Vibration Analysis - Demystifying Modulation by Mobius Institute

hello and welcome to this presentation on demystify modulation my name is Jason Tranter and I am the founder and Managing Director of Mobius Institute in this presentation we're going to look at modulation amplitude and frequency modulation and we're going to look at some common fault conditions that cause modulation in gears bearings electric motors and might mention some others I'm going to attempt to clarify what the differences between amplitude modulation and beating and clarify what the difference is in between modulation frequency and amplitude modulation that we see in machines and the demodulation technique used in enveloping and so on they might seem like they're related but they're actually not the reason for this presentation is I did one recently in a webinar that was related to gearbox vibration and quite a few questions came up afterwards related to these topics I thought well there's a good opportunity to try and clarify some of those points anyway let's start with the real basics here we have a simple sine wave which when we do an FFT produces a single spectrum a single peak in the spectrum I should say you notice that each one of these is a cycle we've got 20 cycles per second but the amplitude of each cycle is the same it rises and falls by the same amount each time and that's generally what we expect to see if the machine was out of balance then I get some rise and fall in vibration we get one peak in the spectrum and all is good but what happens with amplitude and frequency modulation well let's look at this case again and what I'm going to do is just bump it up to 50 Hertz and I'll just bump this amplitude down just a little bit just makes it easier to see what I'm about to talk about so again we have the situation one peak in the spectrum one source of vibration if that's the signal and now if I just add another signal to this and it has a whatever the frequency of that signal is and we'll give it some amplitude as well what we've got right now are two signals we had our fifty Hertz signal plus this four Hertz signal there's therefore two peaks in the spectrum and how our time waveform takes on a certain shape you notice the both halves arise for rise fall and we could play around with these signals and show what happens when you simply add two signals together but the waveform will change a certain way the spectrum will change a certain way but this is not modulation this is simple addition of signals what I'm going to do now is change what's happening with this little simulator so that it is performing amplitude modulation what's happening now is I've got my 50 Hertz signal which is this one in here but I am modulating that signal by this one which is four Hertz and an amplitude of two this one was 50 Hertz an amplitude of five during about the units on the amplitude it's just arbitrary so what we see now is that the signal itself Rises so if you look at this portion here it looks like a higher amplitude 50 Hertz signal than here and that's actually what's happening it rises in amplitude falls in amplitude rises in amplitude falls in amplitude this periodic change in the signal characteristic which in this case is the amplitude that's changing it's called modulation and in this case because it's the amplitude changing it's called amplitude modulation with radios is FM radios and AM radio this is the technique with a am this was our high frequency carrier signal the one in here which is the frequency you dial in to on the radio dial and superimposed on top in this case is just one frequency and one amplitude but natural fact in the old days when am what am still used now but that would be the voice signal the voice and music and whatever is coming over the radio was superimposed over the carrier that was used to transmit it from the radio station out to the towers and whatever anyway without going to all the detail the amplitude signal the amplitude of the signal varies in this case it is varying four times a second four Hertz so if if I could if I had a sine wave which looked exactly like that it would be four Hertz signal with an amplitude of two in fact I can say show me the two signals and you can see and just hopefully seen there the orange is the sort of carrier signal it's the signal that I'm modulating and bit hard to see but in here is the green this is the modulating signal there's just the combination of the two now while I am doing all this if I increase the frequency we see the the rate at which it rises and falls is now five times a second so they're sort of bunching together we can change the amplitude as well down below is the spectrum and notice I do not have a peak in the spectrum at five Hertz with an amplitude of two it's not to say when you do vibration analysis there wouldn't that there wouldn't be such a peak but with pure amplitude modulation I'm not seeing that peak at all this isn't something funny I'm doing in the software this is true you know I'm modulating the signals I'm doing the FFT what we have here is the center peak which is related to this one so it's at 50 Hertz and has an amplitude of five if I increase the amplitude you can see how the waveform changes and you can see how the spectrum changes but I've got these little Peaks on either side they are called side bands the separation between this Center peak which is also called the center frequency and the sidebands is five Hertz so we've got 50 there for this peak is at 55 this one's at 45 with pure amplitude modulation I just get one peak on either side but notice that if I increase the amplitude of the modulating signal the side bands go