EEVblog Electronics Community Forum
Electronics => Beginners => Topic started by: DOCa Cola on March 05, 2021, 09:52:49 am
-
I am currently reading about common mode chokes. I have seen that there seem to be two winding arrangements. I have tried to find an answer for a while now but haven't found this addressed specifically.
(https://i.imgur.com/lpDBnjp.jpg)
(https://i.imgur.com/uz4kd0v.jpg)
One where the leads are wound side by side and one where they are separated. Both types of chokes seem to have similar ratings. Is there a difference between these two types of chokes electrically or does this difference come mostly from production differences?
-
Using both windings together will have better coupling, and therefore maybe better filtering, but it also has a larger capacitance between the windings, which deteriorates the filtering.
I think the separated windings are preferable for the lower capacitance, but they are also more expensive to make, as the windings have to be wound separately. With the second version, both wires can be put on the big spool, and wound at the same time, and each second of production time counts when you're making millions of those things.
It is also harsher on the isolation quality of the lacquer, but transformers and motors for mains (and higher) voltages have been made for some 100+ years in mass production and this is a pretty much solved problem.
The plastic separator also introduces "double isolation", but I'm not sure if this is mandatory between the two mains wires. The lacquered wires also have multiple layers of laquer, but it is not easy to visually inspect or verify.
-
There a couple of things at play here, all CMT's have a common mode inductance and a differential mode inductance (aka leakage inductance)
The top CMT is normally used for filtering has a large leakage inductance = differential mode, and is used where the amount of noise on each line is different, so the common mode inductance filters the noise current flowing in one conductor and out the other , while the differential inductance filters noise that is just on one line, between the two of them you get good noise rejection. Also the core material used in this style of CMT is very lossy, and helps absorb the noise (rather than reflect it back again). This type is commonly used on mains inputs as it also has large creepage/clearance distances. It is more properly a "choke" than a transformer. This style is less likely to saturate.
See https://www.coilcraft.com/en-us/products/emi/power-line-common-mode-choke/high-isolation/cmt/cmt/ (https://www.coilcraft.com/en-us/products/emi/power-line-common-mode-choke/high-isolation/cmt/cmt/)
The bottom style is a CMT typically used on signal lines , and has very low leakage inductance, typically a signal and it's "ground" line would be passed through such a CMT, if there is a voltage imposed on the distant "ground" line then that voltage gets subtracted from the signal line hence the noise cancels out, the ferrite material has usually a much higher permeability, so less noise current flows for a given noise voltage. This style of CMT can also be used on the low voltage side of a power supply (it doesn't have creepage/clearance for more than 60v) and can be used in certain sorts of power convertors (see https://en.wikipedia.org/wiki/%C4%86uk_converter (https://en.wikipedia.org/wiki/%C4%86uk_converter)) , A drawback of this type of CMT is that it is more likely to saturate at low values of differential current. This style of CMT is sometimes called a "coupled inductor" , they can be quite low loss (if used as coupled inductors) or use a different high loss ferrite (to soak up noise like a sponge, but can get quite hot!)
see also https://forum.allaboutcircuits.com/threads/coupled-inductor-vs-common-mode-choke-vs-ct.152715/ (https://forum.allaboutcircuits.com/threads/coupled-inductor-vs-common-mode-choke-vs-ct.152715/)
and https://www.coilcraft.com/en-us/edu/series/a-guide-to-understanding-common-mode-chokes/ (https://www.coilcraft.com/en-us/edu/series/a-guide-to-understanding-common-mode-chokes/)
and https://en.wikipedia.org/wiki/Choke_(electronics) (https://en.wikipedia.org/wiki/Choke_(electronics))
-
Thanks. That helped a lot
-
First one with windings separated is for mains circuits I guess (where a safety is a main concern).
The other, with windings highly coupled is a low voltage one.
-
A little clarification --
The top CMT is normally used for filtering has a large leakage inductance = differential mode, and is used where the amount of noise on each line is different, so the common mode inductance filters the noise current flowing in one conductor and out the other , while the differential inductance filters noise that is just on one line, between the two of them you get good noise rejection. Also the core material used in this style of CMT is very lossy, and helps absorb the noise (rather than reflect it back again).
