Calculation, Design & LT Spice Simulation of AC Line Filters

hello everyone and welcome on our last day of our digital weay so I'm very happy to welcome all of you to our topic

today calculation design and LT spy simulation of AC line filters and I'm also very welcome uh also very happy to

welcome you Andreas thank you thank you very much for being here of course I'm looking forward to it yeah um so yeah Andreas will hold the presentation today

and also answer your questions those will be at the end of the session you can ask us all your questions in the chat and then yeah later on we will go

through them and if we don't have the time to answer all your questions then we will send you an email with the ANW and yeah if you have any other question

then just email us we are there for you so yeah then also we will share the handout with you so we will also get the presentation you will see and yeah I

think with that that's everything from my side Andreas have fun thank you and yeah enjoy it all right a very warm

welcome from my side as well my name is Andreas Sadler I'm field application engineer in Bavaria Germany and I'm happy to present you this topic today so

all the calculation all the math all the simulation I will show you today is not only valid for AC line filters but also

for dcdc line filters so that means uh unimportant if you calculate a filter for a dcdc converter or for a ACDC converter you can uh use this

calculations and this simulations as well all relevant information I also let's say

comprom comprom in a EP node a new one& 015 you can download this on our web page in German and in English and you will find also the detailed information

which we also share now in this seminar well our agenda for today we will look short at the source of

interference for ACDC and dcdc let's say converters or sources as well as for the required components for filtering I will show you the difference between a one

stage and a two-stage filter and maybe where is the benefit of the two-stage filter compared to the one stage filter I will show you how to calculate this how to simulate this with LT spice

because it's a free simulation tool and it's powerful enough for uh to get a good estimation in the pre-development and I will show you of course then uh at at last the

measurements from the lab in conducted emissions and radiated emissions and of course some let's say Hinds to how to good make a good layout for ACDC filter

rings of course just for completeness um I will show you a brief overview of our uh portfolio here we offer you of course complete line filters like the tfls but

if you want to save some money and would like to include this all on your own PCP of course we offer you the completeness for one phase up to three phase filtering from capacitors from uh single

line chokes fre line chokes and so on just contact us or ask us if you need any support here we are very happy to help you in this topic of course now what's the source of

interference for our demonstration here I choose a dc2dc flyback converter just for a safety reason I do not choose a ACDC so I can touch here anything

anywhere so no problem at all and there's no difference if it's a ACDC or dcdc see because it's in both cases we have just a a higher or lower input voltage but the principles in common

mode and differential mode and all the filter principles are the same we have a switching frequency of 300 khz and an input voltage from 24 and output voltage

of 5 volt at 25 watts so nothing special just a standard isolated flyback converter just for completeness also um a few words about differential mode

common mode mode differential mode the Earth is not included in your system or in the in the current path of the interference and Comm mode is the more

critical and more difficult to filter and to figure out uh because there always the Earth path is included in the return current in your system and this

is most for for the most Engineers the more difficult uh interference current to uh let's say to comply with so

important for differential mode the current paths are like in your schematic that means it's pretty easy to follow we

can filter this with LC Pi T topologies capacitors between the line and the neutral or capacitor between plus and minus and usually the DI to DT

is the dominant uh root cause of the interference current in differential mode so that's pretty easy to filter and to calculate for the most engineers mode

is usually a little bit more tricky because the current paths are not like in your schematic they're always traveling around uh let's say parasitic paths and these paths are usually

capacitive paths so therefore we only talking about usually a few microamps not milliamps or amps it's pretty small currents the return paths are pretty

large and you can fil this only with uh combo chokes y capacitors shielding increasing distance and so on but that's then the only way

you can do it or you decrease your d v 2dt that means you have to switch slower uh the voltage that also means then you lose efficiency in your system and

mainly in most cases it's a radiated issue for many customers but as we have now for today an isolated topology the common mode is also a problem at the

switching frequency as we can see on the next slides so why do we have common mode ISS is here in with an isolated topology because of mainly of the

coupling capacity of capacitance of the Transformer so during the switch uh switch on and switch off transitions of the low side mosfet The High Frequency currents are traveling over this

coupling capacity to the secondary side and from there due to the switch node due to the uh connected cables uh we forming a capacitor from a few picofarads to Earth and then the

currents traveling back to the line impedance stabilization Network and there you get your emission problem or your uh conducted problem because there it's measured and coupl it out to the

