High current ferrite beads.

Hello.

I have a problem(I think), I have a linear regulator that is being supplied by a switching converter and as such I want to use 1 ferrite bead each on the linear regulators input and output. But that regulator is supposed to supply 10A max and while looking for beads I can find beads rated for 10A and 10,5A at the most.

But I have heard that I should use a 20A bead if I want to pass 10A through it, even if that would be a little excessive 10A through a 10A bead does sound as cutting it a little to close, isn't it?

What are my options for high current ferrite beads?

Regards

- - - Updated - - -

(20A rated bead for 10A of current isn't something I have been told as such, but rather using 100% margin)

Well, the bead itself carries no current. Perhaps the issue
is saturation of the "core", killing the effectiveness of the
inductor to snub small incremental / HF dI/dt.

You might look at cores for tiny toroidal transformers, and
try to find the same material as the beads so as to get
whatever attributes matter. But perhaps ferrite beads
like lossy cores, which a transformer core would try not
to be. Still, this is my best guess at options.

dick_freebird said:

Well, the bead itself carries no current..

Click to expand...

I have a few 1206 ferrite beads that are 10R (or is it 0.01R) and 5A. In good old days we used to have beads with a hole that were very convenient to slip on one leg of a resistor.

My suggestion is to just parallel a couple of them; unless you are tight on real estate. I really do not know the anatomy of SMD ferrite 'beads'.

Ok, yes have wondered about those small round circular cores that you often find around individual IC legs such as on TO-220 packages, as well as those beads that are made out of a conductor which simply goes right through one of those a little longer circular beads... Where have those gone? I see them in th most equipment that I opens but never for sale(and yes those kind of devices that I open are not made this year, by far often).

But it sounds as it would be acceptable to parallel two of those beads rated for 10A and that would probably allow a 10A current to be passed(assuming the series resistance are equal), I think I will try that because real estate is not a problem.

Thanks.
Regards

David, this is a good question. Why not wait a day or two for others to read this for more solutions.

yes that is a good idea.

David_ said:

But it sounds as it would be acceptable to parallel two of those beads rated for 10A and that would probably allow a 10A current to be passed(assuming the series resistance are equal)

Click to expand...

Paralleling equal ferrite beads certainly would change the frequency response of the equivalent component, not featuring the EMI immunity as expected.

I searched around and came up with the same 10 amp limit and came the same conclusion as post #2, find the correct material in a toroid and wind your own.

Another option is putting a low esr ceramic capacitor before the regulator to snub the ripple. Or a LC filter, You can buy inductors for higher current more readily.

I will look into both using a small toroidal core to make my own bead and research the possibility of a small LC filter, I'll return when I have done that.

Thank you all very much for your suggestions.
Regards

I think I recall having seen surface mount "beads"
(of course they do not appear very "bead-like").
I wouldn't rule out finding something like this, although
10A DC in a single trace is not really your usual SMD
territory.

Well, whaddya know... Mr. Google turns up plenty on
"surface mount ferrite", many vendors and distributors.

https://www.koaspeer.com/products/emi-emc-filtering/surface-mount-ferrite-beads/
6A size appears largest

Several magnetics outfits like TDK, Murata, Wurth, Laird.
Digi-Key, Mouser, Newark all have broad lines. Newark has
10 choices for you at 10 amps (and one at 10.5).

Vishay also makes these. Lifted from an app note on their site:

Surface Mount Ferrite Beads
Chip impeders, also called ferrite chip beads, perform the
function of removing RF energy that exists within a transmission
line structure (printed circuit board trace). To remove unwanted
RF energy, chip beads are used as high frequency resistors
(attenuators) that allow DC to pass while absorbing the RF
energy and dissipating that energy in the form of heat.
Surface mount ferrite beads have many advantages:
• Small and light weight.
• Inexpensive.
• High impedance values removes broad range of RF energy.
• Closed magnetic circuit eliminates cross talk.
• Beads are inherently shielded.
• Low DCR ratings minimizes desired signal degradation.
• Excellent current carrying capacity compared to alternatives
• Outstanding performance at removing RF energy.
• Spurious circuit oscillations or resonances are reduced
because of the bead’s resistive characteristics at RF
frequencies.
• Broad impedance ranges (several ohms to 2,000 ohms).
• Operates effectively from several MHz to 1GHz.
To chose the proper bead, you should consider the following:
1. What is the range of unwanted frequencies?
2. What is the source of the EMI?
3. How much attenuation is required?
4. What are the environmental and electrical conditions for the
circuit (temperature DC voltage, DC bias currents, maximum
operating currents, field strengths, etc...)?
5. What is the maximum allowable profile and board real estate
for using this component?

