-
EMC compliance failure rate
Posted by
metebalci
on 22 Feb, 2024 06:46
-
In a video by R&S, it is said that EMC compliance failure rate is ~70-90%. Is this really the case ?
-
#1 Reply
Posted by
MarkT
on 22 Feb, 2024 09:26
-
Certainly believable - for a first test of a new bit of equipment.
-
#2 Reply
Posted by
Berni
on 22 Feb, 2024 11:21
-
This is why it is worth investing a bit into some pre compliance gear. Like a EMI tent, a spectrum analyzer, some EMI probes, LISN etc..
This way you can identify the worst issues by yourself and test various fixes for it, seeing just how effective the fixes are. Once it looks good in your own EMI test setup you go pay for the real deal test. It still doesn't guarantee that you will pass (Since replicating the full proper setup is too expensive), but you will avoid wasting time and money with testing products so bad in EMI that have no chance of passing.
-
#3 Reply
Posted by
metebalci
on 22 Feb, 2024 11:46
-
Investing in pre-compliance makes sense. I am still surprised about the number that it is higher than 50%, I wonder if it is mainly due to not doing pre-compliance testing, because the pre-compliance testing is not that much helpful or because there are many vendors doing EMC compliance testing for the first time.
-
#4 Reply
Posted by
selcuk
on 22 Feb, 2024 12:31
-
If you do precompliance tests at your office throughout the design phases, you will observe the problems before going to the test lab. I believe this failure rate is related to the lack of precompliance tests. Either you don't do these tests, or you can do some of them. You may have LISN, current probe, near field and far field antennas, rf amplifier, TEM cell but you may not have surge tester, ESD gun etc. Then you may fail at the surge test.
-
#5 Reply
Posted by
tszaboo
on 22 Feb, 2024 12:34
-
My personal rate is 33%. Though I'm not sure if I should count "Ethernet fails immunity at 100MHz exactly with industrial level of noise" as a failure.
-
#6 Reply
Posted by
T3sl4co1l
on 22 Feb, 2024 13:05
-
Sounds about right to me. It's either a very (trivially?) simple system, or one they (designers/manufacturer) really know how to handle EMC, that passes first try. My own record is, maybe, oh, 30%? Depending on how you count it. EMC is a complex topic, and clients want all manner of ridiculous arrangements (one I worked on, they wanted USB carried on unshielded cables! And expected it to pass 10V/m radiated and 2kV EFT!), few of which are designed in such a way that they are tractable. Complex cable harnesses and lack of ground planes are prime warning signs.
Tim
-
#7 Reply
Posted by
CaptDon
on 22 Feb, 2024 13:56
-
During my years at G.E. / WABTEC our failure rate was basically 100% fail on first try. Some of the problems come from the very tight specs in the (Apprx) 160MHz, 225MHz and 450MHz areas as these are railroad comms frequencies. The high speed FPGA's, CPLD's, DSP's and so on are very noisy in those areas. We found very strange sources as the noise source. One that we had to battle were isolators housed in DIP packages that had an internal R.F. oscillator which was used to couple power the isolated side. We used these on much of the I/O circuitry and they radiated like crazy. The data rates were very high so 'filtered' connectors were a 'no-go'. Some of our own in-house specs were very close to those used by the military for aircraft and submarines and are much harder to pass. SMPS power supplies were also a bit hard to tame. All manner of filtering was required such as common mode chokes, differential mode chokes, various grounding strategies within the device, not switching at one steady frequency but instead using discontinuous or broad spectrum switching which will 'appear' good in a normal testing manner but can reveal different results if you take the sweep over a period of an hour or so where you dwell on each frequency long enough to catch the 'real' noise level with an I.F. bandwidth which is wide enough to be 'fast'. This was important when we had a piece of gear that looked clean but yet produced a lot of hash noise when a person was trying to receive communications on a walkie talkie and standing near the unit. Then there are those times when your unit seems to pass all tests, but a guy with a V.H.F. 5 watt R.F. walkie talkie transmits with his antenna 2 inches from your wiring harness and everyone wonders why the E.C.U. did a safety shut down of the engine due to readings being way out of spec for the engine. Sometimes it is black magic!!
-
#8 Reply
Posted by
metebalci
on 23 Feb, 2024 06:36
-
Thanks all for the replies, very helpful.
-
#9 Reply
Posted by
Berni
on 23 Feb, 2024 06:54
-
Yep EMC is black magic.
You can't know for sure what is going to happen until you actually test it. Sometimes weird nonsensical solutions make the best EMC fixes. Yet the solution that magically fixed all EMC issues in one product might not work in another product.
