9
\$\begingroup\$

I live in a place where ferrite beads for EMI suppression are hard to get. I thought steel had a very high iron loss, especially at high frequencies. So is it possible at least theoretically to use hard steel nails to attenuate high frequency noise in, say, power supplies?

\$\endgroup\$
9
  • \$\begingroup\$ I don't know. And google didn't grab up anything for a chart on the relative impedance versus frequency of steel nails, either. Since I haven't tried it, I'll leave it to those who are far more prone to using steel nails instead of ferrite beads in their projects than I've been. \$\endgroup\$
    – jonk
    Jan 7, 2020 at 4:22
  • 3
    \$\begingroup\$ So you live in a place where ferrite beads are hard to get, but power supplies that produce high frequency EMI aren't? \$\endgroup\$ Jan 7, 2020 at 5:05
  • 2
    \$\begingroup\$ @BruceAbbott Yes, SMPS modules available here are all very noisy. \$\endgroup\$
    – ASWIN VENU
    Jan 7, 2020 at 5:14
  • 1
    \$\begingroup\$ Most SMPS modules have EMI suppression built in. If it's not enough then perhaps you could scavenge parts from surplus units? (I live in a place where components of all types are easy to get, but I still dismantle old equipment for parts rather than just throw it all away). \$\endgroup\$ Jan 7, 2020 at 5:17
  • 2
    \$\begingroup\$ Certainly it's possible, but (when I had EMI testing needs) the repeatability requirements meant we couldn't use such things in a product. One can also tie a wire in a simple knot to form an inductor. \$\endgroup\$
    – Whit3rd
    Jan 7, 2020 at 5:48

5 Answers 5

18
\$\begingroup\$

I thought steel has a very high iron loss especially at high frequencies. So is it possible at least theoretically to use hard steel nails to attenuate high frequency noise in say power supplies?

No, not really. A ferrite bead (for instance) relies on the outer ferrite material (not the through-going wire) being both a poor electrical conductor at low frequencies but, at high frequencies, becoming a lossy capacitor and capable of turning EMI into heat. Here are a few examples from Murata: -

enter image description here

As you should be able to see, FBs are designed to target a specific range of frequencies and, different values in the same model range can be chosen to give better attenuation at certain parts of the spectrum whilst maintaining reasonably low losses for signals that should not be significantly attenuated.

I live in a place where ferrite beads for emi suppression is hard to get

A nail doesn't have one of the vital characteristics of ferrite that make it very useful as an attenuator namely; that it acts as a lossy capacitor as frequency gets higher and therefore resonates with the parallel inductance of the through-going wire.

There is a good document from Analog Devices that explains things in more detail and that document shows the developed model for a Tyco Electronics BMB2A1000LN2: -

enter image description here

R1 and C1 represent the lossy dielectric of the ferrite material and you just won't get that with a regular piece of iron or a nail. In case anyone notices the typo in the ADI picture above (L1 = 1.208 uF) should read 1.208 uH thus producing a peak resonance at around 112 MHz.

\$\endgroup\$
8
  • 1
    \$\begingroup\$ Did you mean lossy resistor? Because in the link you provided it says, "It becomes resistive over its intended frequency range and dissipates the noise energy in the form of heat." and "To reduce high frequency noise, the bead must be in the resistive region; this is especially desirable for electromagnetic interference (EMI) filtering applications. The component acts like a resistor, which impedes the high frequency noise and dissipates it as heat." \$\endgroup\$
    – ASWIN VENU
    Jan 8, 2020 at 5:14
  • 3
    \$\begingroup\$ @ASWINVENU no, because a lossy resistor is just a resistor. I meant a lossy capacitor and that can be modeled as a capacitor in parallel with a resistor; that resistor representing the dielectric losses. \$\endgroup\$
    – Andy aka
    Jan 8, 2020 at 8:04
  • 2
    \$\begingroup\$ Early on it says: ac resistance (ac core losses) associated with the bead and this is the heart of what I'm talking about. The ferrite core is basically non-conducting for DC because you can regard the ferrite as tiny insulated islands of conductivity that possess capacitance to each other so, as frequency rises, these islands of conductivity get connected capacitively. But they are not perfect conductors and have loss hence, the combined effect is capacitance and resistance in parallel/series arrangements that can be bulk modeled as a single capacitor in parallel with a resistor. \$\endgroup\$
    – Andy aka
    Jan 8, 2020 at 8:37
  • 1
    \$\begingroup\$ @ASWINVENU There's a good chance those toroid cores are made of ferrite or similar, so it's worth a shot. But not all of those are equal or made for the same purpose, so your mileage may vary. \$\endgroup\$
    – Mast
    Jan 8, 2020 at 9:29
  • 1
    \$\begingroup\$ @ASWINVENU the toroids used in some switching supplies are likely to be common-mode filter chokes so using them as they are electrically intended is a decent idea but, you will need grounding capacitors to make them effective. On the other hand, re-purposing the ferrite is likely not to be as effective as you think at frequencies where FBs are effective. FBs start to become effective above 10 MHz and, the ferrite in toroids (unless very specifically chosen) will not be good at suppressing EMI above 10 MHz (just like many FBs are not good at suppressing EMI above 100 MHz). Horses for courses. \$\endgroup\$
    – Andy aka
    Jan 8, 2020 at 11:35
12
\$\begingroup\$

