Learn To Fix EMC Problem Easily And In Your Lab - Troubleshooting Radiated Emissions | Min Zhang

in this video we are going to talk about how to troubleshoot emc radiated emissions problems how to troubleshoot them in your own lab and this can be super useful if for example you have a product which is failing this emc radiating emissions test or it can be super useful if you are designing a product which will need to go through this test but you are not sure if it's going to fail or pass because if you can measure this in your own lab then it can save you a lot of money i'm not an expert for emc so i had a call with min i recorded this call and that's what you will see in this video as an example in this video we are going to use this led lamp which has huge emc radiated emissions problem you can see it here on the spectrum analyzer and i really like the call with main because you will see the complete process of troubleshooting this product and initially there there will be this like we are trying to fix it and nothing works and then suddenly we will find the problem and like this everything will start working if you don't have much time i created chapters you will find them in the description so you can quickly jump to specific topic what you are interested about but this is not very long video uh so if you watch it like in double speed it will be like half an hour and you will learn a lot of things uh which i hope you will find interesting and useful okay that's everything for introduction and we are going to start our call with turning on the led lamp and checking the radiated emissions by this spectrum analyzer watch this purple line and when the led lamp will be on you will see a lot of noise or many peaks on this purple line so here is my call with me [Music] and you really can buy this in a shop or on internet yes let's just quickly have a look at the setup we just demonstrated it's very easy we have a lamp and the lamp is as i said it's a mains product so it's plugged into the socket so in eu we're talking about 230 volts in the us 110 volts and the setup we had we have a little antenna this antenna is pointing towards the uh the lamp is it important have you put the antenna horizontally or vertically yes it is important normally you would put it horizontal and then sorry horizontal and then vertical um but because we are trying to fix something we if we measure it in a vertical direction as as we show here as long as we can fix it in always using the vertical that is it's adequate basically so but maybe initially you try both the directions and you will see which one is more noisy and then that's the one which you will use yes yes that's a very good method because as i said this is a very simple pre-compliance test setup which i hope lots of people can reproduce this in their office or in a lab but if you send this to uh a credit emc lab they will have three meter distance with big um you know uh antenna and then they will do it properly yeah uh can you buy these antennas or uh where did you get the antenna yes so this particular antenna i use i'll show you in the next slide yeah so this is the simple setup i just mentioned you have the lamp put in here and then i have a one meter distance and then you can see i marked everything using tapes because this is quite important radiating mission can can be different with your with slightly different uh setup so make sure that your setup always be very consistent and this little antenna i used is from an rf engineer he had a small company and he designed this pcb based antenna but in reality you can use any type of antenna you can find for example remember the uh little radio station we had that uh antenna is called rod antenna you can use that antenna and it's the same you will pick up noise straight away because the area only happens the noise too exactly normally different antennas will have different signal reception range for example this antenna is supposed to have very good reception between 400 megahertz and one gigahertz but as you can see later on for this noise which is predominantly between 10 megahertz and 300 megahertz this antenna still can receive um the nose so the point is if you can have any sort of antenna even like a tv antenna you pick up the noise and then you make some change and then you see some 10 db 20 db reduction then you know you you you you basically reduce the noise but ideally you would like to have white uh range antenna so you can i think in emc test house they even change the antennas exactly so you can you can do measurements for one range and then when you are happy with the results you can change the antenna and check if there is nothing in different frequency range yes yes so um you're right um the types of antenna can be many you know you have the log periodic so like this shape and then you have the bicone nickel one you have the home one you have uh load antenna many many types of antenna work from different frequencies as we said where is the one meter between where exactly it is so this one is from this point as you can see this uh this green and tape i had to this point is about one meter so these are just like marks with you which you have like reference points exactly you can put in 1.5 meter or you know but yeah roughly normally for a pre-compliance cmz test put into one meter is a good distance okay uh and i'm i'm just this is maybe a weird question but uh i remember i was talking about this uh with someone else uh usually the distance in the test house is so large because we would like to measure the far field so it means we really don't want to go too close because the near field may not be exactly what we would like to measure or is there some kind of relationship between like near field and far field or what is your opinion about this yes you are absolutely right um the idea of testing uh equipment on a product using three meter or ten meter distance is exactly the reason as you said that you don't want the near fuse to to affect your results so basically what we're seeing is when you know it's very similar to lots of videos you did with other people and talking about the energy traveled in space and when you have energy electromagnetic energy travel in space it's from the far field point of view the electrical energy and magnetic energy should be the same but when you get too close to a product it could be predominantly electric field or it could be magnetic field dominant so you don't want to for radiating the emission point of you don't want to get too close that's why i said uh normally one meter is just on the boundary you don't want to put it too close yeah one meter and then there are