up and I notice the time waveforms start to look a little bit odd but same way if I increase this the carrier signal then sure enough the center peak goes up and if we play it around I could make the center P completely go away and just have sideband Peaks in reality we can have situation a little bit hard to explain quickly but you get sort of this addition and cancellation of signals such that the sideband amplitudes aren't the same we can get a little sideband peak on this side and a much bigger sideband peak on this side in fact you might see situations with the electric motors where you have a peak at the rotor bar frequency and on either side of that is a peak at twice line frequency so it'll be separated by a hundred Hertz or 120 Hertz but you may have a situation where the peak at the rotor bar pass frequency is not all that high in amplitude the peak at that frequency - twice line frequency is may be quite small but the amplitude of the peak at wrote about passing frequency plus twice line frequency is actually quite high in amplitude and in fact it might dominate that part of the spectrum so when you look at it you might see a peak and think wow it's not an integer multiple of running speed I don't know what it is but you might find that if you subtracted twice line frequency that there might be a small peak in the spectrum and that would be an integer multiple of running speed it would be the number of rotor bars anyway so we can see various combinations of these sorts of things which we'll explore a little bit in just a second so that is amplitude modulation and shortly I'll just summarize a few things for the sidebands now we're going to take a look at frequency modulation in this case we see we're back to our 20 Hertz amplitude 8 and we've already seen that if I add a signal of just you know a low frequency you know whatever amplitude you know we've already been through this process it changes the waveforms in a certain way but watch what happens when I say let's modulate the frequency of our original signal instead of the amplitude rising and falling the frequency of the signal changes notice if you just look here it looks like a low frequency signal here it looks like a higher frequency signal so the frequency itself Rises Falls Rises goes up goes down goes up goes down increases decreases I think that's the word I'm looking for now I'm exaggerating the amount of modulation in in actual case with rotating machinery this is more common than you may think but what's actually happening is there's a slight variation in the speed of the machine so in a moment you'll see that you know as gears mesh together we can see if there's any eccentricity or whatever we can see some amplitude modulation but as those teeth mesh together we can have a slight sort of slowdown and then a return to the proper speed a slight slowdown a return to the proper speed so we get this variation in speed which is therefore variation in the frequency it's a periodic motion as each tooth comes into mesh and we can see therefore variation we can see it occurring more so at some points in the rotation than at other points and therefore once per revolution the machine can appear to change speed either way even though this looks quite strange down here we see lots of sidebands so if I just change these these parameters here if I just make some changes anyway you can sort of play around with this there's just lots of sidebands in here and just depending on the actual settings and so on we get a real variety of these side bands which is separated by this frequency here it's separated by that and you can see how that's changing but we get these funny cancellations and so on without dragging on about this too much basically you see that it's generating lots of side bands and that's why with machinery there's a few other reasons I won't go into but we see more side bands than just that one on either side that I demonstrated with pure amplitude modulation so when we do our analysis with the spectra we can see all sorts of things we can see you know side bands appearing around one X and maybe two X and maybe 180 n2x and maybe a gear mesh frequency maybe a bearing frequency now the separation you know there's actual sideband frequency as we'll see in a moment could be 1x vibration it could be equal to the fundamental train frequency it could be equal to the pole pass frequency various types of separation so you you know different carriers depending on what sort of machine and component we're looking at the separation put into the fault we can have harmonics and those harmonics can have side bands we can have just one side band on either side we get lots of side bands on either side and there's variations in amplitude so basically we have the center frequency the side bands and there's a certain separation which is you know equal on either side and as I say we can see some sort of harmonics and each of those harmonics might even have side bands you'll see a real-life case of that in just a sec here we see more sideband so there's our sideband our center frequency and side bands again there's this sort of separation between two which is equal there's a harmonics and we can see them for a variety of reasons you know just different types of fault conditions so I'll look at in just a second so let's have a look at some real cases okay this isn't real but this is just talking about gears under normal circumstances with these two gears turning I've got certain vibration because of the two shafts which I'm not showing here what I'm showing here is just the gear mesh frequency itself so it's two fish two fish tooth mesh tooth mesh and now if you look