| Electronics > Beginners |
| Multichannel Common mode choke? |
| (1/1) |
| martin1454:
Hi there - A quick question. Im doing some research on CDN (coupling decoupling network) and looking at the EN 61000-4-6 there is a good description of how they are made. But - they uses common mode chokes, but with multiple outputs on 1 choke - What do you call those? I tried to search for multi channel common mode choke, but it gives me nothing, and I cannot figure out how they are connected if there are multiple chokes parallel. See the picture for how the symbol looks |
| T3sl4co1l:
Multiple windings or phases, still a common mode choke as usual. May also be found as "array" or such. The market for tightly coupled signal chokes is apparently very small. Mostly you find junk like, ferrite bead chip arrays, or multi-hole ferrite beads (which couple loosely between wires -- they aren't CMCs so much as FB arrays). If you want to use off the shelf parts, consider using dual-winding chokes and coupling them together with one winding in parallel across all parts, and the other winding used for each signal. (That is, say we number the chokes A, B, C, etc., each has a winding 1 and a winding 2, and each winding has a start and an end terminal. Connect all the winding 1's in parallel, start to start, end to end. This leaves the winding 2's unconnected. Connect these to the signals, A2 start to EUT channel 1, A2 end to AE channel 1; B2 start to EUT channel 2, B2 end to AE channel 2; etc.) This works best for unbalanced signals, where the ground/shield can be wired through the common winding. (Otherwise, for balanced signals, the common winding should be left floating, or grounded to the AE side.) Note that the inductance of a single choke needs to be N times the total required, since they act in parallel and don't share flux otherwise. This is a big downside to the method. The other is, leakage flux between any given pair of windings is double that of a single choke (except when used for unbalanced signals, with ground tied through the common). Finally, it's bad for more than a teensy amount of DC bias, because the chokes will saturate (somewhere in the low mA range), ruining the network response. If these disadvantages are too much, your only alternative will be to wind one yourself. This is easy enough: get a length of cable and loop it around a core enough times to get the required inductance. Toroids are abundant, but a bit inconvenient if you need many mH; in that case, consider a pot core instead. Also consider using fine enameled wire instead of cable -- much smaller, you can get more turns on a given core, which means more inductance for less winding length. Note that winding length sets the high frequency cutoff, where leakage inductance and stray capacitance take over. So there is priority to keep winding length short. Preferably, nothing actually changes at high frequencies, that is, LL and Cs stay in their normal ratio, which is to say, the transmission line impedance stays constant. This is possible to do with common mode chokes, and certain autotransformer designs (Guanella matched delay transmission line transformers); but it cannot be avoided on diff mode chokes (for use with AC coupled balanced pairs), or isolation transformers. Tim |
| martin1454:
@T3sl4co1l Is the "off the shef solution" like the design I have attatched? Im trying to make a simple PCB where I can configure between 2-15 channels, so off the shelf components would be nice, but I can see that we will need a few components if we need the inductance to be (amount of channels * required inductance) So far, thanks for the info! |
| T3sl4co1l:
Yes, exactly. :) Yeah, 15 channels will be irritatingly hard to do, if you need say 5mH for the choke then you need 5*15 = 75mH for each individual part! You may not even be able to find data chokes of that value, let alone of reasonable DCR and differential HF response. Besides pot cores, nanocrystalline cores are also excellent, having more impedance per turn in the 20kHz+ range. They are more expensive, but that can be easily justified in low quantity or high performance cases. Tim |
| Navigation |
| Message Index |