On Thursday, Oct. 16, Interference Technology hosted a roundtable during EMC Live - 'Elephants in the Test Room' based on our blog series. Below is additional commentary about testing issues, from our expert panel.
Tom Mullineaux - Moderator, Consultant and Author
Patrick Andre - President, André Consulting, Inc.
Fin O'Connor - Defense and Space Consultant and Contractor, Alion Science and Technology
Adiseshu Nyshadham - Senior Consultant, DVT Solutions Inc.
Steve Koster - Vice President, Washington Labs
Elephant Discussion #1 – Poor EMC Measurement Consistency
No one is surprised when a round robin test shows multiple EMC testhouse measurements taken under supposedly identical test conditions are up to 10dB apart. By its nature, the ISO17025 laboratory accreditation standard covers a very broad church of test situations. However, the EMC industry is a distinct, identifiable niche. The various compliance groups providing the audit-service should be able to work together to improve inter-EMC laboratory measurement accuracy.
Can the EMC industry work with ISO17025 laboratory accreditation teams to improve measurement consistency?
Elephant Discussion #2 – Underperforming EMC Chambers
When calibrating a test field to 6GHz for commercial RF immunity testing, to obtain a compliant test field many test houses are finding they are forced to point the antenna at one corner of the room. All standard 3 meter semi-anechoic test chambers are cuboids with flat walls, ceiling and floor. The four walls and the ceiling are clad in RF absorber. The ‘hot’ wall (the one behind the calibration plane) performs the same as the other three walls.
Is a flat absorber lined ‘hot’ wall truly the only possibility?
Elephant Discussion #3 – Automotive Tests that Put the Car Audio System Performance First
Until fairly recently it seemed the RF immunity tests conducted inside the car cabin space were purely to ensure good sound system performance.
What is the future of RF immunity tests that actually check for electronic sub assembly compatibility inside the cabin space of a car?
Q. Can the EMC industry work with ISO17025 laboratory accreditation teams to improve measurement consistency?
A. Yes, improvements can always be made. However the challenge will be to first establish repeatability in a single test setup for each chamber. Often, changes of ±4 dB is seen with the most minor of test setup changes. The movement of a cable or the position of the support equipment can radically alter the results. Added to this the difficulty of different laboratory layouts, with different sized rooms, anechoic material differences, antenna configurations (both type and location), power line and signal line routings into and out of the room, to name a few, and soon you have so many variables between two so-called “identical setups” that getting only 10 dB differences is actually remarkably good.
Q. It might be necessary to introduce the use of a fully anechoic chamber where the floor is now covered with absorber, reducing at least one variable in the measurements.
A. All standard test chambers are cuboids with flat walls, ceiling and floor. Is this shape truly the only possibility?
Of course not. Reverberant chambers are already being made with non-parallel walls. Also, simply adding a metal or anechoic panel at a 45 degree angle in a corner can alter the fields significantly. However, I am somewhat surprised by the difficulties the labs have, since the cone style of anechoic material should work remarkably well at these frequencies.
Boy, the #1 could go on for days!! I’m on the 461 committee and we will not have any uncertainties in the spec (ever as some say) the uncertainties are built into the limits. Now the test problem differences are very difficult to get a handle on. I travel to many independent and gov labs witnessing testing for NAVAIR and the differences in the way these test houses INTERPRET the standard is surprising. I am trying to get more description for each test into the ‘G’ version of the standard (which we are working on now) but I am getting resistance because they feel the standard has enough description for the competent engineer/tech. I was also involved, years ago, with the round robin testing for the NVLAP Mutual recognition project and this was a for the very simple open filed testing with specific parameters. And even with this test control, the differences between the test houses was 10 + db at least. I would love to see some way to get the testing differences smaller so anything that can be suggested and proven would be welcome but it’s a uphill road.