People used to make lossy inductors (chokes) by scramble-winding 100 turns around a 100 ohm resistor; have the choke and resistor in parallel.

The scramble/random winding was to minimize the coherent capacitance between layers of the wire, thus reducing risk of any resonances.

If this RFC Radio Frequency Choke were in the plate of a Class C amplifier, you'd probably use a 2_watt 1,000 ohm AllenBradley resistor; the large resistor was needed for surviving the high voltage.

Read some construction articles in old ARRL (ham) manuals. Or QST magazines.

\$\endgroup\$
1
  • 3
    \$\begingroup\$ Thos were the days :-) +1 \$\endgroup\$
    – Russell McMahon
    Jan 7, 2020 at 9:58
10
\$\begingroup\$

The full theory here is very messy. You have skin effect and hysteresis loss in play. So, I thought I'd do an experiment. I close-wound a coil of #32 magnet wire on a 10D 3" galvanized steel nail. Here's the coil:

enter image description here

Note that the ruler is centimeters.

I measured the attenuation with the coil placed in series between a sinusoidal function generator and a scope with a 50 ohm terminator across its input. From that, I calculated the impedance versus frequency:

enter image description here

Not so different from a ferrite choke.

Edit:

For more detail, I took more measurements with a tightened-up test setup. I give raw measurements for your computational enjoyment. Input voltage was 1V RMS sine waves: I readjusted the function generator output for each frequency. The time delay through the setup with the coil shorted was 29 ns, measured with a square wave. I have not corrected the delay below for this.

  • MHz Volts Delay(ns)
  • 1.0 0.36 100
  • 1.4 0.34 71
  • 2.0 0.28 59
  • 2.8 0.25 48
  • 4.0 0.21 38
  • 5.6 0.19 32
  • 8.0 0.16 28
  • 11.0 0.15 24
  • 16.0 0.16 21
  • 22.0 0.22 20

The "resonance" is extremely broad.

\$\endgroup\$
12
  • \$\begingroup\$ Thanks for doing an experiment. So you are saying it is possible to use steel nails? \$\endgroup\$
    – ASWIN VENU
    Jan 8, 2020 at 3:04
  • \$\begingroup\$ Possible? Sure. May take some experimentation to get it to work. A different nail might give a different result. \$\endgroup\$
    – John Doty
    Jan 8, 2020 at 3:35
  • \$\begingroup\$ Ok, my theory was that iron losses are contributing towards the attenuation. Is that correct? or is there another explanation? \$\endgroup\$
    – ASWIN VENU
    Jan 8, 2020 at 4:30
  • \$\begingroup\$ As I said, the theory is difficult. There's some combination of ohmic loss (eddy currents), and magnetic hysteresis loss going on here, but I have no calculation. \$\endgroup\$
    – John Doty
    Jan 8, 2020 at 4:37
  • 3
    \$\begingroup\$ 6 points isn't enough to tell the Q (and loss) of the inductor. Can you test at some intermediate frequencies (eg. 2,3,4,5MHz etc.) to determine the peak response? Also, what do you get if the same coil is wound on an inert core? (sorry for asking, I would do it myself but I don't have the equipment). \$\endgroup\$ Jan 8, 2020 at 6:58
5
\$\begingroup\$

As Andy explained, steel won't work. Besides, its magnetic permeability gets lower as frequency increases, and if you wind an inductor over a nail, then the nail which is conductive will make your inductor act as a transformer with a shorted turn secondary* and the inductor will be rather useless.

You can get ferrite cores for free on old cables though.

enter image description here

*: Inductors with solid conductive cores (like a nail) are a bad idea. The AC magnetic field will induce eddy currents into the core, which turns the inductor into a transformer. The primary of this transformer is the inductor, and the secondary is the core itself which is the same as a secondary coil with a single shorted turn.

enter image description here

This increases losses (not a problem here) but it also makes the inductor less effective... not what you want for filtering.