closely you can see a little bit of variation in the amplitude but let's just pretend that's not there and just say if the two gears were absolutely perfect not moving in with any eccentricity at all and each tooth mesh was you know the same as the one before it and after it then we would have this nice you know smooth vibration but let's take an exaggerated case where if you look at this gear it's moving correctly but if you look closely you can see that while the shaft is turning correctly on its axis the gear for whatever reason is moving essentially eccentrically we may have another situation where the gear isn't moving relative to the shaft but the shaft itself is well either eccentric or bent or something like that so either way this gear is actually moving in a circular orbit so at one point in that orbital sort of motion we are therefore putting more force in the gear mesh at the opposite part of the cycle there is less force in the gear mesh now my animation is really exaggerating that point but as a result we've got our gear mesh frequency but rises and falls thanks to the eccentric motion rises and falls if it's this gear that's eccentric then the rise and fall the that that sideband frequency and that time between the cycle that you can see here is related to the turning speed of this opinion if on the other hand it was this gear that was eccentric and this one was not then I would see the same sort of shape except given the dis is tuning at a lower speed the rise and fall would be at a lower frequency so anyway gear vibration is a classic source of amplitude and frequency modulation bearings are also another common source so let's just take a look at something here here I have my my simplified 1x vibration and superimposed on top of that I have the vibration due to a fault on the outer race so this point is damaged and each time one of these rolling elements impacts that point we get a little spike now what I'm going to do is I'm going to remove the sine wave the 1x vibration and just look at the spikes and I'll just give it a bit more amplitude so what we're seeing here is it as each rolling element impacts this point here we get a spike of vibration you know this is the ball pass a de Rais frequency that's ping ping ping this little ring down here is the resonance or it's a stress wave let's not get into all that for right now but the fact is that because it's on the outer race and the outer race is not turning then the force in each impact is the same on the other hand if we look at a fault on the inner race as the damaged area moves through this load zone the force of the impact is much greater than when it occurs up here so if we stop it for a second we get it you know the impact is occurring in the load zone out of the load zone in the load zone the load zone and therefore you know that's when it was in the load zone out of the load zone in the load zone out of the load zone this is modulation now it doesn't look all pretty and clean like it has before but instead of a sine wave in here we have this impact which is occurring according to the ball pass frequency in a race be PFI because we got the fault on the inner race so this is our B PFI we've got a ping ping ping we've got a little bit of ringing because of resonance or the stress wave or as I said let's not dwell on that but the amplitude of that signal is rising and falling therefore in my spectrum in addition to having a peak at this ball pass I'll be PFI frequency and in addition to having harmonics of that because it's impacting and that yeah it's not a very smooth source of vibration we get harmonics we also expect to see side bands of the turning speed of the shaft because once per revolution of the shaft we see a rise and fall because it's once per revolution that this damaged area is going in and out of the load zone so we see side bands for that reason on the other hand if the damage was on the rolling element itself well this is our ball spin frequency there's just the resonance and shock pulse or whatever but we got modulation here now - it's at a much lower frequency in fact why don't I just change that setting so that we can see it a little bit more easily so now you can see it it's it's when the cage takes that rolling element out of the load zone into the load zone out of the load zone into the load zone so I change the time scale so that we could see it out of the load zone into the load zone out of the load zone into the low tone so there's our ball spin frequency in actual fact the lamb doing this is twice ball spin frequency because we are impacting on the on the inner ace then the outer ace and the inner a sling the adder ace but but again we've got good old amplitude modulation occurring so we expect to see peaks in the spectrum but this time the side bands will be closer together because it's a lower frequency okay so that's the case with bearings in the case of electric motors we can have modulation for a few reasons one of them is because of broken rotor bars or damaged rotor bars or end rings and so on and one is because of an eccentric motion of the rotor now it's a little bit harder to explain this I'll just mention a few things and hopefully it's sort of food for thought when you pass a current an AC current from a 3 phase source through the stator windings we create a rotating magnetic field and for 2 pole motor that rotating magnetic field is over 50 Hertz if that's country you're from or 60 Hertz that's what the blue wave is here it's like the North Pole the South Pole 2 poles north-south that rotating magnetic field induces our current in our rotor and when you induce current in the rotor it also generates a