Inexpensive alternative to UL/NRTL certitication

[SlowBro]

Planning my first product, which includes a Wifi-enabled device plugged into mains power via a USB charger, and I'm encountering a roadblock. UL or NRTL certification looks PRICEY!

Yet I don't want my USB-charged device causing a fire and me getting sued for everything. Rock and a hard place.

What are some good alternatives? Or do I just need to bite the bullet?


(Now I understand why "I can make that for 1/10th the cost, why are they charging so much?" is a stupid question.)

[Edit] Bit of clarification:

• Yep I do intend to talk to a lawyer and/or product design lab.
• No batteries would be used. The USB charger only acts as a power supply.
• The Wifi module (ESP8266 ESP-07S) is already FCC single modular approved. I just need verification for my device, which amounts to pretty much the same kind of testing done for unintentional radiators.
• Total power used is expected to be around half a watt.
• The USB wall wart would be pre-certified by UL or similar.

[SlowBro]

Maybe, if the USB charger is UL-listed but the device is not... would that stand up in court or does the whole device need safety testing to pass muster?

[JPortici]

find a lawyer/firm/engineer study place that does that stuff in your area and ask for a consult

[SlowBro]

Sure, I'll do that too.

Anyone have experience with this? Not looking for precise advice, just looking to hear what others have experienced, to get ideas.

[amc184]

Get professional advice, listing a device is no joke. The initial testing is expensive and time consuming, and there are continued compliance costs during the life of the product (including follow up inspections).

One way to reduce the budget is to use a lower cost NRTL.  UL and CSA are the most expensive, while others like ETL and MET are cheaper (allegedly half the cost and time, but I've not worked with both).

[SlowBro]

Thanks everyone. Bit of clarification:

• Yep I do intend to talk to a lawyer and/or product design lab.
• No batteries would be used. The USB charger only acts as a power supply.
• The Wifi module (ESP8266 ESP-07S) is already FCC single modular approved. I just need verification for my device, which amounts to pretty much the same kind of testing done for unintentional radiators.
• Total power used is expected to be around half a watt.

[SlowBro]

Part 3 is inevitable, unless your MCU operated at less than 9kHz, which is impossible. For computer connected device (peripheral), you MUST get a certificate from an FCC accredited certification center. For a stand alone device, which is true for this case, you can do the test in house without having it certified by a third party, this is called verification. However, if anyone report your device to be emitting harmful EMI, you need to cease selling them immediately. Keep in mind verification still requires you to submit your test result to FCC and acquire an FCC number, but this process is way cheaper than getting a proper certificate yet perfectly legal.

Pardon me for slightly disagreeing with someone who's got more experience in this, but I found the following on the FCC website. I just wanted to leave this note for the sake of others who may be reading this.
-------
VERIFICATION (47 CFR Section 2.902)
Verification is a procedure that requires the party responsible for compliance to rely on measurements that it or another party makes on its behalf to ensure that the equipment complies with the appropriate technical standards. The responsible party is not required to use an accredited testing laboratory. It is not required to file an application with a TCB, and equipment authorized under the verification procedure is not listed in any Commission database. However, the responsible party must provide a test report and other information demonstrating compliance with the rules upon request by the Commission.
-------
https://www.fcc.gov/general/equipment-authorization-procedures

In other words, you are not required to submit test results to the FCC or acquire an FCC number. And the lab need not even be accredited; just required to use a recognized lab, which may be less expensive. Then keep your results in a file in case you're ever asked for them.

[SlowBro]

One way to reduce the budget is to use a lower cost NRTL.  UL and CSA are the most expensive, while others like ETL and MET are cheaper (allegedly half the cost and time, but I've not worked with both).

Funny you mentioned that, I put ETL testing in my budget sheet last night, assuming that's what I'd have to do. $1,080 initially and $910 every quarter since I'd be imprinting their mark on my devices and just have the quarterly $150 labelling fee.

Don't know though if there are any other fees or expenses I ought to take into account for ETL. I'd be all ears if someone knows, and I'm going to be talking to some local experts about it soon.

http://www.intertek.com/marks/etl/guide/fees/

[SlowBro]

Pardon me for slightly disagreeing with someone who's got more experience in this, but I found the following on the FCC website. I just wanted to leave this note for the sake of others who may be reading this.

In other words, you are not required to submit test results to the FCC or acquire an FCC number. And the lab need not even be accredited; just required to use a recognized lab, which may be less expensive. Then keep your results in a file in case you're ever asked for them.

My bad. Remembered the wrong thing. You're correct, you only need to maintain your copy of test report.

