 This lecture is on data recording and signal transmission. As you know in machinery condition monitoring, it is very important that we get the data out of this machine and then record it, so that we can do these signal analysis on it. Well, so what is the format of recording is what we are going to discuss in this class, the methods to record certain signal and then how do you do the analysis. You all recall that in computer rated data acquisition, we notice the importance of sampling frequency and then you know the number of data points. This plays an important role in the selection of the data recording media or format. Well, before we come to this idea, I would like to tell you something about the traditional techniques of data recording. Well, you almost have visited certain plants could be a power plant, could be a steel plant, so they have a control room in which I am talking about things in the industry about you know 3, 4 decades ago and then all they had were lot of analog meters essentially RMS or peak meters and data logging was done by plant personnel. This was the format of data recording may be couple of decades ago, but now what they were doing there is as a function of time, they would note down the parameters like the current drawn by a motor, may be the temperature of the bearing, may be the supply voltage and this time mind you or may be in every 30 minutes in hours. This was being done by a human being and if there was an and this is this should be the log sheets, they would essentially form the log sheets and then and this were available in the plant room and then if there was an abnormality at certain conditions, people would go back to a certain date and look into the log sheet of the plant and then try to come to a conclusion like what went wrong and so on or suddenly if in the plant operator notices that the temperature is suddenly rising is increasing, then they would certain alarm condition and then people would take precautionary measurements as to how to reduce this kind of abnormal behavior of the machine or how do you control them. But there were lot of problems with this, this was a completely manual mode, there are problems in which human errors could creep in because of wrong recording, wrong writing, wrong reading, this was one possible. Another was it was tiring and cumbersome because you know visually you have to read all the meters in the control room and in the control rooms could be panels of meters and so on and this was a very cumbersome and tedious process and then there are lot of chances by which errors could creep in and there are been instances where people have reported that because of a wrong reading recorded by person the interpretation was wrong and then the analysis of the prognosis was not correct. Well, another method was there were other than writing, so the first one was human intervention with physical data logging. But closely with this is using a either a printer or a strip chart recorder these are the analog and wherein there will be just a voltage signal coming out because of the parameter and then with time basically there will be a paper roll which will be moving at a certain speed v meters per second and then there will be pens on this and they will be writing or plotting the parameters and this would be on the roll and this is essentially what is a strip chart recorder and then people used to have such large rolls of data wherein we can look into the problem and then find out get basically a time history of the signal by looking at the strip chart recorders and some of these data logging meters used to have you know trip set tripping levels or alarm levels set preset that means when the alarm level increased or the signal level increased beyond the set level it would trigger an alarm. And this was what was the traditional method of I would say data recording, but as you can see the problem with this kind of data recording is if I was to less down one was lot of scope of human error it is manual thus TDS time consuming it is not real time it is difficult to difficult to obtain past data because then one has to physically pull out the log sheets and that would be cumbersome and then because of large amount of writings archiving or storage issues present. Now, because of this disadvantages since the sixties people have started to use other means of recording and that was the machine mode and first would be a tape recorder like the spool or the cassette of course cassette came much later and the spool type tape recorder. Now, we will look into this different types of tape recorders and then what are the present format of tape recorder and how we do the digital recording in such tape recorders. So, if I was to look into the different formats of data recording available today one could be the analog recording or the digital recording. What happens in the analog recording we used to have the direct recording method wherein on a cassette or a tape recorder or a spool type tape recorder the data was stored because of the hysteresis in the ferritic because the signal would be converted basically essentially we have a voltage signal which is a time varying or an AC voltage this could be my voltage signal which has to be recorded. Now, because of this varying voltage there will be an magnetic flux created and the signal is recorded through the head of the tape recorder. So, we have a magnetic head and then the tape is made to go at a particular speed and the take up spool and the Holmes spool and they go at a particular speed and some of them could be 6.77 inch per second and this is a value or this is a industry format then it was in the inch per second speed linear speed and this dependent on the rpm you know we used to have the old LPZ in a 45 rpm or 233 and