some equations that can you know adjust your results to say three meter value but it's not in the scope of today's discussion okay but i wanted to mention this so everyone knows that maybe they don't want to measure 20 centimeters from from from the board because they may get maybe not the results what they are looking for that's exactly yeah we should mention this yes yeah this is a very good point okay so i see spectrum analyzer okay uh tech box emc view software okay and it's just a software that we use for scanning and yeah and it's very simple test setup um as i said if you have a spectral analyzer and some some form of antennas then you can easily troubleshoot basically yeah okay i would like to go back to the video because i'm not sure if we if we pointed out uh the noise ah okay yeah sure okay yeah so as you can see before i play the video you can see that i already had the test setup and you can see on the uh let me just zoom in the uh on the on the spectrum analyzer i have two channels basically the first channel is the yellow trace and the yellow trace is i do uh a scan basically without plugging the uh the led light so the led light was not plugged in it is where this is what you call ambient and that's these are basically all the noises picked up by the antenna from i don't know like wifi or something i don't know why yes you get it right the the peaks here probably one of the peaks is actually there's a very i think it's about um 12 or 14 kilometers away there's a big transmission and broadcasting station for bbc so you always have this uh signals and yeah this is the ambient and once we plug in the the product and we power the led then you will see all the noise going up um yeah this is just basically the radiation emission we measured using the uh antennas here okay so can we play the game so everyone can see it yeah sure [Music] okay now we can see the purple that's the noise what the lamp is making yes okay yes you can see the spectrum was pushing high and then also the the the the radio station has been heavily affected basically the music you listen to has all the noise yeah okay so what is the next step we know there is something wrong with the lamp what we are going to do next yes so before we do a full scale operation on this product you know like often you send a product let's say this this lamp is our design we design this lamp and we send it to a emc test lab and then they test and say oh no no it's not good and what normally the emc lab engineers you know test engineers would always give you some advice um these are the advice probably not from the design aspect perspective but they are quite practical so for for example in this case and they will probably tell you okay let's uh put some ferrite on the cable or let's shorten the cable so let's just have a look at what this means okay okay normally radiation occurs um because there is a some cable structure and cable can radiate quite easily so um as a general advice normally people will say let's try to shorten the cable and to see if we can bring down the noise so shorten the cable not normally is what they suggest in the first first step okay and that's what we are going to try now yes exactly okay so the results you see in this video i also basically save it so here's uh save save the results you can see on the on this side as you can see when we shorten the cable we bring down the noise if you look at this point from here 87 to 67 more than 20 db but you see the noise only is brought down from the frequency from so here is 10 megahertz 20 30 40. so from 40 megahertz to maybe 16 hertz we had some noise reduction by shorting the cable and this is because the original one and half meter is good antenna for these frequencies good guess good good yes so i put in here um so exactly as you just said let's go to this page so in this page i basically put in more information here the the electronic circuit is located in a very little pcb on the on the socket or plug and then it has a long cable as i said it's about 1.9 meter long okay and then the lamp is basically led diodes so uh so if you work out a cable lens of 1.9 meters long and then often a very strong resonance point is a quarter wave length so assuming the you know the speed of light is 3 to the 10 power of 8 meter per second using the equation v equals lambda multiplied by f and you can work out at a quarter wavelength at which frequency is resonance very strongly so in this case we have 1.9 meters and a quarter wavelength is 0.5 meters and then we calculated at which frequency resonance very strongly which is 40 megahertz so that's why when we see we shorten the the keyboard length the frequency in 40 megahertz range brought down by 20 db yes was it 40 megahertz wasn't it a little bit higher it's starting from 40 so you can look at here so this is 10 20 30 40. yeah okay yes yes actually this is this picture here yeah 10 20 30 40. yeah so from 40 you can see very very big reduction and why is it a quarter of the length why not i think many people may wonder why it's not half yeah you can't have half wavelengths radiates very strongly or quarter length wavelengths radius very strongly as i put in on on this very simple um diagram here i guess half is for like dipole no yes yes so um if you have a noise source let's say sit in the middle in this case and then you have you know cable going both directions and then that cable lens is halfway but most of the time we find you know you have a product like here and then you only have one cable and that cable is is a quarter wavelength basically okay i understand okay we can go back but yeah but the idea is doesn't have to be a quarter wavelength or half wavelength cable at any distance can radiate it's just you know adds a quarter wave lens at that frequency for ultimakers in this case it's really it's the strongest but other other frequency also radiates using the same cable basically i guess also the like multipliers of these frequencies will be resona will be radiating very well yeah when it comes to um you know radiation and particularly antenna theory is we can talk a whole book of this thick talking about the things we talk yeah it's a very very complicated uh science of physics yeah lots of science behind it yeah okay we can go back yeah okay so we the first thing we just tried is we tried to shorten the the cable lens and we saw the effect and we get a little bit understanding of okay you know the cable radiates and then uh and the second thing again often um you know engineers will find in a emc test uh lab is they will say oh if a cable radiates let's put some ferrite on the table yeah we know you know from the little