I think Fin and I are in violent agreement. As a fan of MIL-STD 461 and DO-160 (mainly due to my aerospace background), I love the fact that cable length and positions are defined and controlled (1 meter, 2 meters, 3.3 meters, and so forth, depending on the test and standard). Cables are placed on 5 cm standoffs, 10 cm from the front of the ground plane. And even then, even with all this control, we get these variations.
I am working on a piece of medical equipment, with five transducers, five monitors, plus peripheral and support equipment. How do I ever lay out the cables and support equipment the same so I can get the same results twice? I have seen 5 dB swings just moving one cable.
I think the only science less accurate than ours is Astronomy, where they state distances of remote objects within a couple decades, e.g. 10^6 to 10^8 light-years away. And I wanted to be an Astronomer.
Then again, I have had no success convincing the police officer of the accuracy of my speedometer. He just doesn’t buy the idea that as long as I am between 6 MPH and 600 MPH I am within the range of accuracy.
Just to make sure you are up to speed with where the elephants are history wise, you should read the article I wrote recently for ITEM.
In the article I am very critical of the design of the emissions test fixture. I have just received 3D EM software from CST so I can come up with a superior arrangement. Might have preliminary results in time for our webinar. But you should know that if you praise the 5cm high cable over a ground plane I am likely to lambast you without mercy, and have supporting data to do it with.
Oh no, not a 5 cm debate. I’m always interested in why the specs are the way they are. I will be real interested in your research, and will pass on to the committee,if it will make the readings better. But being honest, we have been using this method for the longest and our planes fly, missiles find there mark and radios work, in theater/hostile environments and we have feedback from the fleet and we track problems and why they occur (NAVAIR has the ASEMICAP group which tracks fleet problems and finds the reason and solutions) so to get the community to change, you will have to have a really good reason. But sometimes change is good!! Just hard to get through.
Yes, the long history and the fact that planes are not falling out of the sky speaks volumes. Hard to argue with physics though, particularly the possible mismatch between the noise source (EUT) and the test fixture.
Tom, ‘test fixture’?? Not sure what you mean? The 5 cm is to closely resemble the common actual installation (Military environment, loop impedances), to get some consistency and to work with the LISN (below 10 MHz). Unfortunately though, as Patrick had, most of the time, the set-up is a mess and standardization is mostly impossible so these controls are the best comprise we have. But, like I’ve said before, suggestions are welcome, at least by me!
I agree with Fin completely.
A story I was told about 5 cm standoffs: I was doing testing at Celect, a division of Cincinnati Electronics, in the early ‘90’s with a man named John Day. When I made a comment about 5 cm standoffs, he stated that he may have had something to do with the fact they are a bit inconvenient. He told me that in the early days of military EMC emissions, someone thought that it would be a good idea make sure the cables are not laying directly on the copper bench, so they grabbed 2x4’s and put the cables onto them (which are 4 cm high). However, once things needed to be documented, and they were converting to metric, John was asked to go measure the height of the cables off the ground plane in centimeters. When he reported the measurement, he reported the height of the cable, which was a large cable bundle, and had measured to the center of the cable – which added 1 cm to the overall height.
He said the next thing he saw was a document stating the cables needed to be on standoffs 5 cm high based on the past tests. And he did not correct it.
Or this could just be a story he told me.
FYI, some history . I knew Ken would have the lowdown. He has a pretty good museum of old equipment and books on EMI. And a pic of some of our old generators!!
The answer involves some heavy math.
The first part of the heavy math is if one picture is worth a thousand words, how much is two pictures worth?
The first pic shows a radio room in a WW II-era bomber. On the left is the radio, on the right the unshielded antenna lead from the radio is visible held above aircraft structure via porcelain standoffs, to minimize capacitive loading of the high impedance signal.
The second pic shows the standoff dimension, in inches.
The rest of the heavy math is the conversion to MKS, which is left as an exercise for the reader...
- Ken Javor
Oh, of course. And I should have known Ken would know.
And I kinda figured John Day was handing me a line. Kinda.
Thanks! This is great stuff.
To be continued........
For more information on the roundtable, visit EMC Live's website here.