Neven,
>The distortion comes from the impedance in series with the signal which the >CMC can present for the current-steering mode where the Tx current is pulled >into the >IC from the transformer center tap (i.e. the current is not sourced >inside the transceiver and must pass through the CMC in a very imbalanced >way). Place a 2-wire CMC >between the transformer and the transceiver and if >the CMC has high impedance in the spectrum of the Ethernet signal the chances >are you won't have a good link. 3->wire CMCs can resolve the distortion, but >they are not so effective in suppressing EMI. I leaned and now completely disagree with you. Impedance cannot create distortion. You need a non-linear element for distortion to happen. There is no non-linear element in a non saturated CM coil. CM coils (by design) do have low impedance for Ethernet signals, only leakage inductance need to be concerned ( a minimum value of < 150 nH j58 Ohm @ 62.5 Mhz is specified in the LSI datasheet mentioned below. They do have impedance for the imbalanced (Common Mode) signal (around 350 uH see datasheet), thereby improving the quality of the signal and reducing EMI caused by CM current. The current drawn by the 2 Tx outputs can be split into the wanted DM (=signal) part which is symmetrical (= balanced=differential), and a CM part caused by power supply (current drive) plus imperfections in the drivers and is asymmetrical (=imbalanced =Common Mode). The CM current an be split in AC and DC (for a current driven PHY). The DC part sees the full inductance, but that is not a problem for DC. It only magnetizes the core with a DC field, so the CM coil needs to cope with that (design !). Voltage driven PHY can be capacitive coupled, so the Tx lines carry no DC. The AC common mode current is only noise and we want to get rid of that. This is what happens in the primary (driver)transformer center tap. CM current goes through both winding half’s un-attenuated and sees half the leakage inductance + an insulation capacitor to ground. The CM coil in front of the transformer adds CM impedance so the effectiveness of the center tap improves. The center tap also opens a inverse way for CM current from noisy ground (noisy compared to the system, not to the PHY) to go un-attenuated into the Tx pair. For this last problem a second CM coil just in front of the RJ45 connector helps. This can be a small one, as it carries only low level AC signals. it improves the balance of the output signal and reduces EMI. A tree wire CM coil is of no use in a 2–wire system and needs not be discussed. Most PHY datasheets that address EMI prescribe CM coils to be placed between Tx outputs and the magnetic. Notably current sourced ones. Look for the 80227 (LSI) datasheet as a good example (section A5). While this is an obsolete chip, its datasheet clearly discusses/ shows the topics we are talking about. http://www.lsi.com/DistributionSystem/User/AssetMgr.aspx?asset=48591 Regards, Ing. Gert Gremmen [email protected] <mailto:[email protected]> www.cetest.nl Kiotoweg 363 3047 BG Rotterdam T 31(0)104152426 F 31(0)104154953 Before printing, think about the environment. Van: [email protected] [mailto:[email protected]] Namens [email protected] Verzonden: Monday, October 04, 2010 9:33 PM Aan: [email protected] Onderwerp: Re: [PSES] EMI issue with Ethernet I can't go into details in this forum, I do not have the time and I don't have a desire to do so. I just tried to help with some pointers. So, here is an exception, a limited explanation just for you: The distortion comes from the impedance in series with the signal which the CMC can present for the current-steering mode where the Tx current is pulled into the IC from the transformer center tap (i.e. the current is not sourced inside the transceiver and must pass through the CMC in a very imbalanced way). Place a 2-wire CMC between the transformer and the transceiver and if the CMC has high impedance in the spectrum of the Ethernet signal the chances are you won't have a good link. 3-wire CMCs can resolve the distortion, but they are not so effective in suppressing EMI. Voltage mode drivers are supplied from within the transceiver. That makes a huge difference. Nuf from me on this topic. ----- Original Message ----- From: "qualified testing bv - Gert Gremmen ce-test" <[email protected]> To: [email protected], [email protected] Sent: Monday, October 4, 2010 11:12:35 AM Subject: RE: [PSES] EMI issue with Ethernet Can you elaborate a bit more between the 2 modes of operation. I do not see the physical difference between supplying current to ground or pulling current from the supply. After all, it’s AC. And current sources in a 120 Ohm environment can easily be transformed into voltage sources with the same property. I’d like to learn here. I know how to combat emi problems in Ethernet ports, but ignore a bit about the chipset in front. Another thing: How can a CM coil distort the signal ? For distortion to happen a non-linear element is required, but for the signal current in CM coil no magnetization happens, so no distortion. The only thing that can happen is that the leakage induction attenuates the higher frequencies a bit…. but in a good coil that should be neglectable. And as I wrote earlier, each Ethernet magnetic combo contains 2 CM coils already. And a transformer is also a CM coil. If the CM coil distorts, then a transformer will do that equally. May be you have some figures ??? Gert Gremmen Van: [email protected] [mailto:[email protected]] Namens [email protected] Verzonden: maandag 4 oktober 2010 17:09 Aan: [email protected] Onderwerp: Re: [PSES] EMI issue with Ethernet FYI, I see a lot of noise on this topic, so I wonder if I can help a little. I do not have the possibility to drill deep here. There are two types of Ethernet transceivers with regards to the Tx design. They are current-mode and voltage-mode. Current-mode transceivers feed their output from the center-tap of the transformer, and therefore need the center-tap connected to the power supply. They use current-steering logic, maybe this is too much detail for here. The filtering suggestion (ferrite bead in series between the voltage and the center tap) applies only to this type of the drive, and it is a good recommendation (for this type of tx). It protects