\$\endgroup\$
3
  • 1
    \$\begingroup\$ As a transformer?? I think not. I have done experiments using only the primary of transformer making it behave exactly like an inductor. I believe you are talking about eddy current losses. \$\endgroup\$
    – ASWIN VENU
    Jan 8, 2020 at 2:54
  • 1
    \$\begingroup\$ Yeah, eddy currents are what happens when the core is conductive and becomes the secondary of the transformer... I added some explanations in the answer. \$\endgroup\$
    – bobflux
    Jan 8, 2020 at 10:23
  • \$\begingroup\$ Yeah that is the point. The ferrite dissipates the EMI as heat in the form of iron loss. \$\endgroup\$
    – ASWIN VENU
    Jan 8, 2020 at 11:19
1
\$\begingroup\$

No less a guru than Tom Rauch, W8JI himself debunked the idea of using a solid steel core in place of ferrite.

From https://www.w8ji.com/steel_wool_balun.htm:

Bolt baluns, steel wool balun cores, and other solid iron or steel or laminated cores are a HOAX. They are not baluns at all. The text below explains why they are hoaxes.

A problem arises when magnetic fields are changing level. A time-varying magnetic field will generate currents in any closed conductor path, even iron. This current is called an "eddy current". The amount of current depends on the size of the iron particle(s) forming the conductor and the change rate of the magnetic field. The larger the particles, the larger the conductor area for eddy currents become. The larger the area, the lower the frequency where eddy current start to cause problems.

Eddy currents generate their own opposing magnetic field to the incident field exciting the core. If the iron has large enough cross section eddy currents and the resulting counter MMF from the eddy currents will push the magnetic fields back out of the core. With increasing frequency, a given size iron particle has a frequency where inductance starts to decrease. This effect is from the "shorted turn" generating opposing flux. As frequency increases further, inductance decreases. At some frequency the core is no longer able to support the field, and because of eddy currents, actually reduces inductance.

For example, inserting a solid iron slug inside a small RF coil shows a behavior almost identical to using brass or aluminum slugs. Inserting a solid slug of iron might increases the magnetic field concentration and inductance near direct current frequencies, but at some higher frequency eddy currents and the inability of the core to follow field changes cause the flux concentration to decrease.....eventually reaching zero. At some frequency, the counter MMF takes over. Inductance is actually reduced by the core. ...

I have performed this very same test here with iron, brass, and aluminum cores with the same results. I used an MFJ-259B, while Tom used the expensive test equipment in his lab.

... Let's stop this myth about steel wool and steel bolt baluns. Audio core materials are laminated or powdered for good reason. RF cores are small insulated particles packed together for a reason. A solid material does not behave like a smaller cross section material.


The same principles for a good EMI choke also apply here. Sure, a balun passes RF on the center conductor and the inside of the shield. However, its principal function is to choke off any common-mode current on the outside of the shield.

Also, any unwanted energy that dissipates in an EMI choke is going to be in the form of heat. An EMI filter choke needs to have a high resistance at RF if it is not going to pass RF, same as a balun.

\$\endgroup\$
4
  • 1
    \$\begingroup\$ I believe the OP isn't talking about a balun, but a lossy EMI choke. A balun's job is to transmit electromagnetic energy between its ports. An EMI choke's job is to dissipate unwanted electromagnetic energy. A properly chosen ferrite can do either. Bulk steel cannot (although "powdered iron" in an insulating binder can). \$\endgroup\$
    – John Doty
    Jan 9, 2020 at 16:16
  • \$\begingroup\$ Yeah, I'm talking about lossy chokes. \$\endgroup\$
    – ASWIN VENU
    Jan 9, 2020 at 16:54
  • \$\begingroup\$ Answer edited to explain further. \$\endgroup\$ Jan 9, 2020 at 20:43
  • \$\begingroup\$ And your edit didn't address the issue. As my measurements above showed, the coil wound around the nail has a large RF impedance (about 50 times its DC resistance) at MHz frequencies. Unlike an inductive choke, this impedance is dominated by resistance, not reactance. Reactance reflects power, resistance absorbs some of it. This is a simple coil, intended to suppress EMI on a single wire, not a balun suppressing common mode but passing differential mode. 250 Ω is good for this, as the characteristic impedance of a wire that's not part of a transmission line will probably be of this order. \$\endgroup\$
    – John Doty
    Jan 9, 2020 at 21:03

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.