So best you can do is DIY your own simplified poor mans version of the real test and see what happens. Even if the test is not accurate it will still detect the worst problems and give you some feedback if your attempted fix for the problem made it better or worse. This is the most important thing, so you know if you are going in the right direction.
Sure you might not have a ESD gun handy, but everyone can take apart a cigarette lighter and steal the piezzo igniter out of it to use as a ESD gun. Or even more low budget, find a sweater or rug that is particularly static prone, rub yourself on it to charge up and then poke the product. Etc... the test doesn't have to be accurate, as long as it is close enough to test for the same thing.
-
-
In a video by R&S, it is said that EMC compliance failure rate is ~70-90%. Is this really the case ?
Yes
I assume you're referring to my video "Understanding EMC - Precompliance"
That number is based on my own personal experience as well as feedback from my colleagues and our customers. Obviously, manufacturers and test labs don't publish their pass/fail percentages (as fascinating as that would be), but this is what we have seen and heard across the industry.
-
#11 Reply
Posted by
Psi
on 23 Feb, 2024 11:33
-
It very much depends on what you are making and how complex/large it is.
The bigger and more complex it is the higher the probability that something will cause an issue.
There is a big difference between doing EMC compliance on a cellphone verse on a little gadget that is pretty much just a MCU with built in RF output and a few switches and sensors. The phone is way more likely to fail first time
-
-
Investing in pre-compliance makes sense. I am still surprised about the number that it is higher than 50%, I wonder if it is mainly due to not doing pre-compliance testing, because the pre-compliance testing is not that much helpful or because there are many vendors doing EMC compliance testing for the first time.
One of the main reasons I made that video (which I presented at our Demystifying EMC 2023 event) was that a lot of companies often don't think about EMC compliance until (too) late in the design, development, or manufacturing process.
Compared to failing formal compliance testing and the costs of redesign, the cost of a conducted and/or radiated precompliance test setup is quite low, especially since you're not "paying per test" as you would in formal compliance testing. And since your own engineers are the ones running the tests, precompliance testing also helps to make EMC part of the design / debug cycle, rather than being an afterthought.
-
#13 Reply
Posted by
Psi
on 23 Feb, 2024 11:50
-
yeah, for the cost of the compliance testing you could usually buy 4 to 10 cheap spectrum analyzers.
So buying just 1 to do some checking before paying for the testing makes a lot of sense.
-
#14 Reply
Posted by
tszaboo
on 23 Feb, 2024 12:29
-
Investing in pre-compliance makes sense. I am still surprised about the number that it is higher than 50%, I wonder if it is mainly due to not doing pre-compliance testing, because the pre-compliance testing is not that much helpful or because there are many vendors doing EMC compliance testing for the first time.
One of the main reasons I made that video (which I presented at our Demystifying EMC 2023 event) was that a lot of companies often don't think about EMC compliance until (too) late in the design, development, or manufacturing process.
Compared to failing formal compliance testing and the costs of redesign, the cost of a conducted and/or radiated precompliance test setup is quite low, especially since you're not "paying per test" as you would in formal compliance testing. And since your own engineers are the ones running the tests, precompliance testing also helps to make EMC part of the design / debug cycle, rather than being an afterthought.
That's great. I set up a meeting with my manager, feel free to try to convince him.
-
#15 Reply
Posted by
metebalci
on 23 Feb, 2024 12:55
-
In a video by R&S, it is said that EMC compliance failure rate is ~70-90%. Is this really the case ?
Yes I assume you're referring to my video "Understanding EMC - Precompliance"
That number is based on my own personal experience as well as feedback from my colleagues and our customers. Obviously, manufacturers and test labs don't publish their pass/fail percentages (as fascinating as that would be), but this is what we have seen and heard across the industry.
Yes! I havent realized, I should have mentioned you.
-
#16 Reply
Posted by
CaptDon
on 23 Feb, 2024 13:40
-
The local test lab that we use was $1000 for a full day and that was the minimum. You could buy additional 1/2 days at $600 but you better be cleaned up and out by the end of your time slot! That was 2012. I seem to recall they are up to around $3000 per day now with no 1/2 day billing available. If no one was booked for the chamber the next day you could leave your stuff overnight and pick it up next day no additional charge.
I think the longest I was ever at their facility was four consecutive days. That was spent documenting the failed parameters, trying a few fixes while in the chamber and having to return weeks later for final testing and certification. Some of the sweeps can take 20 to 30 minutes each!!
-
-