Emi receiver as well you can have issues with uh connected heat syns there's also a parasitic capacitance which then acts as a antenna to ground or connection to

ground so with a nonisolated topology like a standard Buck converter usually let's say up to 10 MHz you do not have a common mode issue just over in the just in the higher frequencies then the Comm

mode is getting dominant but with an isolated topology like here um you will have common mode issues at the switching frequency and

above if you look at differential mode differential mode is as I said an issue mostly of the high di todt in combination with uh parasitic inductance that means if you have an input

capacitor with a high ESL like an electrolyte in combination with a high Di todt and the high ESR you will uh let's say reflect a lot of Ripple and

noise to the listen and therefore the listen is then also measured in the listen is measured as your EMC problem then inside the listen we have two 50

ohm resistors and if we now for our experiment would like to know how big is our common mode and how big is our differential mode we need a two line

listener uh that means we need a listen with two parallel usable outputs because in most cases the listen has only one output at the same time and this means

you can only measure differential mode and common mode together not you cannot separate them but if you would really like to size your AC filter in a correct way you need to know the amplitude of

the common mode interference and of the differential mode interference at a certain frequency because then you can really size the Y capacitors the X capacitors the combo choke and the let's

say leakage inductance in a proper way because what you do not want is to size your filter too big there because then you lose money and you lose space on the PCB if you do it too small at the

beginning you have to invest again time during the EMC or after the first EMC test and this also not really convenient for you and your company so if you look

at the differential Uh current path we see the current path in the differential mode the red one is traveling from the the the source of interference to the

listen and is travels uh in with the two resistors in a row so that means we have two resistors 250 ohm resistors in series so for differential

mode the listen impedance is 100 ohm this important then for the simulation and for the measurements for common mode where the traveling path is through the

parasitic capacitance we see the the current mode the common mode is traveling on both lines at the same time in the same direction there is no phe shift between like in differential mode

and in this case the blue uh let's say currents traveling in the 250 ohm resistors in parallel com in in respect to PE that means for the common mode we

have 25 ohm listen impedance which is important for the simulation for example so that means if you would like to measure this with with a scope or uh

with uh let's say with an Hardware combiner we if you want like to measure the uh common mode we have to uh add com the two current from or the two voltages

from both terminals if you would like to have the differential mode we have to uh substract one from each other and as we have uh the listen voltage divider we

need to divide this by two in the end to get the say the correct result so if you have a oscilloscope like the ran schwat mxo or RTA whatever you can do this with

the math channels so you can use channel one channel two to connect them to the RF output of the listens and then make a division by two or use the attenuation

of each channel to attenuate this by two okay now for a filter perspective it's importance from a uh let's say simulation uh point of view if you have

the common mode we have a noise Source impedance which is often not really no um we have the two y capacitors as first uh this is important from the simulation

and also from the layout and from the yeah let's say placement point of view we have to place the Y capacitors always at first from to shunt the most of the

Comm currents back and then we have the common mode impedance uh the main common mode impedance of the combo choke and this is respect to a 25 ohm impedance

inside the listen respected to Earth if we have the differential mode noise Source It's usually the input capacitor of your system so it's usually pretty low uh out in noise Source around 0.1

ohm uh you can if if the capacitor is really at this place then you can really take under one ohm then we have our uh leakage inductance or our let's say

differential mode choke and then we have the x capacitor and then we form the LC filter from another let's say vice versa like in a common mode and the listen

impedance is 100 ohm in this situation important to know is uh we have different comboo cores usually you you need a manganis Sy or nanocrystalline for the lower frequency

ranges for the higher frequency ranges you use nickel sync cores in our example here we took uh we took a manganis sync course you see the difference here manganis sync is better in the low

frequency range where we have the highest amplitudes due to frequency to switching frequency and to the harmonics and in the higher frequency range nickel sink and N crystalline is a good choice and also nanocrystal line in the lower

frequency range is better than the nickel sink so it depend on your noise Spectrum how how to use or which core material you would like to use Al leakage inductance is different the leakage inductance we use in combination

with the x capacitor to filter the differential mode noise so if you have a a core with a a bigger distance from each winding to each other we have a higher leakage inductance compared to

ring cores Just for information also for higher frequency you can use a gapped common mod choke with an air gap that's air gap is inside