test we just did we will notice that from 10 megahertz to 300 megahertz we saw lots of noise so when you choose a ferrite it's not randomly you shouldn't randomly choose a ferrite you should choose a ferrite that has a lot of impedance in this frequency range i think that's more important so in this case so in this case uh we we choose uh a ferrolite as you can see from 100 megahertz to 500 megahertz we have a very good impedance you know 534 ohms so we pick up this ferrite and then let's see what happens so you place it close to the adapter yes i put it close to the adapter yes because i already got the feeling that the noise source comes from that side but i will show you the details later yeah so uh let's have a look at the comparison results okay and if you if you would place it okay if the source is the adapter and if you would place the uh ferrite uh far away then it would not have as good effect as if you place it closer so this could be also a way to find out uh what exactly is making the problems yes actually you reached a very very good question sometimes not only say you put the ferrite in the wrong place as you said you don't know where the noise source is and then you put it far away from the oil sauce and sometimes what people found is actually they put the ferrite and then the noise increased yeah the noise increased and it's getting even worse uh and this is not unusual this is what happens um and the simple explanation for that is because noise travels and then if if the ferrite is put too far away sometimes it can reflect it back really depends on the frequency there are many papers talking about it i think in in some particular case where you know you have two systems connected by a cable and one system is working at one frequency the other system working at double that frequency then it is very tricky to put where to put the ferrites in the cable i can't remember from the top of my head but i remember if you put in the wrong place you actually see the noise and fixed yes so uh but generally when you place it further from the source you will not get as good results if you place it close to the source yes because at the time when you put it too far away the energy has already leaked into into the space basically okay so let's have a look what happened so here as you can see here we have uh the purple line is the original scan okay where you know without any uh methods without any approach then the uh the blue one here is when we shorten the cable we already knew uh from 40 megahertz upwards we have some uh reduction and then if you look at the green and the green phase you can see it's actually starting from uh i think 10 megahertz yeah from 10 megahertz onwards to uh 100 megahertz it has you know it's 10 mega 10 megahertz is the first one so because this is a logarithmic plotting so i stopped at one gigahertz so i know this is uh 100 megahertz okay and this is 10 megahertz okay this is one universe yeah okay so yeah so um yeah yeah so basically you know we know that's both methods work um but the idea is not really to put a big ferrite and shorten the cable because imagine if we we are developing this lamp our boss probably wouldn't allow us to say oh you design a lamp with this long cable or you design a lamp but you tell the customer that you have to buy a ferrite so yeah so i think the next step for us is to look inside what's inside and can we fix the problem inside okay um but i prepared some slides basically to to to explain well what caused the radiation okay um so again it's very similar to the discussion we had last time we're talking about common mode current and differential mode current and as you can see here most of the radiation happens when you have common load current uh in your in the in the in a in a energy traveling path so in this diagram you can see that if i had common mode current traveling on the cable and then you can see the the current will form a loop and then this loop basically is how radiation common mode current is the current which flows kind of outside of our system and the differential mode current it flows in the system and in the connection to the system yes okay yes exactly and um and basically you know we're talking about energy traveled in space you know many times so basically we're saying that same example but if we forget about current it's really this area that is a loop you know between the cables and and the ground or any metal structure you have a loop and that loop can radiate we will we will come to this slide later on with details but um this slides basically is more detailed because i put the uh the circuit into into the drawing as well and basically what's saying is common mode current travels through all the parasitics you know if you have a transformer you have parasitic capacitance and then the product itself will have some para parasitic capacitance to the earth and so current will will always flow will flow um using the uh least impedance path and as long as we have this common load current path we will have radiation basically but we will come back to these these circuits later on in details when we when we dig into the product okay so um the next step really is to uh you know open up the product and then see what's inside and can we fix the problem inside so we're going to have a closer look at the lamp and so how it is connected i can see we were talking about this so there is led directly in the lamp then there is long cable connected to adapter and the adapter is connected to the main so basically the led is pathetic with dc low voltage correct from the adapter yes so i i don't know you know what's the led line this particular case is about until i open it up so um so this is how you know when i opened the little black uh socket plug basically open it and i had a look and i was like ah okay so this is how they they they made so basically the whole circuit as you can see here is a very small tiny pcb and its own job is basically changing the ac voltage into a dc voltage it's uh it's it's like this it's uh it's a flyback uh switch mode power supply basically you have the ac in you have a rectifier you have a dc link capacitor and then you have a flight back transformer which is essentially two inductors and then uh you have some a diode and uh a capacitor and then here is the led and the cable really is just uh what 1.9 meter long cable but only carry dc dc uh voltage low voltage dc yeah okay i see you have the picture operation principle of flyback converter yes so just in case some people are not familiar with flyback switch mode power supply you can visit um i i just picked up uh texas instrument they have this very nice