I guess my accidental finding with the vacuum cleaner cannot be used for serious analysis. The more I learn about all this the more I realize that maybe EMI filters are not necessary for home audio. Hopefully the noisy devices carry their filters to avoid polluting the network. I see all this more like an exercise. But just by the time I thought I have some understanding of it, I find that plain calculations with the formulae won't help. For example your suggestion about 4 turns of the power cords themselves, very similar to my construction and also found elsewhere on the internet. How would that work? I mean it does work obviously but why? An attenuation of a few dB somewhere in the MHz range is enough after all?
yes I counted 4 turns, thanks to your good images.
Sure it works, the CM inductance is quite high, along with Y caps you've carefully selected. Id aim for -50dB at ~ 2 -4 Mhz The purpose is to attenuate the SMPS hash and noise NOT the SMPS fundamental conversion frequency or audio band unless you have (good) data for an unusual interference that low. Have look here note the article by Henry Ott , renowned expert
2 pole filter = -12dB / octave
CM filter fc for -50 dB /@ 4 octaves ~ 250 KHz
Last edited:
My vac is indeed an old model. You don't have to be a dog to drive you mad. Whatever filter I'll end up with, it will have to pass this test since I haven't anything better to use.
No TV at home. As former alcoholics would say; "My name is Kostas and I'm television free for ten years!"
But I have interference problems with my PC monitor. Perhaps this could be useful for my experiments.
No TV at home. As former alcoholics would say; "My name is Kostas and I'm television free for ten years!"
But I have interference problems with my PC monitor. Perhaps this could be useful for my experiments.
> No TV at home.
Not the point of that serendipitous story. If the Hoover, Dremel, Waring messes-up a TV, it will try to mess-up audio (which I assume you have). Your tests show it.
OTOH: what is a little 13KHz through electromagnetic fields, when many of these buzz-boxes make huge acoustic racket? If I run my Electrolux while the hi-fi plays moderately I have about 13dB S/N just due to the sucker's sound polution. Even rock-music suffers some.
Though as the above story shows, EMI can go over the wires even further than the sound waves; our victim apparently did not know his neighbor was running an antique vacuum, except by break-up of TV signal.
Not the point of that serendipitous story. If the Hoover, Dremel, Waring messes-up a TV, it will try to mess-up audio (which I assume you have). Your tests show it.
OTOH: what is a little 13KHz through electromagnetic fields, when many of these buzz-boxes make huge acoustic racket? If I run my Electrolux while the hi-fi plays moderately I have about 13dB S/N just due to the sucker's sound polution. Even rock-music suffers some.
Though as the above story shows, EMI can go over the wires even further than the sound waves; our victim apparently did not know his neighbor was running an antique vacuum, except by break-up of TV signal.
Well, I hope it comes without saying that I don't listen to the music together with the vac. The 6,8/13,6kHz peaks are not a big deal for sure. Usually my audio creations have more residual noise... It just caught my attention. You know, I had the chance to measure something related to EMI with my simplistic test equipment. My old vac was about to be discarded but at the moment it will keep its place to my "bench". I can't tell what exactly causes this noise. But I think it's interesting to find the fundamental frequency of it. Probably quite low or not?. Infinia said that we chase the hash above it and that is the useful conclusion I'd like to keep. It changes my wrong interpretation that we would need an attenuation of 40/50 dB @ 150kHz.
I just realized I haven't updated this.
Further investigation proved that the filter does not cut the vacuum cleaner noise after all! I was wrong, fooled by the electrical installation of the building. What does the trick is the length of the mains wiring between the vac and the preamp under test. Tested thoroughly to avoid posting invalid conclusions once again. When the vac is connected to the same plug with the preamp/filter then the spikes pass through. It takes a minimum wiring length of ~5m to work and the longest the wiring the better attenuation. Note that the vac has a two prong power cord, no PE conductor, and it's several meters long by its own so it allowed to verify that it is conducting EMI rather than emissive. Or at least so I think...
Anyway, I rebuilt my filter/power strip trying to incorporate the info derived from this thread. I'd like to take it for granted that the 100-200kHz noise coming almost exclusively from SMPS is filtered right on the source on manufacturers responsibility and I set the target at 1MHz and up. I used big MnZn toroid cores with three turns of 2,5mm copper wire trying to push the resonance frequency as high as possible. They measure 100μH and I tuned them for a f3~136kHz. That should give -40dB at 1,3MHz, hopefully... I built two filters, one for the linear power supplies and one for the audio PC. Total leakage current is 1mA. And a DC blocker.
Further investigation proved that the filter does not cut the vacuum cleaner noise after all! I was wrong, fooled by the electrical installation of the building. What does the trick is the length of the mains wiring between the vac and the preamp under test. Tested thoroughly to avoid posting invalid conclusions once again. When the vac is connected to the same plug with the preamp/filter then the spikes pass through. It takes a minimum wiring length of ~5m to work and the longest the wiring the better attenuation. Note that the vac has a two prong power cord, no PE conductor, and it's several meters long by its own so it allowed to verify that it is conducting EMI rather than emissive. Or at least so I think...
Anyway, I rebuilt my filter/power strip trying to incorporate the info derived from this thread. I'd like to take it for granted that the 100-200kHz noise coming almost exclusively from SMPS is filtered right on the source on manufacturers responsibility and I set the target at 1MHz and up. I used big MnZn toroid cores with three turns of 2,5mm copper wire trying to push the resonance frequency as high as possible. They measure 100μH and I tuned them for a f3~136kHz. That should give -40dB at 1,3MHz, hopefully... I built two filters, one for the linear power supplies and one for the audio PC. Total leakage current is 1mA.
And a DC blocker.Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.