Reflection requires a reference impedance. A reference impedance might not really apply to a power supply filter, for example. So, it may not be a great description of what's going on.
Though when it does apply, for example when testing the attenuation of a CMC, by itself, in a 50 ohm system -- its purpose is to increase series impedance, therefore reflecting noise power back to the source, where it's supposed to stay.
Losses help more with damping. Damping is a dynamic response, I would say it's easier to understand in those terms. Damping is valuable, as sudden peaks and dips in a component's impedance, tend to cause peaks and dips in a filter's response; damping tends to lower the peaks, but also raises the dips.
The core, may be of two types. Most often it is very high permeability, ferrite or nanocrystalline. The Q isn't typically great for either of these, but most importantly, the high permeability maximizes that desirable series impedance.
Less often, the core is a lower mu material, and may be more evidently lossy. Examples: MPP, powdered iron, ferrite beads (the two-line type). These would not be chosen for the primary filter element, the impedance just isn't as high; they are used, in addition, when a higher frequency needs to be treated (say >10MHz). In that case, having fewer resonances (and those that remain, being well damped) is a bigger priority than series impedance is.
This type is commonly used on mains inputs as it also has large creepage/clearance distances. It is more properly a "choke" than a transformer. This style is less likely to saturate.
"Transformer" is a very good way to understand them; the difference is semantic, i.e. how we use it, or recognize it. (Also, while editing, I just realized you've used "CMT" consistently; "common mode transformer" I believe? I was reading it as "CMC" by default... :P ) Or alternately, a 1:1 transformer (when used as you'd typically use a transformer) can be very well described as a CMC. :)
The other common term, "current compensated choke", also hints at this -- it's a 1:1 current transformer, so of course enforcing those currents to be equal, manifests as a large series impedance.
I like to conceptually divide magnetic components into three classes:
- Inductors: have rated inductance, saturation current (if applicable)
- Chokes: have very high inductance, or impedance; may have saturation flux
- Coupled: has multiple windings
So we have the composite types,
- Coupled inductors: "inductors" with multiple windings
- Transformers: coupled chokes
Some examples:
CMC: transformer. Has very high inductance and multiple windings; saturation isn't typically rated but we're using it in such a way that it's not expected to matter.
Flyback transformer: coupled inductors!
Ferrite bead: choke.
Note that an RFC (radio frequency choke) might happen to have a precise value after all; it might then be called an inductor if that value is critical, or a choke if not.
Now, I don't know how widespread this classification is, like I said, it's conceptual.
Note, I didn't separate anything based on leakage inductance -- we could take this two ways:
1. Ignore it. Transformers for resonant power supplies, typically have relatively low coupling factor (k < 0.9?), that's fine, they're still transformers. Flyback transformers ("coupled inductors") typically have high k; RF/IF filtering transformers might have low coupling (k ~ 1/Q, for a bandpass of the same fractional bandwidth), both are still coupled inductors. Etc. So in this case, we wouldn't care that a pulse transformer and CMC might happen to have the same ratings but one may have much higher leakage inductance.
2. Separate it. For k near 1, we can subtract leakage inductance as a series inductor, and magnetizing inductance as a parallel inductor. The transformer then becomes an ideal one. And then we can classify the two parts. Typically, leakage will be an inductor (granted, its value might not be critical, but in that case its value should be small, not large), and magnetizing inductance will be an inductor (coupled inductors) or choke (transformers). Annoyance: now we need to type each separated element of the physical component; it becomes a matrix of parameters, rather than a simple whole.
For other k, we can still do this separation, but the transformation starts getting less meaningful, or more complicated (we need to consider primary and secondary leakage, or for some transformer types, different magnetizing inductors too!).
There is also a tee/pi transformation for the nonideal transformer, which works even for arbitrary turns ratios -- if you don't mind that some of the inductors may take negative values. This isn't physically relevant, it just reduces to the first case when k ~ 1 and Np/Ns = 1.
