Even for those interested in delving into this field, radio frequency electronics seems like a black art. Unfortunately, the poor soul only needs to deal with it occasionally, such as when seeking to minimize electromagnetic interference. This primer on how RF chokes work to reduce EMI is a good way to explain the theory from a practical, results-oriented perspective.

As an amateur mechanic and machine tool manufacturer, [James Clough] has encountered many ugly cases of EMI. Variable frequency drives are where EMI can cause problems, and chokes on the motor phase output are usually specified. He used expensive chokes specifically for VFD applications on one of his machines, but he wanted to know if cheap ferrite cores could also do the job, so he started looking.

Scanning some ferrite cores with a borrowed vector network analyzer, the results proved unsatisfactory, so [James] conducted a simple experiment with a function generator and an oscilloscope. His demonstration showed how the impedance of the choke coil increases with the frequency of the test signal, which is exactly what you want to achieve in a VFD—passing relatively low-frequency phase signals while blocking high-frequency EMI. For a better measure, he put a capacitor in parallel with the choke and showed how good the low-pass filter is.

We like this kind of demonstration, which not only meets the intellectual needs, but also has practical goals. [James] not only showed that (at least in some cases) a $13 ferrite can do the same job as a $130 VFD choke, but he also showed how they work. This is the basic thing, but it is what you need to know to move to more advanced RF filter design.

no comment yet? It seems to me is a good practical use case and laboratory test method. It is always interesting to observe the simple test method and the rack-mounted version method... even with desktop equipment, because the equipment is getting smaller and smaller. Maybe I'm digging into filters for the time being, and my first SMD 0603/0805 welding is practicing for these. The latter one is my first attempt: https://www.facebook.com/photo.php?fbid=3700447600017015&set=p.3700447600017015&type=3&theater

Yes, interesting video. I actually watched it to see how placing a cap parallel to the ferrite will improve its performance (well, I admit that I am just going to watch it and make a critical comment, because obviously it is incorrect) but watch the video I realize the misunderstanding. Cap is parallel to the source, not parallel to the ferrite.

This is good practical information for those who want to learn about ferrites and filters. Phasor circuit analysis will enable you to go further in the field of noise.

Yes, the strange thing is that there are only a few replies. I actually watched the video and liked it because I have never experimented with these kernels...now I want to test them and see how they behave xD. Thanks!

Bigger is not always better, try using smaller ID ferrite. For the same cross-sectional area, fewer IDs are more effective and cheaper and easier to find from reputable sources such as Laird products from Allied or Digikey. The relative efficiency from one ferrite to another with the same composition: = number of turns * (cross-sectional area)/ID, IIRC.

I have asked EMI questions many times and learned a lot in the process. Include the knowledge/noise ratio characterization of select colleagues who always talk about EMI, as if it were a mysterious beast that would come from nowhere and cause problems in your crappy design. no. On several machines that I had to redesign, I didn't think about minimizing EMI in advance. Some of these problems were fixed in 10 minutes by adding a ground terminal block near the end of the cable in the wireway and reducing the 25-foot shielded extension cord, thereby shortening the length of the new TB. There are a bunch of GRN/YLW at my feet. This makes the omniscient residents who are "investigating" the problem very angry. There is nothing to investigate, this is the basic content I told him. It's too late now, I cut all the wires, goodbye! walked away. I don't want to hear other stories about how water has memory or how many Italian villas he owns. I'm not kidding...

Except ferrite, any phase cables must be shielded, and the shielding layer must be grounded with a shield clamp such as Icotek 37620. I really like these clips. Do not solder the extension cord to the braided shield and connect it to the terminal or other ground, no matter how short. You need surface area. I got the best results with only a common mode ferrite, the wires are twisted together instead of separated. As close as possible to the VFD/drive. Cylinders are better than donuts, you need extra cross-sectional area and the smallest ID. Always do a temperature test afterwards. After finding that the ferrite became too hot, I had to reduce the number of turns or increase the size. If you focus on the drive and servo system, you can eliminate a lot of noise, but don't ignore good shielding practices for other wiring. In one instance, I went from a noise of about 300v to 40mV. Not sure if I believe in 300V at all, but that is an oscilloscope with a diff probe.

*As he showed, the line-side EMI filter used for the driver *Properly shield all cables *Ferrite on the phase cable *Isolate the noise level *Ground all metal plates and ensure that the input PE is good. *Ground loop awareness

Thank you for your long post. I have a question about "Cylinders are better than doughnuts": Do you have experience with other shapes and their relative effectiveness at a given size? (Like double E, or pot type)

Please be kind and respectful to help make the comment section great. (Comment Policy)

This website uses Akismet to reduce spam. Learn how to handle your comment data.

By using our website and services, you explicitly agree to the placement of our performance, functionality and advertising cookies. learn more