here so we lose uh uation from the amplitude point of view but we gain a lot of uh bandwith for the required damping in common mode as in differential

mode important from a layout point of view um it's always good to separate the input filter and the output of a power stage otherwise you can have maybe some

coupling from the input to the output and then your filter is compromised so always try to be as far away as possible or maybe uh make some partial shielding

in this area if possible okay let's start with the the math it's not that complicated uh I promise so uh I this is my setup for my

pre-compliance measurement you see the road oscilloscope here we have a DC Supply an isolation Transformer photoscope to block all common modes

from the grid I have a 50 ohm uh self-made listen 50 micro handy selfmade listen the filter PCB with the single stage and a two-stage filter our flyback

as a noise source and electronic load everything is on the coupling plane to get those perfect results in terms of common mode with comparison to the

lab important if you use y capacitors in your system you need to con connect them as with a low impedance connection as possible that means uh the best thing is

to use if possible ceramic y capacitors and connect them with a very low distance with polygons as short as possible to the protective Earth or to your system coupling plane or virtual LP

what you have in your system depending on what's your application is like not every application has a protective Earth but then maybe you can do a virtual coupling plane under your PCB where you

can shun back the common mode currents locally so this is then measured with my uh test setup in my let's say home lab uh you see the switching frequency at

300 khz and of course all the harmonics up to 30 MHz and I measure differential modes so one channel minus the other channel uh divided by two I measure 77

DB micr volts and the limit is of course or should be Class B so I need a lot of damping at least 30 40 DB here for differential mode the common mode I add

one channel to the other channel so Channel One Plus channel two I see the the com mode is much more dominant here due to the coupling capacity of the

Transformer we see 84 DB microvolt so we need for sure over 40 DB damping here and if you measure both together so now I just measure at one channel of the

listen and I make a 50 ohm uh resistor at the other channel and we can see we getting a little bit higher the the combined one com mode plus differential mode we are around 86 and this is what

you measure usually in the lab so you in the lab you cannot see really the difference between com mode differential mode but in this case you see the common mode is dominating the

system so at least 40 DP of damping to get under the limits of a little bit of Headroom this is our goal in this filter if you use a one stage filter you can

achieve with an LC part at least 40 DB per per decade so a good way is either to uh calculate the required uh damping

at switching frequency this and if you now need the required filter Corner frequency you can rearrange the formula the required Corner frequency of the LC

filter where you get at 40 DB damping per decade because only at FCO this is starting 40 DB per decade damping you can type in here now your switching frequency and the require damping so

that's an our example 4 300 khz and 40p uh required damping a two-stage filter we get ADP per decade at the

corner frequency so uh under we rearranged this formula a little bit we CH we changed the 40 to 80 and then we can calculate the required Corner frequency of the two-stage filter which

will be of course higher that means the filter components for for each stage will be smaller and this is only valid for a two-stage filter with the same

components at two at a two stage okay let's do the calculation we know we need 40db that means with a one stage filter we had 40 DP at 30 khz

because 30 khz is one decade below 300 khz our switching frequency this our FCO is 30 khz so that means uh we are we can

achieve 40 DP if we go one decade below we have to start with something and start with something means you have to Define at least one component let's say

free of in the air but as we all know the Y capacitors are limited use or limiting our Uh current through protective Earth a little bit so because of touching safety we are limited in

most applications to 3 milliamps three and a half milliamps leakage current in our system so I starting usually always with the Y capacitors to size them not too big so 4.7 nofar is a good choice

for starting with the Y capacitors because then usually you stay below 2 milliamps or in uh 60 and 50 Herz grids with the leakage current so we have uh s

they are in parallel when they are connected to Earth we have 2 4.7 by two we have 9.4 nanopar total y

capacitance so now we can uh as we know the corner frequency which is needed for this filter we have to rearrange the formula from the Thompson equation then we therefore we can calculate now the

required common mode inductance with this formula with the 30 khz and the Y capacitors we get three millihenry uh math this is this is not a value which

is common so we choose a 3.3 million and as the all combo chokes has a wide variety of tolerant I always recommend you to stay with a good Headroom Above