uh explanation very simple one so you can understand it so this adapter is changing the 230 volts to 12 volts and this is how it looks inside okay yes but before they changing to 12 volts the first thing is they rectify it so ac becomes dc and dc is about 400 volts high voltage and then then through the uh flyback circuits to lower voltage so um yeah okay and when they make it dc then there is this special ic which makes from the dc some kind of switching uh voltage which goes through the transformer correct yes so uh i see it's just like a switch basically it's uh it's what we call isolated dc-dc converter you know like the the buck converter we talked last time is non-isolated and this one is isolated because it's using transformer yeah you use the transformer for isolation purposes okay because sometimes you're dealing with ac is dangerous and then normally they put isolated power supply yeah okay yeah and yeah so um here is another test setup i used for troubleshooting uh radiated emission and it's a very useful one because the problem is you know the the antenna setup we had is nice but it's taking space and every time you make some change you have to make sure that the antenna always stay and it's just not um efficient to to to solve radiation emission problems one of the efficient ways of solving radiating emission is to use something called a temp cell as as i show here so this this one basically is you can treat it as a very small scale um like a like a chamber because then you can put your pcb basically inside and then you do a scan and then you find where the the problem is and then you can quickly get the pcb out work on the same bench and put it in and then do another scan so very quick uh troubleshooting tools but uh then is not this working with the near field radiation it's yes that's a good question i'm sure it will pick up some mere feuds but um the the way it works is more suitable for troubleshooting radiative emissions yes so because probably if you if you fix this field in this small temp cell then it will also fix the big radiation correct yes yes okay and you have how is it connected the cable is connected so so because because it's a good question because because we know that the original product the ac cable is very very short because it has the plug and then long dc cable um and so i extended the ac cables by another 10 centimeters simply because i can then put the pcb in the 10 cell so you can see the plug is here but i get the pcb out and i extended it by 10 centimeters yeah okay i understand i think yeah yeah i think we should mention this otherwise yeah this is not clear because yeah people will think oh you still have the plug here but this plug is is nothing it's just just ac lines so the pcb is now sitting in here how this stem cell looks inside when it is open it is just metal or it's a metal it's two metal on top and bottom and in the middle they have some pcb structure and basically it it's a transmission line basically it's a it's a a transmission line that can pick up noise okay so on one side there is a termination or something on the other side you are connecting the spectrum analyzer yes exactly same way as it works as a transmission line you have a 50 ohm termination yes standard 50 ohm termination and then the uh the output is connecting to the spectrum analyzer yes okay so now we are going to confirm our theory that the a lot of problem is caused by this uh yes power supply yes so uh yeah again first thing is we we need to do uh a benchmark so this is a benchmark test with the tensile results so if you look at the the the problem is again from 10 megahertz to maybe 300 megahertz and it's at this time it's 47 db micro volts okay so that's the benchmark level we're aiming to bring this down down down hopefully by 30 db that's our our goal okay yeah why there are three different graphs orientation one two and three oh okay yeah this one basically again it's uh it's something you know very similar to uh to what we discussed before for emc testing always make sure that you test different uh orientation it's like antenna as we discussed you have to test horizontal and vertical so put things in temp cell because the weight works you also want to triple check you know this direction this direction and this direction making sure that you know as you you put it always pick up the worst case and then fix on the worst case yeah same idea basically okay but the results are very similar i can see on the graph yes the results are very similar so at least from what i understand is if the orientation doesn't make any difference then probably it's because the cable cable radiation because the cable regardless where you put you always please okay yeah so um okay so now we know that there are some problems on the pcb yes and we would like to try to fix something that's right so what is going to be the first thing what we are going to try capacitors yes capacitors my favorites um yes i think uh from the last time when we discussed the conductive emission we already discussed many aspects of capacitors so the principle is again very similar is the the dc link capacitor is very important for switch mode power supply because it is the main energy source and if you want to have less radiated problems and less conductivity problems always make sure that your dc link is good so in this case there are a few problems with this design first is i think their their capacitors value is very small okay and uh you can see from this pcb you can see the the electrolytic capacitor is banded yeah so it's not really sits on on top of the pcb spend it and the reason for bandits is because the they want to pack everything into the um more space yeah but this is one of the problems i i've seen in many other cases is when you bend a capacitor this bending you can see this is a long long lease and this will introduce extra inductance so this is not good good exercise but sometimes engineers have no choice you know because the height of the product is is defined already by product engineers and then you know you can't have have otherwise the capacity will be too tall so you have to bend it so that's a that's a that's an introduction of some um uh parasitic inductance let's say i can see on the other picture you i put it on yes i put it straight yeah and i i chose a larger one so i changed 4.7 microfarads and then i changed it to two a 10 microfarads okay so that's the first thing we did and this is the improvement we made you can see yeah so original cap is the yellow trace and the purple trace you have about uh 10 less than 10 db improvement from um from 10 megahertz or maybe less than 10 megahertz you