Charles,
Sometimes, I wish I was a “lurker” and I could just keep my big keyboard shut. But I can’t so…. I can tell that my previous sarcasm may not be welcome. My apologies. I will readily admit that others have provided you excellent information regarding how to perform this measurement. Heck, Dave Cuthbert has even offered to measure a bead for you on his Agilent unit. I’m playing devil’s advocate and asking the questions: Why make the measurement? Or, what measurement should I make? Or, if I make the measurement, what does it tell me? Without sarcasm, the blunt fact of the matter is: your component engineering group may be sending you off to make a measurement that you are not equipped to make. I have no doubts about your diligence or ability to perform the measurement. However, consider that, even if you make the measurement correctly; your data still may not directly correlate to your final goal, which is to ensure EMC compliance of your overall system By the complexity of responses received, you can appreciate that ferrite characterization isn’t a simple no-brainer. You need to account for measurement error, parasitics, impedance mismatch and sample to sample variation. Obviously manufacturers are set up to make this measurement. Many of them perform this measurement by accepted industry standards. Ferrites are usually characterized by insertion loss or impedance. When you compare ferrite spec sheets, ensure that you compare insertion loss vs. insertion loss and/or impedance vs. impedance. A more fine point, when you compare insertion loss, you need to ensure that the insertion loss measurements were made with the same assumed “system impedance”. (usually 50ohms) If not, there are formulas that can be used to convert one system impedance to another. I can give them to you, if you’re interested. Lets say that you have overcome all of these hurdles and you have properly characterized a ferrite. Now, you want to correlate this characterization to EMC compliant system performance. How will you make that determination? Of course, you will make that determination based upon the data (be it insertion loss or impedance) that you have taken. DIDN’T THE MANUFACTERER ALREADY PROVIDE THESE SPECS? So, you have learned how to make a measurement, possibly rented equipment, built a test setup and spent all sorts of time, only to replicate data that you may already have. Now, let’s consider the question of whether the data correlates to EMC compliant system performance: In order to do that, you would need to know the impedance of your system. What is the input impedance that your original ferrite “sees” when installed in your system? Why is this important? Well, a ferrite with a higher impedance would seem to be better; but what if the higher impedance causes an impedance mismatch and reflections (already mentioned in one of your responses)? Your system may be worse off with the higher impedance ferrite. Your system may also be fully EMC compliant with a lower impedance ferrite. After all, your original system (probably) only has a few trouble frequencies that required the ferrite in the first place. So, if you performed full blown EMC testing, you would possibly find that any number of ferrites would work. Unless you know: The impedance of your system, AND Your previous EMC test margins AND How to simulate how a ferrite will interact with your system impedance, THEN Having a set of your own insertion loss and/or impedance measurements will put you no further ahead then just comparing the spec sheet in the first place. After all of this “nay saying” I do have some positive advice to offer: If you are going to characterize a ferrite, it would be more appropriate (IMHO) to your overall goal (an EMC compliant system) to do the following: 1. Compare specs first. Ensure that your new candidate(s) has comparable insertion loss and/or impedance at your frequencies of interest. Ensure that the specs were taken by the same test method, just call the manufacturers and/or read the data sheets carefully. Just doing this (in my eyes) puts you in the same position that you would have if you had measured the ferrite characteristics yourself. Remember, you can mathematically convert insertion loss data taken assuming different system impedances. Step one should be good enough unless: you simply can’t get spec sheets and/or your manufacturers are using different and varied test methods and/or you want to see if you can get away with a little “weaker and cheaper” ferrite and/or your original test margins are very tight. 2. If you want to go further, then you don’t have to perform a full blown EMC test to do this, you can perform some in-house measurements (depending upon what you have available) of the ferrites in your system. I have performed similar tests whereby I take a near field probe measurement from a cable, place a ferrite on it, then retake the measurement with identical probe placement. Some may say “schlokky” to this. Good for them. At least, it characterizes the ferrite in its final usage environment. I would assert that this fact more than makes up for any lack of measurement equipment. 3. I would consider characterization by individual measurement (as previously proposed) in a test setup as a third option. 4. Full blown EMC testing always lurks as the expensive last option. Of course, since Dave is offering to measure for you, you could simply mail him the samples and short circuit the whole mess :-). I hope that I have helped you and I wish you happy holidays. Chris Maxwell Nettest Utica, NY 315-266-5128 chris.maxw...@nettest.com _____ From: Grasso, Charles [mailto:charles.gra...@echostar.com] Sent: Friday, December 19, 2003 1:28 PM To: Chris Maxwell; 'Ken Javor'; 'Price, Ed'; 'emc-p...@majordomo.ieee.org' Subject: RE: Measuring a ferrite performance Well Chris - There is a long story attached to this requirement. Let me summarize: 1. We manufacture a significant number of settopboxes for satellite TV. I hesitate to put actual numbers in public - lets just say a sh*t load. Cost therefore is paramount...as you can appreciate. Mulitple vendors of components are therefore essential.EMI testing can get prohibitive in terms of time and cost so .. a). We take our baseline unit and will compare the freq response of the "golden bead" to an alternate. b). As long as the alternate matches or is better than the "golden bead" no brains required and no emissions test required. c) If the desired bead has inferior performance then a decision can be made to either schedule a test or nix the vendor. Makes sense ? I hope so..:-) Best Regards Charles Grasso Senior Compliance Engineer Echostar Communications Corp. Tel: 303-706-5467 Fax: 303-799-6222 Cell: 303-204-2974 Email: charles.gra...@echostar.com; <mailto:charles.gra...@echostar.com; > Email Alternate: chasgra...@ieee.org From: Chris Maxwell [mailto:chris.maxw...@nettest.com] Sent: Friday, December 19, 2003 7:29 AM To: Ken Javor; Price, Ed; Grasso, Charles; emc-p...@majordomo.ieee.org Subject: RE: Measuring a ferrite performance As usual, this thread has been very educational, the mere asking of the question has unleashed many answers, from which I have gained in my own knowledge. The most practical question that I would ask is: Why the h___ would your component engineering group want you to measure the impedance of a ferrite? There can only be two reasons: either they can't get a graph of impedance from the manufacturer, or they don't trust it. If either of these are true; then I wouldn't buy the ferrite in the first place. I hope that this isn't a "slippery slope" issue for you. Otherwise, your component engineering group may be asking for V-I curves whenever you want to qualify a new resistor :-). All that said, I can appreciate the fact that you may want to perform this measurement just to learn how it's done. If you do perform the measurement, I would recommend one little detail. I would recommend that you first find a "known good" ferrite complete with a manufacturer's impedance graph. Compare your measurements to the graph just as a sanity check. Happy holidays Chris Maxwell Design Engineer Nettest Inc. Utica, NY chris.maxw...@nettest.com