This calculated Factor because plus minus 20 plus minus 40 can always happens with a combo choke depending on the core

permeability so on the next step we uh can calculate the actual uh like say expecting damping because we have a little bit higher com mode inductance if you really choose an inductor with this

value we will get around 41 DP with this inductor with this capacitors next step we calculate the leakage inductance of

this uh combo choke this can be done in the with the uh com with the differential mode impedance curve you can use red expert for this and just

divide the impedance by the frequency you read it and it's around for 15 microhenry then at next we can because this is then valid this is a value we need for the

differential mode filtering this the the line filter for the differential mode and now we need a capacitor for the differential mode it's a y capacitor and we also rearrange this with this formula

and we get around 1.9 microfarad in the end we have 2.2 this is the a standard value and we chose this and then we can achieve around in reality with the

components tolerances uh when we are close to the re calculated one around 41 DB the two stage is pretty similar uh we

just have a higher FCO this gives us overall smaller to let's say components we have 80b per decade so that means uh I'm starting with smaller y capacitors

so I get smaller leakage current and smaller and cheaper components so only 2 by 2.2 nanofarad compared to 44 nanofarad then our required combo choke

will not be free m mli Henry it will be only one mandry which which result in a smaller Comm mode choke then our uh damping is here around and in the

reality 480ps because we calculate 0.6 and we get one mhenry or we choose mandry in reality so it's bigger than calculated

therefore we get not 40 DB but 48 next step is the leakage inductance it's around 6.5 micro Henry therefore we can now calculate the x capacitor which is

around then 0.4 microfarad we choose in a 0.56 or 560 nanofarad type which is a a common value here better to stay a

little bit above the calculat than um below and this gives us 45 DB for the differential mode in theory and of course if you if we would like to add

you can add also the differential mode the varistor which is usually used here with a few picofarads and the two y capacitors uh but this is has not big influence in this calculation but you

can add it in the simulation because then the simulated insertion loss has an influence in the higher frequency range if you have a two stage with not

the same component um you have two poles where you have to calculate so that's also not a big issue that means you have for example um a stage with a lower capacitance and a lower inductance for a

higher frequency and for lower frequency damping you have a higher inductance and a higher capacitance and therefore you need to calculate this a little bit separately with this formulas here we do

not go into detail now you can read this in by your own it's not a high high sophisticate map in the end you can also simulate this of LT spice to see at the

two stages so that means we have at the stage with the bigger components we have a lower um FC Corner frequency then you get from this corner frequency 40 DB and with the second Corner frequency with

the smaller components we get the 80P per decade as this is the then valid for a filter with two different stages let's say that's not a big diff uh big issue

to simulate or calculate okay let's go to the filter schematic I with alum I in then let's say developed uh the two filter stages

the upper one is the one stage filter with the calculated values also we see here the varistor included here the Y capacitors the combo choke the x capacitor and the bleeding resistor and

for the two two- stage filter we have the same we have a fuse holder a varistor and the two stages with the same components and the PCB looks like

this for it we see here the combo chog and the Y capacitors which has a the Y capacitors a very low ESL connection to protective Earth the x capacitor the VAR fuse holder and the same at the two

stage filter below we can see the components are much smaller of course maybe a little bit more space after all but you will see in terms of EMC the performance of the two stage filter is

much better simulations we'll see here I can use l spice for it important the first one here is the differential mode

simulation so that means I use 100 ohm listen impedance for both stage for the one stage and the two stage and I use a voltage source uh you can also add in series a resistor if you know your

actual noise Source impedance otherwise we have an ideal voltage source and then we have the common mode the leakage inductance which is uh really effective here with the x capacitor the same with

the second stage here and what we can see this is important we have uh here our calculated damping from the math it's around 40 to

42 and we can see the one stage filter starts uh to Decay uh earlier than the two- stage filter so in the conducted s the radiated Spectrum we can expect a

better performance of the two-stage filter in differential mode the common mode in common mode we I choose a current Source a high impedance a low impedance current Source here

where you can also add in series a resistor if you really know your actual uh impedance of the system this is usually uh depending on the capacitor the parasitic capacitor here to the

coupling plate of the EMC lab and we have the Y capacitors at first stage and then the common mode uh and the second stage of com mode inductance and this is then referred to the 25 ohm listen imp