can see yeah and then so there is little improvement but not like what we would need yes but you know this product we know that is the the emission level is so high so uh what we want to do is every step we bring down maybe 10 db so apply five steps we will get 50 db basically that's the idea okay i understand yeah it's it's about 10 db i would say 10 db reduction okay that's a good good good news okay then uh then the next step the next step is again trying to optimize the dc link a little bit better so in this case we we the first we will change the capacitance value but it's still an electrolytic capacitor and next i wanted to use some low impedance film capacitors so uh i tried to put a 10 nanofarad capacitor in the dc link yeah as i draw here and the idea is the same basically you want to make your dc link capacitance a low impedance path basically and alternatively because it's the x type capacitor it can also work on the ac side you can put the same capacitor between the ac lines that's also uh many products do this that puts the capacitor in between ac trying to provide some low impedance path but uh in this case i put the capacitor in dc i put the capacitor in ac they actually make things worse as you can see yeah they make things worse so what we're saying is we tried two methods which in theory is should work but actually in reality it makes things worse okay and why so i think the next question probably why it didn't work you know is that making the dc link low impedance it should work but why it doesn't work i think again this topic you probably haven't already had a discussion with many other people in the past but it's it's worth uh see it's really how how in reality it works is um the the x type capacitor we choose is a 10 nanofarad capacitor so you can see here is 33 nanofarads here is 6.2 nanofarad so the resonance point of this capacitor is is somewhere here okay somewhere here and but because the layout of this pcb is terrible which we'll see later on which means the location i put these capacitors is not ideal so when i put it there i already introduced some inductance in the in the system and when you put inductance okay so this is the simulation i did i have a 12 nano farad capacitor let's say but i introduced another 10 nano hemi inductance 20 okay so so before it's 10 and then i added another 10 gives you 20. yeah and this is the result as you can see so this is 10 the dashed line here is 10 nano henry originally but this red solid line is 10 plus 10 and the other 10 is basically the parasitic inductors we talked about okay what does it mean what does it mean it means you what you you think yourself okay i look at this dashed line i think oh you know at 15 megahertz i should have very low impedance yeah you see a residence point and you see oh it's very low it should work but because of the extra inductance we bring in the system this is the reality it's not low anymore you look at the impedance here and the impedance here you can see a 10 times difference so i think that's a very good point for anyone to take is you spend a lot of time looking at the data sheet of a capacitor but then you put in a circuit but the circuit actually introduced more inductance than you thought so you you think you pick up a capacitor for i don't know 15 megahertz but once you connect it to your circuit because of the leads to the circuit it actually may not work for 15 megahertz megahertz but for something lower and in 15 megahertz it can be even worse because for 15 megahertz instead of placing their capacitor you place their inductor exactly exactly yeah that's the point we'll try to make okay okay so um that's basically we tried all the methods of capacitor but because of the layout we cannot change the layout of this small pcb so we are limited basically but we know that um the 10 microfarads electrolytic capacitor helps that's the first step okay and uh the second step is to shield the transformer okay so as you can see here i made a small shield and i put on top of the transformer and the shoot as i demonstrated here needs to be bonded or grounded as some people say to uh lowest impedance path because in this system there's no real low impedance path because of the layout is so bad so we can only choose a relatively low impedance path which is um the negative vdc so that's how i how did you make this shield so this shield you can you can make this shield very easily you can uh use copper tape you know um but in this case i have something from worth electronics as you can see we sell this um little shield sheet and it's with all this little um square and then you can easily bend them to any shape you want so i use it i've never seen anything like this yeah yeah it's a very useful uh shielding sheet so you'll see you cut the shape what you need you bend it and then you use the copper tape to put everything together yes that's how i did it yes yeah and you can solder on the copper tape or you saw you can solder on the shield yeah you should i basically i soldered a very tiny uh wire from between the the shield and the uh minus v dc so you you can solder on the shield yes you can okay yeah okay um is it expensive i would say this this this shield i think is it's not not expensive i i think it's less than 10 pounds for a big sheet yeah okay and uh um yeah i would say this method 90 works but in our case it doesn't work okay i think this is an interesting part it doesn't work but we'll talk about why well yeah so the next step is important to say well fix it yeah yeah um um yeah i think one of the the the you know great tips from keith armstrong you know you recently did the interview he he gave is he said you know if you make change step-by-step change on some product uh and you still one step a change doesn't help second stage doesn't help seriously didn't help he said as long as they are not making things worse you shouldn't get away with them leave them there leave them there okay so as you can see now we're in step three so step two doesn't work but i still have the shield so i still have the shape and uh and next thing i'm doing is i'm adding some y capacitor and the y capacitor works very effectively for common modes noise as i demonstrate here we know that the common mode current travels on both lines so if you have common mode current going through and you're live and neutral and then you put two white capacitor and y capacitor again in this case also provides a low impedance path so they provide the low impedance path and then your common mode current should be going back should be returning via these y capacitors and this is a very effective way of making a filter so i i put two white