I wouldn't have said anything but I know others may come along later with the same questions and read this 

[SlowBro]

Be sure what certifications your ESP8266 module really has though as a pre-certified RF module.
I've heard some skepticism about the way some of the modules really having any / all certifications but that was in the early days, there are probably reputable suppliers now.

Refresh your page, I anticipated that comment and added a link to the FCC cert for the device. Shows single modular 

An FCC and CE certified device, WPA2 802.11n Wifi capable, with a microcontroller at 80MHz with 4MB flash for $2.14 in quantities of 100. What a world we live in.

Your comments are good. I may be overthinking this out of fear, uncertainty, and doubt.

[SlowBro]

Sent an email to Harri and Ai-Thinker's technical support about that last night. They're supposed to supply those details on request but I wonder how well they respond to requests...

From the docs I gather it's a 3dBi antenna, 50 ohm, and looks like one of those with an adjustable angle like on the back of an older Linksys router. But that's all I can tell. Don't know about lead length either.

The way you wrote that it sounds like you know Harri? Does he post here?

[SlowBro]

It's a small world after all, it's a small world after all...

[SlowBro]

I read somewhere that if I get my device built by a UL-certified plant I can get reduced or eliminate the certification fees; is that correct?

Probably have to leave the design and everything up to them but that may be a less expensive option.

[sparkswillfly]

I don't get it... your device is a powered by a USB power supply?  So just ship a UL certified one in the box.  Only the power supply needs to be UL certified not the whole device.

[SlowBro]

I don't get it... your device is a powered by a USB power supply?  So just ship a UL certified one in the box.  Only the power supply needs to be UL certified not the whole device.

Really nothing needs to be UL certified. It's not a legal requirement, it's just a wise practice. And yes, the USB wall wart would be UL or similar.

[SlowBro]

For fire concerns the fact that the PSU should be limited energy and so low current / voltage alone should be sufficient and if you use UL94 (e.v. 94V0) rated PCB material and for the majority of components then there's little to worry about except the enclosure which is fine if metal or if plastic then you can figure out if any kind of flame retardant resin is even applicable to your design or not but really there shouldn't be the energy around to be a source of ignition so AFAICT it should be irrelevant to any safety requirements that I know of for IT purposes.

There could be some tips about avoiding tantalum capacitors and so on that could help for reliability and you can put a fuse in there rated at 500mA or whatever is appropriate.  Ferrite beads and series resistors and such of course can help with EMC.

So build it fire-safe to keep families safe and pass inspection. The goal I now have is to get the design so fire-safe that it passes with flying colors with minimal or no modification perhaps hopefully on the first try. In my spreadsheet I've already switched the case to an aluminum one and will be adding a fuse, will demand from my PCB supplier the right materials for the board, and so forth.

I searched Google (several layers deep) for a guide for developing electronics so that they meet UL standards but couldn't find any. The UL standards doc itself is over a thousand dollars. I found some illegal copies shared online but I'm not going that route.

Anyone know of a guide or book that explains these things?

[SlowBro]

I am not a compliance engineer / lawyer, so take my thoughts for what they're worth.

I have been involved in designing a few commercial and consumer oriented electronic products and only in special cases (medical applications, special types of equipment that do have UL related fire/smoke safety requirements mandated by electrical codes and compulsory regulations in many places) has UL fire safety compliance testing and certification been considered relevant / necessary.

Yeah, as mentioned I may be asking out of FUD. These are valuable points, so I can be more intelligent when I work with engineers/lawyers. I've already got some quote requests out there at some labs so I'll digest your information and go work with them for the next steps.

Thanks much.

And yes I know this is not legal advice. I hardly expect "But your honor, some guy on a forum told me.." to be a valid legal defense 

[SlowBro]

So I thought I'd give an update about the antenna type. I reached out to support@aithinker.com and after some delay I did get a response. They sell the exact antenna plus coax used in the FCC certification so I can copy the same design for my product. At a good price, too.

If you're trying to do the same, get in touch with them about it.

[anfang]

Yet I don't want my USB-charged device causing a fire and me getting sued for everything. Rock and a hard place.

==LEGAL==
Whether or not you get sued has zero bearing on your certifications. I can promise you there are plenty of ladder makers that have followed all the rules and they still get sued. That is what an LLC is for. Set up a company. It's quick and easy. And if someone sues the company, then you are insulated personally (as long as you are following certain rules--those are the things to get educated about).

==SAFETY==
The risk of fire from a 5V/1A wall wart that has UL cert is very small. Maybe you'd see something strange if you built 1M products. But if you build 100, the odds are on your side.  If you want to convince yourself, get a 5V/1A wall wart and connect it to the 5 ohm 5W resistor. And then try to light tissue paper on fire with the resistor. You can't. And that's not even in a case! For the case, use metal case or flame retardant "drip proof" plastic. None of your electronic components (or PCB) will blink at being exposed to 250C. If you have a larger case, the problem is even easier (less watts per m2).