the common-mode point of the transformer, i.e. the center tap, from the power supply noise. A recommendation to put a CM choke between the transformer and the Ethernet transceiver is a bad recommendation if this type of the transceiver is used, because it can distort the Tx signal. Also, in general, the magnetics packages that contain transformers and CM chokes should be sufficient, without externally added EMI-suppression devices. In vast majority of cases. The voltage mode transceivers do not need connection of the center tap to power supply, thus only a cap from the center tap to GND is sufficient. The cap at the center tap does indeed provide considerable CM rejection in both cases, up to somewhere in the 50MHz-100MHz range. At these frequencies it is very much needed, and a recommendation to remove the center tap caps was not such a good recommendation. The (im)balance of these transformers is typically not terrible, i.e. the center-tap cap helps and does not increase the emission due to the transformer imbalance. After somewhere in the 50MHz-150 MHz range there are other more significant factors which affect the EMI-rejection and the center-tap bypassing (cap to GND) is not of a considerable importance at the higher frequencies. The most limiting factor is the total inductance in series between the center tap (measured fomr the body of the transformer core not from the pin of the package) and through the capacitor to GND. When this impedance gets too high the center tap does not bypass the CM energy any more, which happens in the aforementioned frequency range. Above 100MHz (+/-), factors such as construction of the magnetics including the location and type of the CM choke, Ethernet connector pins, layout and factors that allow coupling noise across the magnetics, play more role than the center tap. Even if all these factors are tuned to the best possible performance, the imbalance of the connectors and the cables are the final limiting factors, which one can hardly affect - at least practically, to the EMC performance of this type of interface. Going back to the OP, he has a major issue somewhere in his design that he needs to resolve first. Best, N ----- Original Message ----- From: "John Woodgate" <[email protected]> To: [email protected] Sent: Monday, October 4, 2010 2:49:01 AM Subject: Re: [PSES] EMI issue with Ethernet In message <[email protected]>, dated Mon, 4 Oct 2010, "Pawson, James" <[email protected]> writes: >If the magnetics you use for your Ethernet have a centretap on the >input side of the transformer (sometimes connected to a voltage rail, >sometimes to ground) then you should try filtering the supply to this >with an RC or bead-cap filter. You mean the sending end transformer, I suppose. Exactly how do you connect the filter? After all, if the CT is grounded, and you introduce a resistor or bead into the grounding wire, where do you connect the 'cold' end of the capacitor? It surely can't go back to ground! -- OOO - Own Opinions Only. Try www.jmwa.demon.co.uk and www.isce.org.uk John Woodgate, J M Woodgate and Associates, Rayleigh, Essex UK If at first you don't succeed, delegate. But I support unbloated email http://www.asciiribbon.org/ - ---------------------------------------------------------------- This message is from the IEEE Product Safety Engineering Society emc-pstc discussion list. To post a message to the list, send your e-mail to <[email protected]> All emc-pstc postings are archived and searchable on the web at: http://www.ieeecommunities.org/emc-pstc Graphics (in well-used formats), large files, etc. can be posted to that URL. Website: http://www.ieee-pses.org/ Instructions: http://listserv.ieee.org/request/user-guide.html List rules: http://www.ieee-pses.org/listrules.html For help, send mail to the list administrators: Scott Douglas <[email protected]> Mike Cantwell <[email protected]> For policy questions, send mail to: Jim Bacher: <[email protected]> David Heald: <[email protected]> - ---------------------------------------------------------------- This message is from the IEEE Product Safety Engineering Society emc-pstc discussion list. To post a message to the list, send your e-mail to <[email protected]> All emc-pstc postings are archived and searchable on the web at http://www.ieeecommunities.org/emc-pstc Graphics (in well-used formats), large files, etc. can be posted to that URL. Website: http://www.ieee-pses.org/ Instructions: http://listserv.ieee.org/request/user-guide.html List rules: http://www.ieee-pses.org/listrules.html For help, send mail to the list administrators: Scott Douglas <[email protected]> Mike Cantwell <[email protected]> For policy questions, send mail to: Jim Bacher <[email protected]> David Heald <[email protected]> - ---------------------------------------------------------------- This message is from the IEEE Product Safety Engineering Society emc-pstc discussion list. To post a message to the list, send your e-mail to <[email protected]> All emc-pstc postings are archived and searchable on the web at http://www.ieeecommunities.org/emc-pstc Graphics (in well-used formats), large files, etc. can be posted to that URL. Website: http://www.ieee-pses.org/ Instructions: http://listserv.ieee.org/request/user-guide.html List rules: http://www.ieee-pses.org/listrules.html For help, send mail to the list administrators: Scott Douglas <[email protected]> Mike Cantwell <[email protected]> For policy questions, send mail to: Jim Bacher <[email protected]> David Heald <[email protected]> - ---------------------------------------------------------------- This message is from the IEEE Product Safety Engineering Society emc-pstc discussion list. To post a message to the list, send your e-mail to <[email protected]> All emc-pstc postings are archived and searchable on the web at http://www.ieeecommunities.org/emc-pstc Graphics (in well-used formats), large files, etc. can be posted to that URL. Website: http://www.ieee-pses.org/ Instructions: http://listserv.ieee.org/request/user-guide.html List rules: http://www.ieee-pses.org/listrules.html For help, send mail to the list administrators: Scott Douglas <[email protected]> Mike Cantwell <[email protected]> For policy questions, send mail to: Jim Bacher <[email protected]> David Heald <[email protected]>