Anyway, that's more than enough philosophy for just a magnetic component... In short, a nonideal transformer model is truly general, and meaning should be understood from its parameters, relative to the circuit it's in.
I just really like transformers... :P
The bottom style is a CMT typically used on signal lines , and has very low leakage inductance, typically a signal and it's "ground" line would be passed through such a CMT, if there is a voltage imposed on the distant "ground" line then that voltage gets subtracted from the signal line hence the noise cancels out, the ferrite material has usually a much higher permeability, so less noise current flows for a given noise voltage. This style of CMT can also be used on the low voltage side of a power supply (it doesn't have creepage/clearance for more than 60v) and can be used in certain sorts of power convertors (see https://en.wikipedia.org/wiki/%C4%86uk_converter (https://en.wikipedia.org/wiki/%C4%86uk_converter))
Quite correct. Mind, what's pictured, seems to be very much a power choke; I mean, it would work fine on signals too, it's just really oversized. ;D But yeah, the same type of construction is typically used in those. Or a cylindrical build, where the parallel pair is wound around a cylinder, which is then capped with blocks, capped again with a plate -- like so:
https://www.coilcraft.com/getattachment/d066e321-89a1-4bc7-873e-6228c01382aa/0603usb.jpg (https://www.coilcraft.com/getattachment/d066e321-89a1-4bc7-873e-6228c01382aa/0603usb.jpg)
These have the advantage that, although leakage inductance is modest, the more important part is that the pair acts as a transmission line. Indeed, the leakage we measure at low frequencies, is the LF equivalent inductance of that transmission line.
Some data chokes, you'll see DM impedance plotted as well; at high frequencies, it peaks above and below some mean value, which is Zo. (A pure leakage would only peak!)
A drawback of this type of CMT is that it is more likely to saturate at low values of differential current. This style of CMT is sometimes called a "coupled inductor" , they can be quite low loss (if used as coupled inductors) or use a different high loss ferrite (to soak up noise like a sponge, but can get quite hot!)
Two things --
1. CMCs always saturate at relatively low unbalanced currents. (If saturation current were nominal, it would imply nominal inductance as well, making it a coupled inductor, not a choke.)
2. Coupled inductors are made in both styles -- unfortunately. It would be nice to have some nice high-k coupled inductors! Alas, they aren't always made (or specified) that well.
An extreme example, here's one I made using four strands in "star quad" configuration:
https://www.seventransistorlabs.com/Images/Dual_Inductor.jpg (https://www.seventransistorlabs.com/Images/Dual_Inductor.jpg)
With the thin enamel insulation, the characteristic impedance of this transmission line is very low, probably under 20 ohms. I measured something like 10nH I think, for leakage. It performed very well in a Ćuk converter. :)
Don't think anyone goes to that length for commercial products. More often you see parts like these:
https://www.digikey.com/en/products/detail/pulse-electronics-power/P0396NL/2266728 (https://www.digikey.com/en/products/detail/pulse-electronics-power/P0396NL/2266728)
https://www.digikey.com/en/products/detail/pulse-electronics-power/P0599NL/2266354 (https://www.digikey.com/en/products/detail/pulse-electronics-power/P0599NL/2266354)
(which, sadly, neither one gives a coupling factor in the datasheet!)
(Obviously, these are using a low-mu core, giving nominal inductance and high saturation current. The same designs apply for CMCs, just with hi-mu cores.)
Also the shielded bobbin type is very common, particularly in smaller ratings:
https://www.digikey.com/en/products/detail/bourns-inc/SRF1280A-1R5Y/5030940 (https://www.digikey.com/en/products/detail/bourns-inc/SRF1280A-1R5Y/5030940)
I think these are just a spool of wire embedded in a core (whether ferrite shapes, or molded powder composite), so should have modest k. (I don't think anyone makes CMCs in this format? Molded powder doesn't have a high enough mu, but I suppose the ferrite shapes could be fitted tightly enough to do the job. Probably, using toroids is just cheaper.)
Tim