Differences

This shows you the differences between two versions of the page.

--- emi_testing [2019/02/26 16:30] – created jrseti
+++ emi_testing [2023/10/28 18:36] (current) – external edit 127.0.0.1
@@ Line 7: / Line 7: @@
 =====NOTES=====
-From Roy,
+**From Dave Deboer, Feb 22, 2019:**
+Hi - there certainly are standards and standard techniques.  We used to have a wideband conical antenna — I can have a look.  Multiple discones are often used.  Of course, we do have our feed to use.   You could use it at room temp and turn off everything you can on the feed used to take the measurements (as opposed to a cooled feed you may wish to measure).  I’d suggest this approach.
+The measurements occur in a screen to guard against outside RFI, but also because the RFI from the DUT (device under test) bounces around and eventually will get into the measurement system (it is an echo chamber).  Ideally mode stirrers are used so that you don’t get standing waves that can produce nulls.  You could measure twice, once with a big piece of metal in a corner and once not.  You want to measure with the DUT on (of course) and then off with a reasonably well known source at various frequencies across the band (ideally a comb generator, but could transmit a set of tones out from a signal generator transmitted out of a reasonable antenna (a few simple dipoles fashioned for a few bands would be a start).
+I think sufficient for our purposes would be to use a room temp feed and then maybe a tone/dipole combo every octave across the band.  I will look and see what antennas we may have here that could help.
+**From Roy, in an email Feb 22, 2019:**
+I spent the second half of the '70s doing almost nothing but EMI testing and remedial design on products to get them to meet FCC regulations. I also set up a number of EMI chambers for companies I consulted with to enable them to do their own testing and compliance design.
+David's comments are right on with some of the difficulties of radiated signal measurement. There are a lot of ways to fool yourself into thinking something is clean when it's not, or vice versa.
+I'd like to put radiated testing into perspective.
+The concept, as we know it today, grew out of two situations that left the FCC with egg on their face. Small portable FM broadcast radios were interfering with aircraft navigation instruments and the suddenly popular CB radios were interfering with California Highway Patrol low band VHF 2-way radios. The local oscillator in the FM radios operate in the aircraft navigation band and the local oscillator of the CB radios are in the police band.
+In both cases, the LO signal generated in the offending receiver leaked through the front-end filter to the antenna and then coupled to the antenna of the victim receiver. In an attempt to make the testing of the offending device a black box affair, the FCC came up with the radiated testing standards that products now have to comply with. Radiated tests are an artificial way of abstracting the measurements so all devices are treated the same, are applicable to measuring the effect between two physically separate devices, and are very difficult tests to get right.
+Another point, the fundamental problem in both examples is a signal that is conducted to somewhere it doesn't belong. All EMI problems start with a conducted problem.
+In the case of the controller board clock signal getting into the receive signal path, you already have the perfect test setup. You know the source of the signal and the spectrum analyzer output of the LNA shows you exactly how bad the problem is. Trying to find the interfering signal from several feet away is a real struggle. As a rule of thumb, two efficient omnidirectional antennas separated by 1 meter have a loss of about 40 dB (a factor of 10,000 in power) between them. The offending signals would be far below the noise floor of the test equipment.
+The experiment Franklin suggested where a small loop of wire between the center conductor and the shield of a piece of coax hooked up to a spectrum analyzer is the most sensitive way of finding a signal problem. The small loop is not picking up the radiated field, it is coupling with the magnetic field that is very localized and very indicative of a signal going where it isn't supposed to. Not only do you get a sense of the level of the problem, you can see physically where the signal currents are flowing.
+So, I think it would be more fruitful to focus on directly measuring the feed equipment and discuss how to reduce the interference at the source.
+**From Franklin, Feb 22, 2019:**
+Roy, I agree with everything you said.
+However, there's some other stuff going on, where we've been discussing doing EMI testing on some new devices that are soon to be brought into the ATA facility.  In the past there has been a shocking lack of process, for example no emi testing before introducing new equipment.  In two cases I know about, the new gizmos are going to be mounted outside in the field among the antennas.  Seems like there should be some sort of EMI qualification before introducing new stuff among the world's most sensitive receivers, and broad-band receivers to boot.
+There is some desire to institute a more rigorous process.
+So Jon is trying to understand what he can do.  He has access to a spectrum analyzer and screen room at Minex, which are major parts of what's required.  They don't seem to have any appropriate known-gain antennas.  A significant problem is that he lacks the background to know how to proceed.
+I think you can help them on this.  One way would be for you to teach them how to make use of the tools they have.  Alternately, it might be more appropriate to simply contract out for EMI testing at a local EMI test lab.  Once you visit Minex and see what they have, work with Jon a bit, etc, I suspect you can figure out the best way for them to proceed.
+Among the many little issues... They have no "spec".  There is a standard that astronomers use, ITU rec 769.  (I'll send you a copy.)  I read it quickly and while its something one could certainly use, it is not particularly suited to our situation.  For example, the standard seems use very wide measurement bandwidths.  In SETI we often look at 1Hz wide frequency bins, so this standard doesn't seem particularly appropriate for SETI.