0

If I use this ferrite bead for a USB cable, what can be said about the noise attenuation for 1 MHz and 10 MHz.

The plot only shows the impedance versus frequency, whereas I would like to see the attenuation roughly for a USB cable's interference. I would like to see how much a 1 MHz and 10 MHz interference would attenuate.

Noise originating from the PC and through USB Hub goes into the System. This happens when a nearby device next to PC is activated like motor. All I want to suppress and try if I can attenuate the noise coming through earth or mains through USB cable first.

enter image description here

GNZ
  • 1,559
  • 4
  • 25
  • 52
  • 3
    Attenuation is dependent on source and load impedance. So you would want to know these impedances in USB transmission. – Stefan Wyss Dec 12 '18 at 11:08
  • I see but how can i know that? I there a typical value for USB port input impedance? – GNZ Dec 12 '18 at 11:09
  • You need to show how it is being used - try adding a schematic and show where the noise source is and what circuit nodes you want attenuated. – Andy aka Dec 12 '18 at 11:09
  • @Andyaka It is used from a PC USB port to a USB Hub's port – GNZ Dec 12 '18 at 11:10
  • Show how you would intend to use it. – Andy aka Dec 12 '18 at 11:11
  • @Andyaka See my edit. System is isolated and only have contact to USB Hub. – GNZ Dec 12 '18 at 11:19
  • It's likely that if the noise is coming from a motor (as you say) then adding the ferrite won't really achieve anything. Have you tried it and noticed any improvement? – Andy aka Dec 12 '18 at 11:20
  • @Andyaka No I didnt try it. It happens only when it is switched like a pulse EMI Not continuous I mean opponents during switch on'. – GNZ Dec 12 '18 at 11:21
  • It might be ground bounce - try and cut-out the problem at source by suppressing the motor or re-routing any ground connections to it. If this can't be done then the process is this: try many different "things" additively until the problem goes away then randomly remove some of those potential improvement "things" until the problem comes back. Iterate a few times until you know the main "thing" that gives the highest suppression of the problem and then theorize why that may be so in order to make a more focussed and better solution. It's hardly ever straightforward with these sort of problems. – Andy aka Dec 12 '18 at 11:24
  • I see I will look for ground bounce I need to make a research about it. Removing the motor is not a fine solution I would rather like to suppress or block that bounce. It is because there might be some other devices is a bit long story. In case if it is ground bounce would cutting the PC's earth make sense to try? (Seems like USB ground goes all the way to the earth through the motherboard) – GNZ Dec 12 '18 at 11:34
  • 1
    Nothing you try basically makes sense until you find the silver bullet. After that it starts to make sense. Be random; try things that don't make sense. No need to worry if they don't make sense just try them and keep trying things additively until you hit the spot. Then remove stuff that you tried until it goes belly-up again. That's the only way I know how to fix these problems generally. – Andy aka Dec 12 '18 at 12:17
  • Db attenuation will unlikely to solve anything. Your problem is likely in incorrect managing of shield connections along your entire USB network, especially inside your USB hub. See this post,https://electronics.stackexchange.com/a/408044/117785. – Ale..chenski Dec 16 '18 at 14:38

2 Answers2

1

All filters are voltage dividers, like this

schematic

simulate this circuit – Schematic created using CircuitLab

analogsystemsrf
  • 33,703
  • 2
  • 18
  • 46
  • In summary, I think you can say this: until current flows through a ferrite bead, you can't know its attenuation. If R2 is infinite, R1 doesn't attenuate. – glen_geek Dec 12 '18 at 16:45
  • But dont you guys have any idea what can R2 be roughly if it is a USB port? – GNZ Dec 12 '18 at 16:51
  • @analogsytemsrf I like this kind of circuit involved precise explanations thanx – GNZ Dec 12 '18 at 16:52
  • What will R2 be (what is the impedance) if that impedance is the input to a USB hub? Suppose you want 12,000,000 bit rate? You'll need about 20 nanosecond time constant. Suppose you need 100 ohms to terminate the USB line. The USB line needs to look resistive up to about 100MHz to minimize reflections. Thus you can insert 100 ohms into R2. What if your USB is much faster? Then the USB line still needs to look resistive, but to much faster data rates. – analogsystemsrf Dec 13 '18 at 14:19
0

I've used something similar to get 'hash' off a usb line. I will sometimes loop the cable through a few times. (more turns) I think of this as a common mode choke. But a lot depends on where the noise/hash comes from.

George Herold
  • 4,754
  • 1
  • 18
  • 25