How Important is EMC design

[harnon]

Hi guys,
I've been reading up a bit on this forum and the suggested reading at http://www.compliance-club.co.uk/ about designing your boards / selecting components etc to be EMC safe.  In particular the articles from Keith Armstrong are very detailed.  I am interested to understand the sort of hoops products jump through to get CE certification in the EU.

What I want to know is just how important this stuff really is? I'm talking about fairly straightforward, relatively low component count designs (i.e. lets say as an example Dave's uCurrent, or an Arduino Uno).  How serious are the EM emissions / interference likely to be for a product like this? 

Looking at the Arduino eagle files from http://arduino.cc/en/Main/ArduinoBoardUno they don't seem to have considered many of the measures that Armstrong suggests yet it is CE and FCC stamped.

Will

[jgbena]

Will,

I work with EMC stuff related to vehicles.  When it comes to EMC there are two sides to the coin.  Emissions, and susceptibility.  Whether you need to take this serious or not depends on the rules and reg's that you need to meet the standards that you are striving for.   Some are more stringent than others.

There is another good book by a fellow name Henry Ott called Electromagnetic Compatibility Engineering.  Its a good read for an EMC professional.

[jgbena]

Oh One more thing!

Also when it comes to EMC compliance, a lot of manufacturers state that their products have been tested and pass EMC requirements.. but they never tell you what standard, and how the tests were performed.  Remember also that when you deal with discreet devices, each in their own right may be in compliance, however its the integration of these items when developing a system that also needs to be tested.  When doing so its very helpful to have the Emission and Susceptibility of each device, what standard it was tested to, and see if you can get the data from them.  Most often you can, sometimes not.

Jon

[harnon]

Hey jgbena,
Thanks for the reply. That sounds interesting re the vehicles (I'm studying motorsport engineering).  Are there any particular parts that have the most EM issues on the vehicles? Funnily enough, on the electric vehicle that I work on all of the electronics that we made are on basic FR4 board or perf board with no flood fills or shielding and are in plastic ABS boxes.  Even the ones that are under the 200V+ of batteries and next to the motor and motor controller   We haven't had any issues yet but sounds like something to think about.

Back to the original question though, sounds like your answer to "does it matter" is ... "it depends!"

[Neilm]

I have to confess that every time the matter of EMC is raised on this forum, I point the poster to the articles by Keith Armstrong.

I have done quite a bit of work with EMC standards and having to meet same and I would recommend thinking about EMC before even starting a schematic.

When I joined the company I am with, I started by having to get a product through the compliance testing. Launch was delayed by months as EMC compliance was though of as something that has to be done at the end of the project - the result being it was incredibly noisy and the readings wandered all over the place.

I have just finished its replacement. As I worked on it through most of the project, I had a large input into the design and insisted on some of the measures Keith Armstrong advocates. The result was it sailed through all the EMC tests, with the readings hardly moving.

EMC compliance can be summed up as "do you feel lucky?" If you do then don't give it a thought and hope that when tested it passes. If you don't feel lucky, then analysis the circuit and decide what measures you can take. Doing this means that it will either be over engineered, or you will be aware of the pitfalls and possible issues so fixing them may be a bit quicker than otherwise. A third option is to provide a technical justification on why you don't need to test the product. This would generally only be done for something that is entirely passive, for example a resistor box.

Neil

[jgbena]

That sounds interesting re the vehicles (I'm studying motorsport engineering).

Funny you should say that.  I spent 10 years in professional motorsports designing SCCA Trans-Am, IMSA GTS-1 and GTP cars.  Then got into NASCAR Truck (which was not near as much fun as the others).

The answer is yes it should be done as preemptively as possible, that's just good design practice, but to what length depends on the product and how or where its used.  you should always strive to design for the environment.  Its always good to know what devices you have, and they can potentially radiate or be susceptible to.. that way when you get down to the debugging phase of the build, you can be ready and have a leg up on understanding what might be causing some goofy behavior in your product.

Jon

[jgbena]

EMC compliance can be summed up as "do you feel lucky?"

Neil!  Aint that the truth!!!

Jon

[gregariz]

Launch was delayed by months as EMC compliance was though of as something that has to be done at the end of the project.

I've seen this a few times. Often the reason EMC doesnt get factored in up front is because the project manager is not an electronics engineer but rather a PMP who doesnt know how to task or schedule it. Sometimes EMC tests are straightforward to get through. Other times they are horrific.

Its not always in the best interests of the engineers in any given project group to jump up and down about it. In my last job that had a positive outcome and meant that engineers were appointed to oversee the project manager on the next job.

[harnon]

Its not always in the best interests of the engineers in any given project group to jump up and down about it. In my last job that had a positive outcome and meant that engineers were appointed to oversee the project manager on the next job.

Gotta love that Ah, reminds me of getting really expensive consultants in to help us on a project and then having to train them to do the most basic tasks!

 

Funny you should say that.  I spent 10 years in professional motorsports designing SCCA Trans-Am, IMSA GTS-1 and GTP cars.  Then got into NASCAR Truck (which was not near as much fun as the others).

You don't say... whats your area of expertise? Next time you are in Oxford I'll buy you a beer and you can give me the low down!
Funnily enough about 1/4 of my course don't even really like cars... another 1/3 just like the technology (thats me) and the rest are petrol heads.  For the petrol heads the lunch time conversation is the same every day, basically a random combination of the phrases "bore and stroke", "how many horses?", "6 litre V12", "front wheel drive sucks", "Snetterton", "Lewis Hamilton" and "turbo".