Ence the same for the two-stage filter components and if you look at the common mode simulation here we also see the calculated values are fitting good to the simulated values and we also see

here the one stage filter is decaying faster and gives you a lower overall accentuation than the two-stage filter from the radiated and the conducted

spectrum of course I can tell you everything uh but I have to prove this is the reason why we measured this here in unic in our uh EMC lab where you can

also if you are here around this area you can book hours here with the support of our Engineers to make pre-compliance measurements and we can see here the conducted emission uh measurement setup

with the coupling plate I used and this is what we measured here this the same like I measured with my test setup in my home lab so I also got there uh with I make the combined measurement Comm mode

plus differential mode 84 DBS and also the whole Spectrum looks good like compares to my pre-compliance setup and if I now add the one stage filter I get

exactly what I calculated and what I simulated around 41 DB of damping so I'm pretty good up to this frequency range but here we see we are pretty close to the average limit so that means in

reality I would not do this with the one stage filter or would at least modify this one stage filter to get this frequency area around 23 megaherz uh

with a higher let's say distance to the limits to the class B limits here but therefore I choose the two-stage filter the two-stage filter gives you what we calculated a little bit more in the

lower frequency range because the components were a little bit bigger than the calculated one but it's fitting well to the simulation into the math and we see the big difference here in the

higher frequency range we are around 30 DB better than the one stage filter uh this is because the smaller components have the better parasitic uh properties

in the high frequency range and the next setup or the next measurement was the radiated setup we also made in our EMC chamber here in

Munich so you can see the antenna here and absorber and here the test set up of really long cables or one 2 m cables so not not shielded nothing really open and

this is when we have no filter at all so the flyback with no filtering we see a Broadband uh interference which is typical for switch more power supplies

we we do not see single spikes we see a Broadband uh let's say uh interference this is valid for many switch Mo power supplies as they look pretty much the same depending on uh their cable length

and so on it's a little bit different amplitude and frequency but it's all Broadband so if it add the one stage filter we are under the lines uh

vertical horizontal and quasi Peak so everything is fine with a little bit less head room in this area maybe if we took the two- stage filter here we get

the required Headroom around 8 to 10 theb over the whole frequency range and this is also or a good idea or this also shows you why it's a good de maybe to

take sometimes a two-stage filter compared to a one stage filter so where can be possible deviations between your measurement your math in your simulation of course one

big topic is the the tolerance of the component especially Mo chokes but also X capacitors older ones maybe have bigger tolerance the PCP layout has a

big influence so the near field coupling if the components are too close to each other or too close to the power stage the grounding of the YC caps is not good

enough in terms of ESL so a lot of possible variations why your math and your simulating is not fitting to the reality and last but not least you can

have from us some filter components uh so already we could offer you a filter set with bare pcps where you can assemble by yourself for ACDC as well

for dcdc and this is available with this order code just go to your local wood electronic partner and say you need some of this design kits and that's all from

my side I hope you enjoyed the session and you learned something you can use for the practice and I'm open and happy for your questions yeah then Andreas

thank you very much for the presentation a lot of information but very cool and we also got some questions I see first of all I now want to share the handout

with you so you are now able to download the slides and yeah then enough for me for that so then let's just start into

the into your questions and here we see the first one Andreas how do you handle cooling is a two- stage filter easier to

cool since the components are smaller or more distributed cooling first to say you should you should you should size your

filter that is not required to cool of course because mainly uh it's a RDC topic here so the AC losses in the filter are usually not the big issue

compared to power stage itself so uh but yeah if you want to cool this um I think it makes no real difference if a two-stage or one stage filter a good way to cool a filter is to uh use potting uh

some potting materials because it's a better termal resistance than uh normal still air let's say so but yeah from a pure cooling point of view maybe it

makes not a big difference then maybe it's better to choose components in the from a capacitor point of view and um combo choke point of view which are in the equal size then it's easier to putt

them or cool them with a with a fan for example but yeah I would from the beginning I would them from an RDC point of view that is not required to cool

them okay thank you very much then the next uh question is regarding the slide 18 uh is isolation Transformer effective

enough for rejection of higher frequency common mode noise from Mains due to Capac capacitance so always a yeah my

colleague help a little bit of the slide here ah perfect okay uh the coupling capacitance is a issued during the design process of the Transformer we also with our daughter company midcom

designed this Transformers and there are are two aspects designing a Transformer from the EMC point of view is the leakage inductance and the parasitic coupling capacitance primary to secondary and both influence each other

in influence it's each other so that means you have if you would to achieve a lower leakage inductance then you will achieve mostly a higher coupling capacity because then you have to

interwine primary and second secondary in a very close manner B manner let's say so if you have done a Transformer with a low coupling capacity primary to secondary most of the time your leakage

inductance is higher because then you separate more primary to secondary winding so it's always a little bit kind of trade-off a good idea is then to use the LLC topology because with the LLC