capacitors uh in you know in the system like like like like it is here um but again no no improvement yeah no improvement this is getting more interesting um and another thing we tried is we said okay lots of the methods we tried is on the high voltage side yeah you look at all the high voltage sides we haven't tried anything on the low voltage side so uh i put some ceramic capacitors which again supposed to work really well in a frequency range were interested so two capacitors one works well between 1 megahertz and 10 the other works well around 100 megahertz i put it on the on the low voltage side no no improvement okay so step four we're still stuck okay so as you can see all the all the results we we put in here the the purple is still our original scan but then all the rest is just stay there didn't make any improvement maybe even a little bit worse either even a little bit worse okay and uh i think that's one of the advice i want to give to all engineers when you have problems like this stay calm drink some tea maybe sleep come back the following day and then restart never get frustrated um okay so i i basically um stopped and then i i come back later i said okay maybe we should you know we we look at the circuit and we think oh yeah we fly flyback switch more power supply but we really we didn't go into the details so this time we went to the details and this is you knew there is probably something really wrong with the circuit itself indeed so um this time i really studied the whole pcb component by component and i marked all the components here and i draw the circuit diagram basically as the pcb layout and here is what we found it is uh the iic this little ic has seven pin and basically i found a seven pin device somewhere on the internet and i think it's very similar to this so i basically draw where they had and you can see here the hv pin basically these are the switch pin to this uh transformer and the vdd of course getting some voltage through this potential divider and the they have the feedback pin but in this case they never use and what's strange is they have a c3 capacitor basically connecting the vdd pin to the secondary side the low voltage side um diode and this c3 in my opinion is a wrong design and it should never be there for many reasons so that's that's the first sign i saw on this on this um schematic and i i think something wrong with this design basically so in the left bottom corner there is the proper circle this one is one this this uh this circuit is the one i found on the internet and i think the chip is similar to the chip in this case because you can see even on their data sheet they have the led lights on so uh i think it's a very low cost solution they use all the time and there is no capacitor between between vd and yeah okay i see yeah uh it's basically yeah i think this is what i draw based on on this pctv and i think that's that's that's how it is basically yeah i'll show you i'll show you a proper one then you will be amazed okay so this is a demo board from st electronics i found it online and you can see the quality components they use did you see did you see that they have a few capacitors yeah one two three four a few capacitors and then you can see this is the transformer and i if you look into details yeah zoom in did you see this this thing here it's uh it's what we call a belly band shield so it's a shield on the transformer basically yeah same as we try to apply so they have a shield on the on the transformer and they have this i think this must be an x-cap as well or maybe a move a common constantly from this angle and then they have a common mode choke as you can see um yeah all co all good quality stuff and then if you look at the schematics and then you will see wow look oh okay on the low voltage side they have lclc filter on the high voltage side they have um you know capacitor common mode choke yeah this c1 is the uh x cap i i was using very similar techniques and you can and then you can see their chip is much more complicated than the one we just saw so you know that you know for these low cost solutions they basically they took everything else out basically so that's the challenge they removed component and say still works yes still works oh always works yes yes i'm pretty sure that's how they how they uh how they did it um so uh basically as i said now this time i i studied this oh if you go back if you go back yeah yeah i see the they connected capacitor between the on the transformer but it is in different place yes this capacitor as you will see later on is very important it's very important yeah i will show you i will show you so uh that's a good good spot yeah um so so basically this is a summary we put on on this slide we're telling you know the layout is horrible it's horrible right i can't describe how bad the layout is it's just terrible and uh it's not safe right that's why i said you know be aware when you buy these cheap leds is even you know the high voltage design is not safe right pbr aware and uh and basically there's no stubborn circuits no feedback yeah the whole design is really just functional yes and as long as i see voltage then they're happy basically okay so uh we we said okay this c3 we discussed before is is strange so uh i basically moved that c3 and i think that c3 is already broken it's just connected there and then you can see without doing anything i just removed uh c3 you can see we have some improvement again okay again i think it's about 10 db yeah maybe 5 to 10 db you can see the comparison and uh at this point is a good sign and i think okay at least now we're seeing improvements again okay so um so basically yeah we have you know we we did lots of steps in before but we didn't see any improvement but now i found this c3 so i said okay you know what let's make this point as our new starting point and we try to bring down the noise from this point the yellow trace is the ambient and the purple trace is the new starting point without c3 so then we can have much better comparison because so did you change also some settings in spectrum spectrum analyzer or or in the connection i will remove yes i put all the change here so you can see now um the previous one we measured up to 300 megahertz but this time i measured up to one gigahertz again and i might have changed uh the uh pre amplifier or the um or the attenuator in the extra analyzer as well so i marked them so now you can have a better uh look so the idea is really um as you said you know initially if i don't apply any attenuator to my spectrum analyzer it will go above the the limited line and then speeding and could potentially uh damage my yeah so so