-test following a firmware update?

Some firmware updates are specifically dishonest. Run firmware that turns on spectral spreading. Test. Change firmware to turn off spectral spreading. Ship.

Of course that's possible, but why what would be the motivation? I guess it would apply to applications requiring as jitter free clocks as possible including ADCs, but will only pass testing with spread spectrum clocking enabled.

The only reason I could think of for what you suggest is because you need a certificate to sell your product into a particular market but believe the probability of being challenged for non-compliance subsequently is slim.

So do you require EMC certfication (for non-intentional radiators) for the US, Australian, Chinese etc. markets?

If not, why bother with testing? CE compliance doesn't, AFAIK, require certified EMC testing (for non RF devices at least) but you might want it for protection in the event of a complaint about your products emisions.

[wizard69]

If you are connecting to mains and switching or otherwise interacting with the mains, I wouldn't ship until you have certification, insurance and some sort of corporate protection.   Certification isn't enough.   The plant I work in requires UL certification on all (ALL AS IN EVERY THING) equipment connected to the mains.  That means every thing including a desk lamp   As for insurance & corporate identity, that is to protect you some bozzzo that kills himself doing something he shouldn't have done.

It is a good idea before going into "business" to realize what your exposures are.   Here in the USA we have the SBA offering up courses for people that want to go not business for themselves.   I actually took one of these seminars and frankly it turned me off to the idea of going into business for myself.    the legal landscape is that bad and you really need a product that can sustain a business out of the box or have a deep bank account.

[Psi]

Yes, if you are making/selling your own gear as a hobby business always buy and then resell a pre-certified DC power brick to your customers to power your low voltage product.
Never try to roll your own mains voltage PSU as part of your product.

And make sure the power brick is from a known good brand.
Go to their website and download/save all of the PSU's certification docs before you sell it. Have the docs on file in case you need them.
You don't want to use a certified power brick and then not be able to provide the documentation to prove it later on.

[skyjumper]

My product is an "intentional radiator" because it has WiFi and BT/BLE, all from a certified ESP32-WROOM module. I understand that using a certified component does not make my product certified, but I was thinking that it would make designing a product that will meet certification easier. Given it is an intentional radiator is there a mandatory testing requirement?

[ataradov]

You can transfer module certification to the product certification. But this works if you only have one intentional radiator. And I'm not sure what happens if the module itself contains two radiators. You may want to research that.

[skyjumper]

You can transfer module certification to the product certification. But this works if you only have one intentional radiator. And I'm not sure what happens if the module itself contains two radiators. You may want to research that.

Interesting... Well the ESP32-WROOM-32U uses a single hardware radio for both WiFi and BT, so I think it's fair to say thats only one radiator, unless there is law or regulation that says otherwise.

But what does that mean exactly, transfer certification?

[ataradov]

I'm not sure about the details. The concern there is that combined power may exceed limits even if single radios pass. So if ESP32 uses the same antenna and by design only one radio can operate at a time, it might be fine. But I would definitely do more research.

Yes, and the key point is that module must actually have modular certification. It is different from a final product certification.

[skyjumper]

I'm not sure about the details. The concern there is that combined power may exceed limits even if single radios pass. So if ESP32 uses the same antenna and by design only one radio can operate at a time, it might be fine. But I would definitely do more research.

Yes, and the key point is that module must actually have modular certification. It is different from a final product certification.

Yes just one antenna and, I think, either WiFi or BT at any given time. Thanks very much I'll investigate.

[palindrome]

If your product does not connect to mains and does not have any intentional radiators, you should be fine in most cases.

This is an excellent thread. I have been working on a product for a couple of years and I think I am at a stage where I could manufacture a small run.
My device does not have any intentional radiators. The problem I am facing is that my device is using an internal power supply - it is a certified PSU with a metal chassis.
I see that other niche products from the same segment use exactly the same power supply and I am fairly certain they did self certify their products (small volume and full certification costs wouldn't make it viable).
Is having such power supply going to get me in trouble (I am also using IEC inlet filter and I think all the wiring from inlet to PSU is properly done) or is it better to use an external power adapter?
I am planning to order a first batch of enclosures and it would be a pain if it turns out I should have used an external power adapter (patching IEC inlet cutout to fit XLR connector would look awful).
I am probably not going to sell more than 50-100 units.

What I have against external power adapter is cosmetic and customers would prefer to connect to mains and that I would have to buy a large volume of these power adapters (likely more than I would make my devices with) and they are more expensive than internal PSU (it is a renowned brand and has all certification).

[ataradov]

Is having such power supply going to get me in trouble (I am also using IEC inlet filter and I think all the wiring from inlet to PSU is properly done) or is it better to use an external power adapter?

Only lawyers in your jurisdiction would be able to tell for sure.

Anything having to do with mains is better safe than sorry, IMO.

