Way this works: You go into the chamber where there is a computerized setup that logs the whole chebang, conducted radiated horizontal/vertical. Out comes a list and then usually there are some close calls. If one or more were way above limits then you pack up and come back after some redesign. If none were above or just vy a smidgen everything gets trundled out to the open space setup. There's a hut with a big fat receiver in there and (hopefully) some hot coffee. The DUT gets placed on a rotary table outside. Now they take the list of close calls and check all those again.
This is because chamber measurements are not very precise.
Not 30 millihertz :-)
Goes from close to DC to a little above 3GHz.
It doesn't matter, you still have to pass regular (unintentional) EMC outside your intentional transmit band. That part doesn't get any easier. But why pay the extra chunk of money for field strength measurements on the ISM band or wherever you are transmitting, when you don't have to? That's like paying sales tax for the same item in two states.
Ok. Then why not just slap on the sticker, do only the regular EMC and be done with it? Of course the EMC lab needs to know about the intentional radiator part so they don't flag that.
Hmm. The chamber is the gold standard. A 3M isn't worth much, but we've had no trouble with the 10M. The rotary table and all that jazz is in there.
If you read the regs, at least for FHSS radiators in the ISM bands, anything outside the band, and outside one or two do-not-collect-$200 bands, must only be 20dB down from the in-band radiation. Easy-peasy, when you're radiating a hundred or two milliwatts.
Because the intentional limits are almost impossible to fail. We pass easily, now, so there isn't any point in gaming the system, but the rules leave the door wide open.
I didn't work on the anenna test range. It was on the flat roof of the factory, and quite hot except in the dead of winter. There was a tiny fiberglass shed for the unlucky to cool off between tests, but it was well over 100 degrees most of the time.
The only time I went up there was one night to see what a nasty job it was. That was around 8 PM, and it still took your breath.
Politicians should only get paid if the budget is balanced, and there is
enough left over to pay them.
That would be up to the EMC lab I guess. If they sign on the dotted line without open range measurements, well, dicey. The labs I went to never did that unless all the nasties were way under the limits.
Mine is the Icom R-1500. Has a mini console for quick checks when I don't want to fire up the PC. If you go PC-only then there's the similar PCR-1500 for about $100 less.
Nice thing is, you can listen to stuff in SSB, on every frequency. This is valuable beyond belief if you must knock a few dB off stuff that's already in the noise. With an analyzer that has only a display and maybe some rather crude AM detector with headphones your eyes will hurt at night, especially for guy like me who need a different set of glasses between 0603/0402 soldering and looking at screens. Civilian class A/B are easy. Aircraft category M, different thing.
I can crank up the big stereo speakers here and don't even have to don headphones. Of course, this means I can't listen to Bluegrass while doing EMC work. But the Shepherd doesn't like these funny gargles and warbles, leaving her pillow and walking out of the room, giving me "the look". She likes Bluegrass :-)
Huh? So you can be way above class B limits in, say, the aircraft bands? Do you have a link there? If you also know anything like that in the tax code I think lots of people would be all ears :-)
Sure, when you are a good designer or use pre-cooked modules they are easy. But why spend that extra money to test if it's already done?
My worst case was on my board but it wasn't my design (legacy crap). I guess I should have expected problems with the 100th thru 120th harmonic. ;-) I'm slowly convincing them that split ground planes are a bad idea, though (just found the source of some of our 2.4GHz "buzz" problem).
They've never even suggested open range measurements. The company with the 3m chamber couldn't get the same numbers twice, though.
I'll take a look at it. Thanks.
Evidently she doesn't like acid rock. ;-)
"outside one or two do-not-collect-$200 bands"
I meant to look it up again today but got busy doing real work. (Some nitwit decided to save $20 and leave off the solder mask on a proto board).
Because it makes passing a piece of cake. Can't fail! It probably wouldn't have worked for the product we had real trouble with because we needed CE (and all the rest of the alphabet soup, too)
Some of my stuff must be measured at 1m. At 3m you wouldn't be able to see much anymore, certainly not on a spectrum analyzer. At 10m there'd be nothing. But the standard is rather detailed about how the chamber must look like. For example, numerous antennas, several rotating stirrer plates, and so on.
An hour ago it did another fine job. Got three little transmitters here that are stuck. In order to diagnose this I had to see if the various data packets were still being transmitted, over all that din these things created. Fired up the Icom and sure enough, could hear a distinct faint "brrripp" whenever a packet was sent and the sound of it changed when I sent different packets. Very deep inside of what sounded like Niagara Falls from 10ft away. Now try that with an analyzer ...
Well, I bet you'd get in trouble no matter what if anything leaked. FM band, police, air, even TV. And if you step on some Lt.Colonel's radio link, oh boy ...
Oh great! He's earned himself a spot in the hall of blame I suppose :-)
Well, I know for sure the Europeans won't let you get away with that. But I can't imagine the FCC would either. Because then just about everybody could put in intentional radiator in their product whether it's needed or not, just to dodge the smog check.
?? If I can't see it at 3m or 10m, for that matter, why do I care? I let the compliance lab worry about their chamber. That's what they get paid the big bux for.
Why? It's an intentional radiator. It passes that test.
It's worse than that. The board is 2-layer (fortunately rather simple) with ground and Vcc pours top and bottom. It's almost impossible to solder anything without bridging. Getting solder to flow under the LEDs was a RPITA. "He" doesn't much care about blame, though it's not going to happen to me again.