magnetic field that magnetic field sees the spinning rotating magnet the rotating magnetic field around it and tries to chase it it never quite catches up but if you watch closely you'll see that the rotor is turning more slowly than the magnetic field you see these lines on the rotor are sort of slipping behind that's the slip frequency now imagine if there was a broken rotor bar right where my mouse is on the rotor I'm trying to keep it in the one place what we find is that the ripple of vibration due to that when it's right in the center of this strong magnetic field the forces are greater than when it happens to be at this point so as because of the slip frequency as the broken rotor bar sort of is in the magnetic field versus outside the magnetic field well in the strongest partner in the weakest part we have this rise and fall in vibration that's the pole pass frequency it's the slip times the number of poles on the other hand if we had a centricity you see that the rotor is now moving in a sort of a circular motion so there's one point which is closer to the state at any point and that also causes modulation you can sort of see it here there's one point the air gap is not uniform around it but it's it's rotating a centricity and we get modulation it's a little bit harder to visualize it in this case and that's not exactly the point of this little presentation here point is we do get modulation in particular amplitude modulation we can see it in the waveform and we can see it in the spectrum so let's have a look at a couple of real samples of vibration here is vibration from a bearing and what I've tried to illustrate here is you know there it's in G's so there's the spike as it's going through the load zone out of the load zone into the load zone out of the zone into the load zone out of the load zone and so and so forth so we get this modulating effect as it goes in and out of the load zone of the bearing and as a result we get other modulation we see the side bands in the spectrum in this case and so these arrows are going to get funny little technical difficulties that to uptime but anyway in this case we've got damage on this gearbox some s intricity on both of them a little bent shaft in one case is interested in the other case just to simulate this fault and so there's our modulation there's the turning of the lower speed gear the driven gear so there's the modulation modulation modulation modulation as as its teeth sort of go harder in mesh we can mesh harder in mesh we can mesh and then there's the pinion so the pinions the smaller gear in this case it's the output as well so we can see the fluctuations in the time waveform there as well but if we look closely at the spectrum I can clearly see the sidebands as a result and so that the spectrum did give me an indication of this particular fault but I can set the time waveform to sometimes you might be surprised that if you look in the log version of your spectrum you'll see side bands so in this case yes it's high amplitude vibration but you can't really see any side bands but when we switch to logarithmic form at lo and behold it's got these little side bands around each of the peaks you can see them quite clearly here the side bands you couldn't see them in the linear in this case you can see some side bands in the linear but in the log you know log vertical or amplitude scale you can see them quite clearly all these side bands here anyway one of the other points I wanted to make in this quick presentation was how beating is often confused with amplitude modulation in the time waveform they look similar so let's go back to this situation what I'm going to do is make this 50 Hertz signal and I'm just going to drop this amplitude down for reasons you'll see in a minute so here I've got a signal of 50 Hertz and an amplitude of 5 don't worry about the units and it looks exactly as we expect I will add a an additional source of vibration you know give it some frequency give it some amplitude now right now you it's what we saw before let's do something interesting our first signal is 50 Hertz with an amplitude of 5 so what I'm going to do is I'm going to change this one to 50 Hertz with an amplitude of 5 and what you see is what you might expect I've got the two signals adding together in fact I can show all signals but it's a little bit hard because there's sort of being overwritten by the the final result but what you can see here is two signals adding together so I've got amplitude of five for the second one amplitude of five for the first one till we get an amplitude of 10 but that's because the two signals are in phase with each other they both had zero phase what would happen oops-a-daisy if I change the phase of that second signal now if I slowly adjust it what you'll see is that the amplitude appears to be dropping so we can keep changing this and when we get to 180 degrees lo and behold we get zero amplitude nothing in a spectrum nothing in the time waveform because what we have are two signals which are 180 degrees out of phase and it'll be a bit easier if I actually temporarily dropped the two signal frequencies down and you can see it very clearly here we've got the green signal rising falling rising falling the red signal rising falling but out of phase with each other and they cancel each other out when there is zero phase difference between the two of them we just wind that back you if you look really closely you see these two signals moving sort of relative to each other timing wise but when they are in phase with each other you see that they add together when they're out of phase they cancel each other now why did I just tell you that