Do consider what might happen if the power supply fails. A PTC is cheap at the entry port of your product. It has some voltage drop, but if things get hot it will open and reset as they cool down. It's a great second line of defense.

== PCB ==
Layout out your PCB so that you have a solid ground plane on the bottom. Looks at standard shields from Digikey. If you are the paranoid type, include the ability to use a shield in your design if needed. if you have a solid ground plane on the bottom and the parts with clocks are under a shield, you will sail through EMC without any issues.

==EMISSIONS==
Buy yourself a spectrum analyzer. There are kits that let you "pre-cert" so that you know if you if you have problems before you get to the test lab. These include antennas for a variety of frequencies. But more important, the makers explain the steps you need to take. Even though you don't have the chamber, they help you understand how to work around that (the best answer is head out to the sticks where signals are down in the noise). If you can pre-cert yourself, you go into the compliance lab and sail through without any surprises. The Signal Hound folks have a lot of info here, and their USB Spectrum Analyzers (both low end and high end) are far superior to desktop units.

==SUSCEPTIBILITY==
Things like ESD and proper operation in the presence of interferers are important in Europe, but the US doesn't worry too much about it (FCC does for cell phones, but not for wifi). Use standard parts to ensure ESD is clamped at your input ports. For your own warranty concerns, ensure your product can take a zap on any port with an ESD gun.

== Reliability==
Get yourself a temperature chamber. Set it to 80 degrees (assuming you are making a product for consumers). Run your product continuously at that temperature. If you've missed a voltage spec on something, this will find it. Silicon (and passives) that are overstressed will fail very quickly (hours) at elevated temperatures. It depends on your product, but roughly for a product designed for room temp operation, running it at 80C for 3 days will simulate 6 months of continuous operation (depends on your precise product though). You need to study this in more detail, but the search terms are HTOL and arrenius.

== Mechanical==
Take 10 of your prototypes. Drop them over and over from a height of 1m in different orientations onto concrete. You will see some common failures. Be on the lookout for cracked parts on PCB (too much board flex, not enough support/screws/etc), solder joints that fail, etc. Chances are if the case is off the shelf and plastic, it is thick enough you won't have any problems. But do the drops UNTIL you see a failure. Your aim here is to find failures. If on 2-3 units you see the same failure after 2-3 drops, be very concerned. If you get to 20 drops on 10 units (200 drops total) from 1m without any failures, you should feel pretty good.

==US==
US will be fine just ensuring the FCC is happy.

==EU==
If you want a distributor in EU, you'll need EU compliance and as others have mentioned, there are a myriad of specs. Many of these you can simply attest and self-cert. For emissions, you'll need a test house to sign off (so request EU and US emissions at the same time). Find a product that is similar to your from a big company. They will specify what IEC specs they have passed, and what sections were relevant for that product. Do the same. Usually, this paperwork is called "declaration of conformity". If it's not in the owner's manual, it's often on their website.

If you just want to sell to the EU from the US, just put them up for sales and ship them when the order comes. Even without CE you'll see 99% of them make it through to the customer. 

The above is just friendly advice. Verify it all yourself.

[l0wside]

Some comments from a Europaean perspective (are you working in Austria, are you Austrian, or both?). I largely share your views, with a few exceptions.

== PCB ==
Layout out your PCB so that you have a solid ground plane on the bottom. Looks at standard shields from Digikey. If you are the paranoid type, include the ability to use a shield in your design if needed. if you have a solid ground plane on the bottom and the parts with clocks are under a shield, you will sail through EMC without any issues.

Disagree. A ground plane is a good idea to shield bus signals, but a ground plane below a DC/DC inductor can make an antenna, ruining your RE performance. Look at your current paths and try to minimize loops.
A spectrum analyzer is a helpful tool, but only one. For RE, you need a proper antenna and room, or a TEM cell. For CE, a LISN. Neither is cheap. DIY solutions for a LISN (and for TEM cells) exist, but likely you will not be able to calibrate them even roughly. Pre-compliance in a lab is not extremely expensive, ask around a bit.

==EU==
If you want a distributor in EU, you'll need EU compliance and as others have mentioned, there are a myriad of specs. Many of these you can simply attest and self-cert. For emissions, you'll need a test house to sign off (so request EU and US emissions at the same time). Find a product that is similar to your from a big company. They will specify what IEC specs they have passed, and what sections were relevant for that product. Do the same. Usually, this paperwork is called "declaration of conformity". If it's not in the owner's manual, it's often on their website.

