EMC; 0dB, 3dB, 6dB or ? margin to conducted emission limits

So, we use a specified margin to the regulatory conducted emission limits.

For sake of discussion let's say 3dB. So during product development we stay at least 3dB below the limits specified in the relevant standard, CISPR 22 or equivalent.

The rationale for doing that is for taking into consideration repeatability of the measurement, tolerances of RFI components (CM inductors, Y capacitors etc) and cover my ass margin to avoid problems with design changes along the way.

But, if the end product is well below the limits, then it's money and project time out of the window

Also, we are the nice one in the bunch, working with safety margins etc. Then we sell the product to a consumer, a kW unit, and that is blown away by a Chinese 10W wall adaptor, with no EMC filter.

So, to the point of the post. What do you guys do? Go to the limit or adopt a safety margin also?

Cheers

Klaus

Reply to
Klaus Kragelund
Loading thread data ...

3db, but quasi-peak is a differnt measurement, so without the dedicated EMC SA I'd like to be more like 6db under. Plus the test environment and setup wil vary from site to site, so going to the limit is not such a good idea if lab expenses are of concern.

Cheers

Reply to
Martin Riddle

Design something that probably has margin. Send it to a test lab. If it passes by 0.1 dB, we're done.

Sometimes a friendly test lab will help a little. Move a cable or something.

--
John Larkin         Highland Technology, Inc 
picosecond timing   laser drivers and controllers 
 Click to see the full signature
Reply to
John Larkin

I generally try to get at least 3dB but have run out of chamber time before I got there. When I was doing such testing at independent labs I found a huge difference between the labs and their equipment. It turned out to be a *lot* cheaper to rent a 10m chamber from one company than the 3m from the other "local" place. The 10m chamber (and lab) gave us very consistent results. The other company (and 3m chamber) was like playing Whack-A-Mole for a few days.

As far as prophalactic mitigation, I use resistors and ferrites without thinking about the cost too much. Inductors and CM chokes need a lot of rationalization. If I don't already have a cap on an I/O, I'll be sure to leave pads so I can easily add them later.

Reply to
krw

It might depend also on whether you are directly selling a consumer product, or selling a sub-assembly that goes into something that your customer will have to take through EMC.

e.g. a certain SMPS supplier beginning with R sells a SMPS that is advertised on the datasheet as passing EMC standards, but if you request their test report, they will eventually begrudgingly admit that it only passed by

Reply to
Chris Jones

Or, go to a different test lab until you find one where the measurement has margin. Yes, you're allowed to do that

Reply to
RobertMacy

It's my understanding that 'margin' is kind of carry over from the VDE days. Where statistics were applied: If you test ONE unit, need 3 dB margin to enusre 'Product' compliance. Else, test at least THREE sample units and THEN you're allowed to be within ?? of the margin, but NONE can be over. However, [again, my understanding] that *if* a compliance body, like the FCC measured a sample unit; you were allowed one tone to be out, don't know by how much. The testing criteria was all whether you were 'demonstrating' Prdouct compliance as a Manufacturer, or you were a test lab that was 'verifying' compliance based upon a single unit, thus a tone out can be considered 'statistical distribution'.

Today, a Manufacturer only needs a 'passing' compliance report from an NRTL and they're done.

What do I do? It's worth it to do it right from the start. Then when you hit small, but necessary, Engineering changes, or go to version 2, no worries. At least 3dB margin, prefer 8 dB margins, and am ecstatic at the

15+dB marging, which usually get.
Reply to
RobertMacy

You are allowed to do that. But it will not help when somebody else tries to verify your data. ;-)

--
Reinhardt
Reply to
Reinhardt Behm

For normal commercial stuff that is it normally. I am into avionics. If there is interference I can happen that somebody questions your measurements and asks for verification.

--
Reinhardt
Reply to
Reinhardt Behm

Once a product passes EMI testing, you're done. File the compliance paperwork in a folder that nobody will ever look at, and buy a reel of CE stickers.

--
John Larkin         Highland Technology, Inc 
picosecond timing   laser drivers and controllers 
 Click to see the full signature
Reply to
John Larkin

We test with both Average and QP, so no problem there

Cheers

Klaus

Reply to
Klaus Kragelund

I was asking about conducted emissions, which are much more deterministic (at least below 1MHz) compared to radiated emissions, which change heavily with the setup

Cheers

Klaus

Reply to
Klaus Kragelund

The division I am in, is selling directly to the consumer

They are shaving off pennies for bigger profit margin, but are losing customers on the other hand. When we sell to sub assemblies we have much more margin

Our challenges is below 1MHz, where the cost of extra components to atthieve the margin matters

Cheers

Klaus

Reply to
Klaus Kragelund

For high volume products, 5 samples must be tested. That is relatively new. We used to do 3 samples only

Cheers

Klaus

Reply to
Klaus Kragelund

Ah, that's a lot easier. If it passes by a quarter of an inch, it passes. A you say, it's a lot more deterministic so there isn't any reason to "guard band" the tests to any significant degree.

The only issue I've had passing conducted emissions was when we didn't have a good enough ground on the line entry module. It took a little paint scraping and a star washer. Oh, and there was one time we had a trace routed all the way around a board before connecting to a cable that exited the case. That took some piggy-backed caps to solve and a re-spin of the board, later. I haven't made that mistake since.

Reply to
krw

Bad grounding and paint is a pain. We see that often, prototype screws and washers being used, that does not work correctly, and the room resonance increases dramatically

Cheers

Klaus

Reply to
Klaus Kragelund

Klaus Kragelund schreef op 05/09/2015 om 10:20 PM:

Complying with standards doesn't mean a product works well in every situation.

Reply to
N. Coesel

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.