because if we just wind that back again and we change the frequency of one of them just by a bit we start to see this curious thing happen this blue signal is the combination of my 19 Hertz signal and my 20 Hertz signal you can see that here if you look closely at the two raw signals they start off at the same time and right at the same time they look very close in phase to each other and therefore the result is that they add together but because there are different signals you know if you keep watching these two you'll see at a certain point they are basically completely out of phase and they cancel so let's go back now to 50 Hertz and keep doing that and we'll just make its about 49 Hertz and we'll leave the amplitude the same but now in fact what I'll do is to make it more of a difference we see this blue resultant wave form now and turn off those other ones because we understand what beating is now we see something that looks just like amplitude modulation but it's not it's actually two signals going in and out of phase and look at the spectrum even though those waveforms look really similar if we studied them more closely you'd see that there is a difference but notice I've got two peaks in my spectrum one at 47 Hertz and one at 50 Hertz and depending on the relative amplitude you know you can see that the rise and fall is different you know if they were the same amplitude as each other I forget what I had for this one five of course I did you know at this point they add together to be ten at this point they cancel each other out to be zero right at this point that just depends on the relative amplitude but this is the situation as I say where we've got two two sources of vibration interacting with each other and we get this this beating phenomenon but as a result I've actually got two peaks in the spectrum and if I have resolution that was high enough I would see that there were two peaks those two peaks could be the 2x vibration and twice line frequency it could be other things that the beat together so it's not the same as amplitude modulation even though the waveform looks very similar now last but not least I want to raise this issue or try to clarify the issue of you know we talked about modulation and we talked about demodulation so surely those two things are related aren't they well no actually they're not everything we've discussed with modulation is in regard to sources of vibration in a machine that happened to rise and fall in amplitude or vary in frequency as we've discussed so got bearings with inner race going through the load zone rising and falling in that in the source of vibration we get sidebands now yes it's true we can then use demodulation to diagnose that fault but for an entirely different reason because let's face it we can have an ADD erase fault there is no amplitude modulation with an outer race fault but we still use demodulation to detect it so let's see why it is and I could use the bearing simulator or this one if I look at this vibration again and let's make it that fault in this one the situation is that in the early early stages of a fault we have the situation where we've got a little bit of vibration well we've got the One X vibration and two X vibration and superimposed on that is just a very little bit of vibration now what I'm displaying here yeah the spikes a way to beat them is INRI is the case in reality so we have an early stage fault there's just a little bit of impact vibration in that little red bottle where the damaged area comes into contact with the rolling element so here and here now in my illustration here my little simulation I'm exaggerating that amount of vibration so the fact is that if we simply took the raw vibration from the accelerometer and looked at it as a time waveform we may not see it in fact in the very early stages of a bearing fault we won't see it if we looked at it in a velocity spectrum we won't see it there are some other issues at play here related to duty factors and limitations of the FFT and so on but we can use some smart technology step one of the smart technology is to filter out all that low frequency high amplitude vibration now it's potentially a little misleading what I just did then because there are two things kind of going on there is just a periodic source of vibration but every time there is an impact here so every time there's there's damage on the inner race or on the outer race that's not confused these sources of modulation every time I get this little ping there is a sound ping I'm waiting for it ping waiting for it ping and that's something that if the amplitude was higher we might pick up in a normal spectrum but every time we get a ping something else happens inside the machine just any time you've got metal to metal contact you get this stress wave goes whooshing through imagine a great big metal bell and we got this little metal clacker whenever those things are called and we just give the Bell a little light tap we may not hear the bell ring we may not be able to hear the tap at all but natural fact because of the metal metal contact we can detect this stress wave or the shock pulse that ripples through if we hear it just a little bit harder again with all the other sound that's going on imagine that this Bell is surrounded by all this traffic and whatever other than you know just a lot of confusing noise at to lower frequencies we couldn't possibly hear a person just dinging the just tapping that Bell however if that bell rings at a high enough frequency we can filter out all the vibration at a lower frequency than the ringing of the Bell so all that high amplitude noise and vibration that's distracting us from