Transformer you use the leakage inductance as a part of the power stage and you can design then a Transformer with the lowest possible uh parasitic capacitance from primary to secondary so

but on on the most uh sides they say higher leakage inductance will lead to a higher voltage Spike and and higher problems in the lower frequency range and and the coupling capacity leads

maybe then also in the higher frequency and the lower frequency range depending on common mode so difficult to say and there's no perfect uh world for this to be

honest okay and yes thank you very much you're welcome so then let's just go on in what case we need um different values

of uh CX LCM or centim probably I don't know in two stage filter yeah and this the question where I showed you if you have a LC two- stage filter with

different values so this is maybe then a good choice if you have a A broadb or system which has a Broadband Emi interference that means you not only have big issues in the common mode in

the conducted range but always in the radiated range but so this maybe can be a good idea to say okay okay for the conducted range I choose the Mist sync and a bigger capacitor and for the

radiated one I choose the N uh the nanocrystal line the nickel sink or the air gap combo choke in combination with a smaller capacitor so depending really on the Emi signature of your device so

because if I only if I have really big combo chokes uh for example I do not in my example I had one one one mandry chokes these are pretty small compared

to other combo chokes but if you for example need to 10 mli Henry ohms the influence in the very high frequency range is pretty poor of them so maybe then this could be a good choice to uh

skip one of the 10 mli Henry chokes and add a smaller choke here for the higher frequency range because the parasitic capacitance is smaller and therefore the

frequency range which the core and the winding works as an filter is higher okay then let's go on with the next question so I see there are a lot

of questions coming in that's really nice to see a lot of interest here uh if we can't answer all your questions now within this live weinar we will answer

them later on via email or yes Andreas will do I will answer every question so next question Mars are also required for

most main filters is there any help with the design yes in this EP note I showed you at the beginning I detail I show you detailed how to calculate the varistor

for uh one phase system for the three-phase system is pretty much the same just maybe the search uh values are higher not 2 kilo volt maybe 4 Kilts um

important is always yeah to size the ver not too small from a diameter point of view so 14 to 20 mm is always required and then an important point is also how

where you place this varistor it's not maybe only a good choice to place the varisto is like I did at the first place after the fuse because then maybe get uh

during the search event depending on your inductive uh uh components in the system a higher voltage Spike so in many applications can be also a good idea to

place the Vester behind the input filter in front of the for example Bridge rectifier then has a better uh influence on the inductor Spike or sometimes also

required to place two vestors one before one after or damp this system with a capacitor with a which has a little bit lossy

ESR but as I said in the& 015 in the new one you can see how to calculate this VOR okay then here uh next question from

ysf uh the SM smps had a constant load or a dynamic load so a switch mode power supply has a constant load so in the end

uh you should do if you have only one uh let's say working mode so when it's only continuous conduction mode mode and then usually the good choice is to choose the lowest input voltage and the highest

output current as a worst case but if you have different working low modes due to light load you have burst mode or DCM mode in your uh switch mode power supply then you should do more than one

measurement to really overcome all the the working points of your system so it's because not always the highest output current means the highest the highest EMC uh let's say noise signature

in your measurement okay thank you very much then we have a very open question here how to recognize a common mode or a differential mode

noise yes therefore you can use uh there are more possibilities you can use a current clamp from Rand Schwartz with the required bandwith you uh feed both

cables uh life neutral through the com through the the current uh the current clamp then you measure the common mode if you only measure at one of the two cables uh you have the differential mode

over over the say the Spectrum or you use uh the two line listen I used here with a dual use output then you can see with the scope on for example uh channel

one channel two minus each other uh you have the differential mode with the with the math Channel then and channel one channel two plus each other is the common mode on the second math channel

for example then you see pretty good uh where is the the main interference at which frequency good okay so we still have a

little bit more time a lot of answers coming thank you very much uh so then let's go on is that is it possible to

use LT spies for predicting radiated emission uh so common let's say to predict conducted emissions uh yes we