Except for wireless devices, CE is all about self-certification. From the US, the pragmatic way is to simply slap a CE sticker on the device and ship it. The probability of being sued by some European authority is effectively zero.

[anfang]

A ground plane is a good idea to shield bus signals, but a ground plane below a DC/DC inductor can make an antenna, ruining your RE performance.

TI disagrees with you. See AN2155 "Layout tips for EMI reduction in DC/DC converters" and they experimentally demonstrate that a ground plane reduces EMI and that that a ground plane is "High" importance in reducing EMI. See section 4 for the summary. There are no caveats here.

If you use pre-compliance SW and tools on spectrum analyzers, along with a range of antennas, you can learn very quickly how bad (or good) your board is. If you go 50 km from a city and nothing is around you for 1-2 km and you've turned off all nearby emitters, then the spectrum you see on an antenna will be very clean. You mark what you do measure on the spectrum, and then you turn on your device you've made. And then you look for changes. It is is the changes you are concerned about. And you will learn loads about your design doing this, especially what makes it better or worse.

The investment is pretty low here: A USB spectrum analyzer (real time is important) and a range of uncalibrated near field probes, log periodic antennas, etc. These are all fairly cheap as they are uncalibrated. And again, you aren't looking for absolute. That is what the test house if for. You are only looking for relative. "Does this change make my result better or worse?" is the question you are trying to answer.

And once you've got that down, as I said, you can enter a proper test agency with great confidence.

Tek has a good paper on this. They are doing pre-compliance in a basement. http://www.tek.com/dl/37A_60141_1_HR_Letter.pdf

BTW, I'm in the US

[SlowBro]

Excellent thoughts here for low-cost certification in general. If others encounter this thread some day they may find it good info.

For my particular application though I only need FCC Verification which will be about $1000 to test, which makes buying a $2500 device a bit much for one or two uses. Would be nice if I could rent some of this equipment, and maybe I can.

anfang I like your idea of going out into the wilderness to do a check. Is something like this what you had in mind?

http://www.ebay.com/itm/PC-USB-2CH-20M-1MB-CH-Digital-Oscilloscope-2CH-Spectrum-analyzer-Data-Recorder-/141105887222?hash=item20da90f7f6:g:LgkAAOxydB1SiDDN

[SlowBro]

Could you please point to some inexpensive devices which do?

[SlowBro]

I'm facing a ~$1000 test (FCC Verification) so I think I'll pass buying that equipment. ($919 just for the 4.4GHz analyzer.) I do have an SDR dongle so if that'd get me 80% of the way I'll try it.

[anfang]

Is something like this what you had in mind?

No, not even close. You need something with great sensitivity and very fast response. The Tek and Signal Hound both digitize massive swaths of data at a time and crunch on that. Both are around $3500.

I'm facing a ~$1000 test (FCC Verification) so I think I'll pass buying that equipment. ($919 just for the 4.4GHz analyzer.) I do have an SDR dongle so if that'd get me 80% of the way I'll try it.

The SDR dongle won't come close to doing what you need to do I suspect.

Consider this scenario: You go to the test house with a marginal board, it fails, you need another 4-6 weeks of new PCB, new build, new verification, and then you come back again. Most test houses will charge you AGAIN (with maybe some discount). But remember, if your first board passed tests 1, 2, 3, 6, and 10 and failed 4, 5, 7, 8, 9, then the board house has to test everything again. The fact that you previously passed some test means nothing to them. So, they will make a strong case for charging you more (almost 2X more).

This is why I said earlier that if you did your board such that the bottom layer was a full plane and the top was shielded, then your radiators are all shielded and in that case you can go into testing with great confidence. And you can send them a board without a shield and see if that passes. In both cases, you'll leave the test with something that passes on first shot. In that case, you can get by without any pre-cert if you are diligent.

Some things that seem expensive are actually pretty cheap when the larger picture is considered.

PS. What is the highest frequency on the board (not including any modules)? If you are 0 to 100 MHz, you'll very likely easily pass with just a ground plane if you do a competent design.

[SlowBro]

This is why I said earlier that if you did your board such that the bottom layer was a full plane and the top was shielded...

Don't misunderstand, I'm taking careful note of every single design recommendation here. I want to pass first time. Retries are pricey.

PS. What is the highest frequency on the board (not including any modules)? If you are 0 to 100 MHz, you'll very likely easily pass with just a ground plane if you do a competent design.

It includes a Wi-fi module. Model number and FCC certification on the first page.


Great info here and I'm all too happy to get it.