[floobydust]

Just because a power supply module is "certified" doesn't mean the implementation ended up safe. Tell me about your AC mains fuse.
The power cord, strain relief, terminal blocks, fuse, grounding etc. - all must be approved components and used properly.
Are you using child labour for the soldering, an example of why an approved manufacturing facility is also something agencies require to guarantee the assembly is professionally done.

A common mistake is thinking a PSU module is certified, when cheap manufacturers will stretch it, i.e. in the little datasheet schematic it shows you must add EMI filtering and a fuse+MOV, and a certification agency will treat it just as a component that gets to go along with the formal product testing.
Even a wall adapter/brick a few don't actually have proper approvals, it passes one subset or EMC partially and that is all over the datasheet to mislead people. Always get a copy of their safety certificates and go over with a fine-toothed comb, to know if any loopholes were used.

It's unfair and terrible for the small guy that chinese products enter the country by the zillion having no approvals whatsoever, and we allow it to happen.
For small batches sold locally, I've used special inspection services where an inspector looks at each device and issues a sticker if it passes hi-pot and other requirements.

[nctnico]

If your product does not connect to mains and does not have any intentional radiators, you should be fine in most cases.

This is an excellent thread. I have been working on a product for a couple of years and I think I am at a stage where I could manufacture a small run.
My device does not have any intentional radiators. The problem I am facing is that my device is using an internal power supply - it is a certified PSU with a metal chassis.
I see that other niche products from the same segment use exactly the same power supply and I am fairly certain they did self certify their products (small volume and full certification costs wouldn't make it viable).
Is having such power supply going to get me in trouble (I am also using IEC inlet filter and I think all the wiring from inlet to PSU is properly done) or is it better to use an external power adapter?

Go for the external power adapter otherwise your device is connected to mains with all the safety requirements that come with it. I 100% agree with what floobydust wrote. 'Certified module' is an empty phrase.

[palindrome]

Thank you so much for help! I am going to use an external power adapter then.

Always get a copy of their safety certificates and go over with a fine-toothed comb, to know if any loopholes were used

Fortunately they sent me "paperwork" with the samples, so I will review it.

[Kerlin]

As some one who has spent the last eight years in international Avionics, believe me you don't want to ignore the RF interference requirements.
It HAS happened.
All aircraft instruments communicate and take many types of sensitive and exacting navigation measurements by using RF.
On frequencies from H.F. to Microwave.
They will fly over your equipment.


I will give this yet advice again, it was taught to me by a former employer and has worked for me, hundreds of times.

How they "over there" get their products approved and still manage to flood markets with rubbish is that they have their own world wide certified test houses.  I know I worked "over there" for years as an E.E., and sometimes still do projects there.
If you go about it the right way they can be very "co-operative" and don't charge any where near as much.
 USE them or definitely risk sending yourself broke one way or another.

[penfold]

Thank you so much for help! I am going to use an external power adapter then.
[...]

Alternatively, there are some PSU product families (at least from TDK Lambda... depends on your price point though) that, whilst intended for "internal" installation, have an integrated IEC socket intended to be butted up against the side of the enclosure quite similar to how a classic ATX PSU is installed. I only mention because an external PSU can be a bit of a minefield unto itself, though applicability may vary, just some things to think about: firstly (as with any solution) it must be appropriately rated for the useage environment (i.e. overvoltage category and industrial/commercial EMC) which may not necesarily be an easy find for industrial environments; secondly you typically lose the mains earth connection, given that most external units are isolated, and that may be useful in a broader systems EMC; and whilst you have still guarded yourself against the noisyness of the switcher, there is still possibility of radiation from common- and differential-mode noises, from your system on tthe DC side that will (with an external power adapter) be allowed outside the enclosure. Again, any of those points may or may not be applicable to you, mileage may vary... just something to think about.

[fcb]

Palindrome.

If you take your product to a testing lab and spend (£3-4K) on testing it and get a nice report (usually another £1K) all you really have is their OPINION that this unit, on this date, under these conditions met this standard, etc..

You can put this report in your TCF (technical construction file) or whatever they call it these days, and this will help you if you ever get hauled-over-the-coals.

The reality is that almost no one in the last 25years has been prosecuted.  There have been odd cases of laptop maker A paying for latop maker B's product to be tested and then whining to some authority to block their import, but nothing I have seen for many years. Authorities spend their time focused on firey escooter battery chargers, toys with accessable batteries, lead paint, choking hazards, etc..

If you fit a nice TDK PSU you are just as liable if it squirts all over the spectrum as if you fit a NONAME unit - however the reality is no one will really care and you are fairly safe. YMMV.

My main recommendation would be setup an LTD company (£13+ongoing accountacy!), get your company some product liability insurance - Hiscox for about £2-300/yr. depending on product.  Remember that regardless of paying an EMC lab or not - you are always self-certifing the product.