now we're in a in a stage that i can you know get rid of some attenuator so now it's much better yeah it's a good importance yeah yeah okay okay so now we call this the new step two because as we said the step two three four should work but it didn't work okay so now let's let's call this the new step two as you exported already in the design people often put a wide class capacity i have a question i'm sorry for interrupting so you removed all the previous changes so in these changes i had no yes you're right i removed all the changes and we are starting again basically yes except the low voltage capacitance remember we we also put some low voltage ceramic capacitors on the low voltage side and so i didn't change these because you know i knew they they're probably not doing any bad or probably not doing anything good but i'll leave this okay and i see there is also 10 microfarad capacitor on the input uh yes yeah sorry yeah i should have marked here is this this just do the 10 10 micro farad capacity as you can see here so the only change is just the capacitor let's say we're changing to 10 microfarads okay and uh yeah so uh so what what we did first is i put a y2 capacitor which is one nanofarad between the primary side and the uh secondary side ground so that's exactly the capacitor what we could see in the big schematic yes yes they also had two yes two points i put one nanofarad i also have 2.2 and i believe if i put 2.2 it will even better be better okay but uh there's a point here i just want to mention the it's not like you can put very large capacitance you know we know one nanofarad is good 2.2 is better but i can't put 10 let's say and the reason is because if you put too much too much capacitance between the two sides the leakage current going through will be too big and it's uh it's lethal so uh standards basically define that you shouldn't put too too large capacitance between the two to prevent leakage currents that's the from safety point of view so that's very important why why did you choose this capacitor is the major or one of the changes what we would like to try uh because because when i had a look at the pcb you know the first thing we saw is that the layout is terrible which means they don't have a good solid ground plane and which means if i measure the common mode voltage between this point and this point is going to be huge common mode voltage and that's the main noise source of the whole thing the problem is the huge common mode voltage between this point and this point so uh naturally the first step you know i'm thinking is i put a y capacitor here and i can make that common mode voltage become less and that's that's why i chose this method in fact i measured the common mode voltage and you can see it's it's really bad yeah yeah i have the measurement video if you want to see it yeah and uh yeah it's uh it's pretty pretty uh obvious that the common voltage is very high and if we place the capacitor then we kind of make short circuit for these high frequencies and then the voltage will be same on on the cables yes yes you are already jumping ahead so basically you you understand everything so that's exactly what we're going to see so uh this is the result right um so this is the results um so the purple line is again you know we said this is the new starting point that's okay the point where we removed c3 and now we are putting a y capacitor one nano farad y capacitor between the primary side and the secondary side and you can see it's a clear 10 db reduction from very low frequency all the way up to high frequency so this change is is major change you know across a wide frequency range and it's 10 db at least yeah okay so why this helped yeah look i had these slides why this why yes um as you already um you know you already understand it yeah it's uh really just to bring down the voltage difference from rf radio frequency point of view yeah and i use a you know analogy here basically we were saying that um you know the previous tests previous steps we did why didn't help is simply because we we we didn't fix the big pipe you know imagine you've got a leakage problem in this pipe first thing you want to do is fix the big pipe because that's the main main source so what we did just now we put a wide capacitor here it's like when we fix a problem we fix the big pipe of a you know of the system so uh from this point if we bring back all the methods we use like shielding the transformer printing white capacitor we should see step by step change now okay i understand so because there was this huge noise we could not really see any other improvements yes yes and basically huge noise was caused by incorrect connection of the capacitor because the original capacitor probably should be placed on the ground instead of the power yes and and i don't know i think no i think they didn't even intend to put a y capacitor because the capacitor needs to be y class rated as you can see it's a it's a big disk ceramic capacitor like this you know y-class because then you fail you know you fail in a safe way but they use a ceramic uh surface mounted capacity so uh i i simply don't get it why they why they did not design so yeah okay so can we explain a little bit more about this so what happened i explained it but i would like to hear you to explain what happened when you place this capacitor there yeah so um to understand it i think you need to understand the uh common mode voltage and in this case the common mode voltage is from is this point which is your primary side zero volts point and then this point which is your low voltage zero volts and if you measure the voltage difference between this point and this point you will see a big voltage jump between these two points and particularly for radio frequency range that's causing the problem that's the rf voltage so if you put a capacitor between these two points and capacitor works with frequency so when your frequency increase the impedance drops so basically is making sure that this point and this point their potential from rf point of view is more or less the same so you don't have a voltage noise source between the two sides the two sides mean primary side and secondary sides so you don't drive any emissions up yeah so basically this 1.9 meter cable suddenly will have some kind of ground but before it was not really ground yes i think this is a good understanding of it before that cable is antenna and is an antenna and and the noise source is because there is really this big voltage difference between here but now you have a capacitance and then you basically reduce that noise source so what we are saying is you know actually this is one of