[harnon]

Ok, thanks for all the replies.   I think the message is to design the PCB with EMC in mind as its much harder to rework later if your product isn't up to spec.  Its relatively easy/ inexpensive to select the right components and group and lay them out intelligently and it sounds like that would make a fair bit of difference. 

The separate Ground/V+ plane I suppose will depend on how much money I have in the pocket when prototyping    I don't think seeedstudio even do 4 layer boards and itead charge 4 times as much?!

[gregariz]

Its relatively easy/ inexpensive to select the right components and group and lay them out intelligently and it sounds like that would make a fair bit of difference. 

Thats the rub I think. You can go along to a high speed EMC course or pick up a book like Clayton Paul in order to get tips on pcb layout. You can usually incorporate some of the suggestions into any given design, but not all of them. The usual reason is that you simply don't have the product space to do it. Seperating grounds, pushing your high speed stuff to the end of the board and filtering your output lines cost real estate that a project manager may not give you. So you end up licking your finger and feeling the breeze to hope that the tips you did incorporate solve it.

[jgbena]

You don't say... whats your area of expertise? Next time you are in Oxford I'll buy you a beer and you can give me the low down!
Funnily enough about 1/4 of my course don't even really like cars... another 1/3 just like the technology (thats me) and the rest are petrol heads.  For the petrol heads the lunch time conversation is the same every day, basically a random combination of the phrases "bore and stroke", "how many horses?", "6 litre V12", "front wheel drive sucks", "Snetterton", "Lewis Hamilton" and "turbo".

Make it an Old Speckled Hen and you're on! How far is that from Milton Keyenes?  My sister lives there.
Well I wear a lot of hats!  Back then I was a Machinst/Welder/Fabricator/Designer/Engineer... I specialize is Chassis and Powertrain, and I also did harnessing and data Acquisition.  I still keep a cross functional role howver my job now is more electrical in nature (i work on megatrucks  very large mining haul trucks).
Yeah I had a roommate in college that used to get all mad and said"All you all talk about is numbers.. I had a 66 with a 396 4 barrel and a 391 12 bolt...." so one day he jumps up and says.. "OH YEAH??? well i got over the ram ram, with dual duals man!"  LOL

[Neilm]

The separate Ground/V+ plane I suppose will depend on how much money I have in the pocket when prototyping    I don't think seeedstudio even do 4 layer boards and itead charge 4 times as much?!

In that case shop around. I would not expect to pay more than about 25% more. I have used www.pcbtrain.co.uk for small runs of PCBs for prototypes for work.

Neil

[janekm]

The separate Ground/V+ plane I suppose will depend on how much money I have in the pocket when prototyping    I don't think seeedstudio even do 4 layer boards and itead charge 4 times as much?!

In that case shop around. I would not expect to pay more than about 25% more. I have used www.pcbtrain.co.uk for small runs of PCBs for prototypes for work.

Neil

Well they charge $10 for 10 2-layer boards so 4 times the price for 4-layer boards is not unreasonable, you'd be kinda hard-pressed to find anyone doing 4-layer for $12.50 even in China


---
I am here: http://maps.google.com/maps?ll=55.941477,-3.183925

[harnon]

The separate Ground/V+ plane I suppose will depend on how much money I have in the pocket when prototyping    I don't think seeedstudio even do 4 layer boards and itead charge 4 times as much?!

In that case shop around. I would not expect to pay more than about 25% more. I have used www.pcbtrain.co.uk for small runs of PCBs for prototypes for work.

Neil

I have done quite a bit of shopping around, and its hard to compete with Chinese prices!! PCB train 44GBP for 2 layer, 75GBP for 4 layer for one board excluding shipping, about 70% more.  (Or, I can get 10 two layer boards shipped from China for 30!)

[harnon]

Well they charge $10 for 10 2-layer boards so 4 times the price for 4-layer boards is not unreasonable, you'd be kinda hard-pressed to find anyone doing 4-layer for $12.50 even in China

Don't really expect $12 boards, but it would be nice if they were good quality

[harnon]

Just stumbled across this link on the TI Website via google - thought somebody else might find it useful...

PCB Design Guidelines for reduced EMI (PDF)

[Neilm]

Just stumbled across this link on the TI Website via google - thought somebody else might find it useful...

PCB Design Guidelines for reduced EMI (PDF)

That dates from the last millennium, so some of the information is now outdated. Chip dies have shrunk over the years so some of the design guidelines can actually cause problems as chips are now more susceptible. For more explanation, see the articles at http://www.compliance-club.co.uk.

A few years ago I had to get a load on instruments to pass a new version of standard. Following the guidance from the above site I not only got the vast majority of the instruments to pass - I also fixed a problem on one of the units that had been decided could not be fixed.

Neil

[gregariz]

Just stumbled across this link on the TI Website via google - thought somebody else might find it useful...

PCB Design Guidelines for reduced EMI (PDF)

Thanks. I'm not sure I'd route the caps under the crystal traces as on page 5 however I like the rest of the document.

[harnon]

Just stumbled across this link on the TI Website via google - thought somebody else might find it useful...

PCB Design Guidelines for reduced EMI (PDF)

Thanks. I'm not sure I'd route the caps under the crystal traces as on page 5 however I like the rest of the document.

Yeah, figured its a bit easier to get to grips with than the (excellent) compliance club resources!