Selection of the right bead for your particular frequencies is not
a simple process. In most cases, since beads are only rated for
impedance at 100MHz, you will need to look at several graphs
to determine the best bead for your frequency if it is different
than 100MHz.
This is a time consuming but necessary process to select the
correct bead value since the highest impedance bead at
100MHz is not necessarily the highest impedance bead at
higher or lower frequencies. DC bias will also lower the
effective impedance of the device. (Vishay Dale has developed
a Surface Mount Ferrite Bead Designer's Kit that includes
product samples, electrical data and a slide chart design aid
that allows you to calculate the impedance of a surface
mounted ferrite bead at a designated frequency. This kit allows
quick selection of the correct bead from Vishay’s line of surface
mount beads without the time consuming process of looking at
a multitude of graphs. The Surface Mount Ferrite Bead
Designer's Kit (available upon request, call 605-665-9301) also
calculates the DC bias derating percentage for a range of bias
currents.

The reality is for any ferrite bead that has a modicum of effective mu, it saturates pretty early on in the current stakes, thus under full DC bias they are effectively not there, great for blocking impulses and noise with low to no DC bias, but invisible for currents above 1-2 amps DC.

For serious DC inductive filtering you need a gapped ferrite or a low mu toroid, a tiny wee bead will only act in the gaps when the big DC is not present...

- - - Updated - - -

1206 type "beads" with an open magnetic structure will be different to a bead with a hole that slips over a wire, but still less effective at high DC bias...

I believe that the most important part will be dissipation. These beads are not transformer materials and have wide hysteresis loops. They are used in the same way we use 0.1 uF bypass capacitors at or close to the IC power pins. 5A or 10A is quite a bit of current for a 1206 bead and I hope they can filter spikes with 100mJ of energy.

They are certainly not meant for high current ripple filtering with high DC bias current.

- - - Updated - - -

In post #12, I meant 100 uJ - 100 mJ is lot of energy for a voltage spike.

First comment, asking for a 10 or 20 A ferrite bead without specifying an impedance respectively inductance suggests that you are not sure about the filter parameters and respective device requirements. Did you calculate a tolerable input ripple for the linear regulator?

Secondly, using a ferrite bead at 50% of rated current will reduce the impedance more than expected by most designers due to saturation. See below the impedance curve of a 6A rated 1206 ferrite bead from Fair-Rite. Most manufacturers prefer not to publish similar curves.



Although you can get 10 A rated ferrite beads, I won't use them for 10 A DC. Instead consider a small ferrite inductor with full specifications like this https://katalog.we-online.de/pbs/datasheet/744308040.pdf

FvM said:

Although you can get 10 A rated ferrite beads, I won't use them for 10 A DC. Instead consider a small ferrite inductor with full specifications like this https://katalog.we-online.de/pbs/datasheet/744308040.pdf

Click to expand...

Very interesting information in the datasheet. They are rated for 25A and they saturate at 26A. Of course this a monster part compared to 1206! The inductance starts dropping only above 20A. The part is officially described as "high current flat wire inductor". I wanted to use the same ideas in datasheet to the graph given for a 6A 1206 ferrite bead. It is quite usable upto 3A and good for attenuating above 100MHz. The typical inductance will be about 70 nH (point of 40ohm and 100MHz) and this is strongly frequency dependent suggesting a lossy character.

I think the OP was right when he wanted to use a 20A device for a nominal current of 10A. As 20A devices are not available, two 10A devices may be used in parallel. It will be quite effective at 100MHz to 1GHz.

The "lossy character" of usual ferrite beads becomes effective above 100 MHz, it's not particularly relevant for filtering the output voltage of a switched mode power supply. The ferrite bead can be modeled as a simple inductor in the 100 kHz to 10 MHz range. Despite of it's rather low inductance and even if it's partly saturated, it can still achieve some interference attenuation in combination with ceramic capacitors of sufficient capacitance and low ESR.