have already made a good video YouTube for it uh anti anticipate EMC with LTI spice is the name for it so there you can see uh for up to 30 mehz is pretty

good then if you have for example Automotive environment cisper 25 you can have a listen which measure up to8 mahz then it's also a good Poss to predict

here otherwise it's getting a little bit difficult because in the over 30 MHz for the standard industry you uh measure the radiated spectrum and then your cable

length and the cable Arrangement has a big influence here and that's a little bit difficult but what you can say is um if I'm getting in my simulation or in my measurements here on the test T up on on

the bench I can measure here with some with the DC listen we also made together Ro WS you can uh order this by uh elector for € 100 this DC listen 200 MHz

Spectrum if you see there uh uh Improvement of 20 DP at 100 mahz on your test bench you can really uh have a good a good result nearly to 20 DP at the

radiated Spectrum in the chamber let's say this correlates pretty good with each other but but otherwise it's a little bit difficult because the cable has a biggest influence and the impedance of the cable is depending on

the length and uh depending on which part of the cable you measure you have different um let's say standing waves and therefore different amplitudes yeah

easiest way is to have a listen here okay next question is from Far hard uh how to calculate and design the filter

up front before designing the power supply yeah like I did here always make a measurement without any filtering just make the pure power stage maybe with a

coupling capacitor primary to secondary yes or some other um onboard uh let's say measurement but if I would always do a first measurements on the testbench

like I did here with my scope in the coupling plate without any filter components because only then you really know how big is the uh the common mode and the differential mode amplitudes and

after that you can start calculating because then you know how much damping you will need at which frequency because then you can start determining your corner frequency start designing with

one or two stage filter or whatever okay thank you very much welcome next question from Maran do we need any damping components in line

filters like input filter for buck dcdc um as I told you maybe in the with a uh bar restorer it's can be necessary

to damp this that you have in very high and you have an let's say com mode inductance with a very high leakage inductance with a very high quality capacitor then you have two components with very high quality factors then

maybe you can get a high ringing frequency or bringing amplitude to the corner frequency of this filter then maybe it's really necessary to damp this a little bit uh with uh serious resistor

of a second uh or of additional x capacitor um but from step ility point of view like we look at an uh let's say negative input impedance as a dcdc

converter where this can happen oscillation this is usually not necessary in most cases for the Line Filter and it's also not the correct place maybe to damp an let's say

negative input impedance there you Dam this usually at the input capacitor at the switch move power supply okay so then I think we're coming

to the last questions um why are there damping resistors in the LC filters uh this what the same what I

told you at the moment when you have no if I add damping resistors um then maybe I compromise if uh if I have just one filter capacitor and put a damping resistor in series then I will

compromise my filter performance of this capacitor of course but I can add a second capacitor with a high ESR electrolyte type but this is a little bit difficult because the input line

filters have need to have vde certificates UL certificates need to withstand some search voltages and so on so the components here compared to a

pure dcdc filter they're only the the biggest difference is safety the compliance of this components and the voltage rating of these components so

not that easy okay so then last question um how to compare the quality of Emi

Cho uh the quality let's say the quality is maybe not the right uh point to look at the impedance curve is interesting because in the end it doesn't matter how

much millihenry or microhenry or microfarad the capacitor or the choke has the impedance curve is interesting I need to have the filter choke a certain

higher impedance than my system and with Filter capacitor I need a certain lower impedance at the requ ired frequency and this is interesting so this is the

reason why maybe an electroly capacitor is not a good choice because his impedance curve is dominated by the ESR in broad range and therefore it's not a good filter capacitor it's good for

damping good for hold up time because of high capacitance but not for filtering okay so always look at the impedance curves not at the pure values because the impedance curve like in our

comot also starting to Decay at certain frequency because of the frequency depend dependent permeability of the core and maybe the 3 mli Henry or 3.3 M

mhenry you do not have at 5 MHz it's only valed between 100 khz and 2 mahz and then all then only then your calculation is valid if you really have in this frequency range where you

calculate this uh the real the the component can hold his impedance okay so then Andreas thank you very much for all your answers you're

welcome and also your presentation and thank you very much for your attention I hope you enjoyed our session today out of Munich and yeah then I would say stay

tuned we have some topics coming up and also with our partners today and yeah then I would say have a nice day thank

you very much and yeah goodbye goodbye