That's what the regs say. I checked my interpretation with our test lab and it agrees with theirs. The problem is that it takes a few more hours to guarantee that it meets the "no go zone" edges. I stumbled onto this because the radio module does *not* meet the unintentional limits by itself. I turned it off and our box passes with 12dB to spare. If they can radiate like radium, why can't we? ;-) Seems all it takes is a redefinition of what the "intentional radiator" is.
The company that does most of our testing does initial testing in the chamber and if everything is 10dB under the limits, they will pass the product. If there are higher peaks, but still under the limit, they will re-measure those at the OATS (Open Area Test Site).
Another company seems to have a little more confidence in their chamber, but they still do OATS measurements.
The OATS is the gold standard, because that is how it is defined in the standards. A test company that passes your product when it is just under the limits in the chamber must have great confidence in that chamber.
If you use a pre-cooked module with a datasheet that says it meets certain limits, it will only meet those limits if you use it exactly as the manufacturer says (ask for those test conditions if they are not in the datasheet!). And even if you do that, as soon as you add more stuff you need to worry that the sum of emissions will not go over the limit. If you only use tested modules in prescribed ways, you may get away with adding the emissions and pass it on paper if that sum is well below limits.
For CE you are not required to do measurements, you just have to show somehow you are under the limits. But doing measurements is in most cases the only practical way to prove that.
Most test companies ask you to put the product in some mode that will produce (expected) worst case emissions. Just switching off emitters on your board for the test is just fooling yourself (and your test company).
If you are later caught in the field with too much emissions, you will have a hard time explaining why it was valid to turn off that emitter during your tests.
Stef (remove caps, dashes and .invalid from e-mail address to reply by mail)
You can't take damsel here now.
Does anyone know the status on the wirelessUSB modules from Cypress. I was planning to use them awhile back, but now I see that Cypress doesn't want us to use them in new designs. Shame, that is, since the modules were pre-certified and in two formats - 10 and 20 meters.
I don't see that the new technology 'CyFi' is modularized like the older wirelessUSB.
The two companies we've used rely on their chambers exclusively. Note that one of them I don't trust - can't get the same answer twice. The other calibrates all their chambers across the company periodically (better than monthly). Neither has ever suggested OATS, even when we were on the cusp.
Like I said, both have confidence in their chambers (even though I have no confidence in one of them).
That's not the issue.
Again, that's not the issue. The FCC has some funky rules for some intentional radiators that I'm sure CE doesn't have.
That's what they demanded. Otherwise it was an intentional radiator.
Because the emitter is over the line for an unintended radiator.
Turns out that our "pre-compliance" runs were worthless. The results couldn't be duplicated.
As long as it's 20dB below the fundamental, apparently the local sheriff can go scratch.
I had a little time today (emphasis on "little"):
"15.215 Additional provisions to the general radiated emission limitations.
(c) Intentional radiators operating under the alternative provisions to the general emission limits, as contained in §§15.217 through 15.257 and in Subpart E of this part, must be designed to ensure that the 20 dB bandwidth of the emission, or whatever bandwidth may otherwise be specified in the specific rule section under which the equipment operates, is contained within the frequency band designated in the rule section under which the equipment is operated. The requirement to contain the designated bandwidth of the emission within the specified frequency band includes the effects from frequency sweeping, frequency hopping and other modulation techniques that may be employed as well as the frequency stability of the transmitter over expected variations in temperature and supply voltage. If a frequency stability is not specified in the regulations, it is recommended that the fundamental emission be kept within at least the central 80% of the permitted band in order to minimize the possibility of out-of-band operation."
Both test labs (the module manufacturer likes one, we prefer the other) interpret this the same way; that the intentional radiator has to be 20dB down outside its band.
Now, this (prior) paragraph contradicts this...
"(b) In most cases, unwanted emissions outside of the frequency bands shown in these alternative provisions must be attenuated to the emission limits shown in §15.209. In no case shall the level of the unwanted emissions from an intentional radiator operating under these additional provisions exceed the field strength of the fundamental emission.
..except for the "most cases". The interpretation from both labs is that (c) overrules (b). Harmonics from the hopping are certainly above 12.209.
Our box, without the radiator passes 12.209 (below for information) but the radio does not, outside its band. The radio has a separate cert.
In the restricted zones, yes. Outside of those zones it apparently does *NOT* have to meet class A/B. They looked at the spectrum and saw one place where we were close. They made sure that pup was outside the (frequency) window and all was goodness. It *was* over the Class-A line but would have passed (and actually did when the module had its cert done).
No, out-of-band must be 20dB below the in-band. ...didn't make any sense to me, either.
Strange. They should have shown you the heavy hitters. Mine worked, every single time.
Below the one watt (!) carrier of your legit 900MHz TX module? That would be 10mW on a police band. Unless I see that in writing (meaning the law) I cannot believe this.
That's indeed remarkable. Seems you can let 20mW spill over past 902MHz or 928MHz. But, reading on ...
Aha! That means no free lunch after all.
It doesn't matter what the lab interprets. If the sheriff has this brought to court and you lose, a major recall may follow because the judge said so. Possibly wiping out the EMC lab and shaking up the place where you work.
If the radio doesn't pass I'd be rather concerned.
Hopefully they are right ...
Way I read the law above it ain't so. Not across the whole frequency range.