our analysis of the bearing or the Bell and all we can hear then not so much the tapping of the bill but the ringing of the bell or the stress wave that I mentioned so now we see stress wave stress wave or the bell ringing ringing and so what we've been left with now is just this wave form and what I'm going to do is just increase the amplitude of it so it's a bit easier to see I've just got ring ring ring now this is all at a you know a higher frequency source of vibration what we want to do is we want to make it easier to capture this because the stress wave of shock wave or the ring could be very short in duration harder to detect so we go through an enveloping process or a demodulation process it's the actually the same sort of process that they use in a.m. radios to extract the the sound of the DJ on the radio from the carrier frequency that was used to get it to your radio so the first step is to rectify the signal push all the negative stuff positive and then put an envelope around it and then put it through a low-pass filter which I can illustrate back here as well so you know we've got this sort of a signal we filter out the low frequencies we make all the negative going vibration positive going we envelope it and then when we put it through a low-pass filter we end up with something that if we look at it as a time waveform we can see the periodicity s so if I if I'll just go back to that same point what it's going to show is you know there's my damage on the outer race of the bearing that's the time between each ball impacting that point on the outer race so as I go through this process it ends up giving us a nice time waveform we can look at or or nice peaks in the spectrum so it can be confusing we can say well isn't that same isn't the signal the same as just the vibration no it's not there's a number of reasons but really what we're doing is we're taking advantage of the fact that you know that the bearing resonates or the sensor resonates or we can pick up the stress waves we can use some signal processing to detect those very short duration Peaks or impacts whatever you want to call them and and then going through this so-called demodulation process we make them visible in in a spectrum or a time waveform so it's a very effective technique but as you can see we use demodulation in the case of you know add erase faults on bearings where there was no modulation in the first place it's a coincidence that we've got modulation in the case of inner ace fault and we use demodulation to analyze it they're just two different processes all together but what I would tell you is that we can use that process even for gear faults and the broken rotor bars as well again nothing to do with the fact that they actually caused modulation it's because as those gears impact each other if they were a bit worn or there was a crack or whatever they also create this high frequency vibration so again I can use that enveloping technique to get rid of the low frequency high amplitude vibration and just focus on what results when you get metal to metal contact same with broken rotor bars you can see a situation with a broken rotor bar because of the crack that it creates the same stress wave and you may pick that up and there's other situations as well but it doesn't rely on a modulated signal in the first place it's just that we're exciting we're generating stress waves or shock pulses or were exciting resonances in the selamat or the machine itself okay so amplitude and frequency modulation are common in rotating machinery we can see the results of particularly amplitude modulation in the waveform and in the spectrum we can see amplitude and frequency modulation quite clearly in the case of the spectrum we see it in terms of the side bands and we can look at how many side bands and the relative amplitude of the side bands and how they change over time with bearings rolling element bearings gears electric motors and other components we can see amplitude modulation occur because there's a rise and fall in the amplitude of a certain source of vibration there is a difference between the amplitude modulation that I've just talked about and beating and there's difference between that same modulation that we experience and the demodulation measurement technique so I hope I have clarified the point not created any more confusion you know these presentations I'd love to spend a lot more time on on these things and explain everything in more detail look all I can say is that's what our training courses are there to do you know it just takes time to really clarify each point but hopefully I've made some things clearer for you today so thank you very much for viewing this presentation just in case you were looking at this through an email link or something like that or via youtube or something we've got lots of lots of other presentations both on our website you can access the presentations just at the moment you you can join as a free member to my mobius but in a few months that will go away actually will be easier to access then but either way there's lots of presentations on lots of topics both on vibration but alignment balancing and the much bigger topic of reliability improvement as well anyway thanks for viewing the presentation I hope you found it interesting

Channel: Mobius Institute

Views: 36,304

Rating: 4.9533076 out of 5

Keywords: mobius institute, vibration analysis, iLearn Reliability, iLearnReliability, iLearnVibration, ISO 18536, training, iLearn Vibration, certification, accredited, CBM Connect, IMVAC, CBM Conference, The CBM Conference

Id: yp_gnvvct6E

Length: 41min 15sec (2475 seconds)

Published: Mon Jan 20 2014