THe beads I have been looking at are of the sizes 3318 and 3312, as for the target frequencies I can't really know.
The hole idea comes from a app note from Linear technology, I watch a film clipp where some of the prominent persons within the company(can't remember he's name) presented what he liked to express as "the secret saws" for linear regulators to prevent high frequency content from passing through to the output.
Here is the app note plus a little info on ferrite beads which I have read some time ago and will read through again tonight:
https://www.edn.com/design/analog/4...regulator-residue-in-linear-regulator-outputs
Its based on that LT app note.

I might have been unclear about what I want to filter, this isn't meant to filter any content of the switchers switching frequency but rather the around 100MHz and >100MHz content generated by the sharp edges of the control signal. The content is shown here:

Keep in mind the attenuation of RF content in SMPS is a function of impedance ratios for series and shunt elements in the circuit. There is also a factor of common mode rejection of radiated and conducted noise and ground ripple from poor ground control.

When you don't have a vector impedance analyzer to tell you why you have a problem, it can be analyzed by each component looking at ESR, ESL and transient current peaks. Ferrite raises the impedance from inductance and eddy current losses for which there are two rough categories of low and high mu materials for high and low frequency problems.

Since Jim Williams wrote about these solutions in 2005, there are additional solutions with ultra-low ESR caps, but ESR cap costs rise with AC ripple current rating while inductors and ferrite bead costs rise with DC current rating.


For example if the DC load regulation is 10A for 100 mV it has a output of 10 mΩ. A load cap with 10 mΩ ESR would not attenuate much and you may need two series chokes, one for below 1MHz and one with an SRF above the rise time equivalent frequency.

The output error is a function of the output impedance to load ratio, where the Cout and it's ESR should be as low as possible, relative to the series impedance between the regulator and load cap. THis can be accomplished by ferrite chokes or inductors rated for the DC output. Keep in mind the Series resonant impedance of the choke and you may want to use two, one large and small to keep Zout high in the 1~50MHz range before the load cap.

The same applies for input noise from the SMPS to the linear regulator as the ripple reduction at above 1MHz will be negligible in the regulator depending on gain bandwidth.

SunnySkyguy said:

For example if the DC load regulation is 10A for 100 mV it has a output of 10 mΩ. A load cap with 10 mΩ ESR would not attenuate much and you may need two series chokes, one for below 1MHz and one with an SRF above the rise time equivalent frequency.

Click to expand...

What do you mean with "the DC load regulation", I have a hard time putting it into some context so that I get what you are saying.

By the way, it has been mentioned that manufacturers prefer not to publish graphs which shows Z vs frequency where information for different DC current levels are displayed, I can remember 3 times where I have found them in datasheets(I can't even count how many times they have only included one single plot line which isn't specified as any load at all)

But why is that do you think?
Might it be because there product may appear more effective than they are in reality?

I have begun making an Excel document listing and including some common characteristics of all the ferrite materials available from digikey, since I can't post that and since everyone might not be using windows I will put what I got so far below. Why I do that is because I feel as this is a very slim collection of materials for a site such as digikey, isn't it?
(By the way, these are ALL of the materials they offer)

The DC output impedance is inversely related to load regulation which is more common. I just wanted to mention it, in case it was overlooked.

If load regulation is 2% and output is 10V@10A
then LDO output = 2% of 10V/10A= 20mΩ DCR
if 0.5% of 40V @ 10A then 0.5% of 40V/10A = 20mΩ DCR , the same. There is usually a wide tolerance.

The output ripple for a flat R load will then be purely input ripple rejection.

FWIW just like an Op Amp with GBW/Gain the break point near 10 Hz for the integrator internal stability compensation cap reduces feedback gain and raises output impedance from that point.

Injecting a sine wave into the input can be done to measure attenuation and AC output impedance between a choke the SMPS source or DC lab supply Then will know how much passive filtering is needed with an LC input filter.

Feedthru noise can also occur with stray reactive coupling between input and output..

Just to clarify, it is unlikely that any SMT or thru hole bead will give you a useable amount of VHF spike reduction if it has more than about 2A DC in it, unless it is very lossy at VHF, or the magnetic material is very low mu...or both...