the most important thing in the emc world is you if you solve the problem at the noise source it's much much easier you know it's much easier as you can see if we don't solve this problem at this point regardless we put the shield filters it doesn't help isn't it so solving the problem at the noise source is always always the best okay so we can we can go to the next slide yeah so next step is we are bringing the shield back we try to shoot before and we didn't see any difference so now let's try try the same method exactly the same method put the shield on bonding it and then see what happens and this is what happens so wow that's a big difference that's a big difference yeah again another 10 db you know from from this this blue trace is where we had the white cap fitted and now we put the shield in then you can see the green trace so now another 10 db i would say and it's quite effective again across a wide frequency range and as i said that's what we expect actually we we expect the shield will work but initially we didn't see it simply because yeah simply because the reason we just explained yeah okay and step four again uh we talked about this before so when we have y capacitor connected in this fashion we should guide the common mode current back to the noise source so uh in this case we put two y capacitor between the live and the ground and neutral and the ground basically yeah and and what we see is it didn't bring any uh improvement okay still hasn't got any improvement but it doesn't make um emission worse it doesn't make it worse so and i think the reason i put in the reason here so um as you can see um this is really what we wanted to have right this circuit is very similar to the circuit we showed you earlier on with that uh st design you want to put the common mode choke somewhere here and on this side yeah on the dc side perhaps and then the common mode choke with the capacitor will have a loc filter effect yeah um but because of the layout i cannot put a common mode on the pcb it's simply too big if you put a common mode choke here it will basically block them block or cancel so uh and you block the canceling is cancelling differential noise so okay the differential signal is cancelled but for common mode it's like a door basically i'm blocking you basically it's high impedance it's a high impedance common current yes yes so uh so as i said we we have you know the the layout is already there so the best way you can try is we have the y capacitor but we put a common module on the ac side it's not ideal but it's it's you know it's uh it's the one way of doing it so you can see i have the white cap here and i have a common node choke here and i'll show the result so this is the result again here uh the the blue trace is where we have the um basically the green trace is the new result so you can see now the the values is below 17 db micro volts in some cases so you can see if i this is the 17 db micro loss line and the green trays basically they stay quite close to the 17 db microvolts line so again i think another 10 db uh reduction because we started as 47 or 57 let's go here yeah so we started at 47 so now we're almost 17. so all together through the steps we bring down 30 db across a very wide frequency range yeah so um each step basically brings down 10 db and that's that's what we hoped as we discussed early on i said each step we bring down 10 db then hopefully three or four steps we bring down 30 db and yeah so so now we have a design that i think at this stage is good enough you know of course it's not perfect um but based on their layouts and their really bad design and i think at this stage applying the methods we applied including adding y cup between the primary side and the secondary side adding a shield on on the transformer adding some y capacitors on the ac line all together we are in a stage that we are happy basically and and the the final step is i can pack everything up in the in original uh little plastic uh and yeah i i i mentioned something here basically is is the the beauty of emc engineering in my opinion is you can fix some problems and then you can still use the same box let's say because everyone can buy a big heavy filter you know like this one and then put it and it works but that's not fun you know the fun part is to stop the problem as the noise source and pack it up still in a original form not stopping the noise but stopping the source of the noise yes yes and yeah then uh then we should do some uh moment of truth test the final test wow the last difference we again now we bring back the original setup where we use the antenna right because all the fixed we're using a temp cell and now we have a small antenna we pointed it exactly the same way and you can see that's before and that's after we are talking about sometimes 40 db reduction 40 db reduction yeah yeah and i think we need to say that the setup for the second measurement it is exactly same as for the first one that's why you had all the tape markers and and pictures and everything so you could make it exactly same yes so um so now let's go back to the same setup again but let's have a look at what's the performance now [Music] [Music] so now we don't see any spikes and the radio performance okay you know so i'm thinking okay that's a job down for the radiating emission well done i'm currently working on a course uh specifically for emc design engineers and the course will be basically theory based but then supported by simulation and real real life case studies and i believe this approach is the best for engineers to gain the maximum of emc knowledge so uh if you're interested in my course please drop me an email which you can find on my website it's min. at macquarie design and we'll let you know the first time when the course is out thank you and uh that's everything for today's video i would like to say thank you so much to min because you know it was a lot of work to prepare all the materials for this video so thank you so much and if you have any questions leave comments if you would like to let me know what you think about this video leave comments if you have any ideas for future videos you can guess leave comments uh if you like this video don't forget press like button if you would like to see my future videos don't forget to subscribe i would like to thank you very much for watching and see you next time bye

Channel: Robert Feranec

Views: 13,292

Keywords: EMC Problem, emc probelm, emc problems solutions, emc problem solving, emc problemas, emc probelm solving, emc radiated emission, Radiated emission test

Id: fkNa-FejWsQ

Length: 75min 13sec (4513 seconds)

Published: Thu Dec 09 2021