one third rpm etcetera. So, these are the particular speeds at which these things used to run of course, in the I am not talking about the LP spools because in LP spools what happens the data once stored cannot be erased or changed. But, we are not talking about the LP spools in this case we are talking about in head wherein because of the magnetism in the introduced in the ferritic tape the data could be stored and then of course, we also have the possibility of having an erase head if the erase head is there you could demagnetize it and then the tape the information stored in the tape could be removed. But, the advantages or disadvantages of this kind of systems are this had a very poor dynamic range by dynamic range I define as the highest recorded signal level to the lowest recorded signal level and they were also having a frequency limit like an upper limit or lower limit and traditionally these magnetic tapes had a very low upper limit could be as low as 10 hertz and may be as high as may be 12 kilo hertz so on. And, this later on improved with the type of tapes you know there were metal tapes, chromium tapes etcetera wherein this could be go up to may be 70 kilo hertz. But, traditionally this direct recording tape recorders were used mostly for recording audio signals and as you know the audio signals are in the range of 20 hertz to 20 kilo hertz and these recorders did a good job in recording such audio signals or for that matter any signals which had which were in this frequency range. So, when you are talking about machinery condition monitoring we could be recording signals from noise or acoustics signals vibration signals and may be temperature signals or certain process parameters like pressure etcetera. And, our requirement is that with each device we have to either reproduce the original signal. But, the problem was this direct recording was data storage because once we store it in a magnetic media there is a every possibility possible chance that in a magnetic environment they may deteriorate the physical tape may get corrupted because of moisture high temperature etcetera. You must have seen if in at home if you store your old and if you ever have your old school tapes or cassette tapes stored elsewhere you will see if you do not care about the moisture around you know there will be lot of fungi formation in those tapes and then there will be you will lose the signal quality etcetera. And, these are the problems and of course, from a signal processing point of view the major problem is it cannot record the low frequency signal. But, you know in these early 70s or late 60s this pool type direct recording tape recorders were quite popular because that was the only technology available then and they were doing fairly a good job with recording audio signals particularly in a music recording etcetera. Of course, if you want to then have a permanent recording in the media people use this the record record player in the record the LP records the EP records etcetera wherein permanently the song tracks could be cast and kept in those vinyl discs. But, we are not talking about you know we obviously cannot store data because data has to be recovered for an analysis and we do not want to record it permanently in a recording media. So, that the provision of erasing was not possible in the LP records of the EP records. But, in the direct recording by having the series head we could have always reduced erase the data recorded. Then, they coming back to the next format of recording that is the FM recording that is the frequency modulated recording. The FM recordings are very popular and convenient because they can measure as low as DC and then can go as high as may be 12 kilo hertz etcetera. So, low frequency low frequency recording are possible by the FM recording of course, the recording media could again be a spool or a cassette tape and so on. But, the advantages the low frequency recording could be done and it has a relatively high dynamic range. So, this was popular from the may be from the 70s to the may be may be mid 80s people used to use this FM recording for recording the signals this is from mid or mid 70s to 80s people were doing the FM recording. But, then this later on from mid 80s we have come up with this format of pulse code modulation recording which were essentially an high frequency recording used in the in the video recording. People use the same pulse code modulation technique which was used in video recording for recording audio signals and people then use them in data recorders or digital audio tape recorders and these are quite popular till very recently may be may be till about late or may be outside till the year 2000 they were very popular digital data recorders and the advantage of this data recorders are they are DC to about 20 kilohertz range high dynamic range compact and so on. If you see here I have just made a comparison in this slide on the data recording methods by the direct recording the frequency modulator recording the digital recording the lower frequency limit in direct recording about 20 hertz upper limit is 20 kilohertz noise floors 10 millivolts that means you cannot record any signal as than 10 millivolts dynamic range is very poor 40 decibels and time compensation is not available. Whereas, in frequency modulation recording I can go as low as DC or 0.1 hertz to about 12 kilohertz of course this has improved lately again noise floor is 10 millivolts but, then we can have a very very high dynamic range of 80 decibels and time compensation is possible. The digital audio recording or the data format which we use can have a low frequency limit of DC and then the high limit high frequency limit is 20 kilohertz noise floor is 0.1 millivolts and 100 decibels. But, the advantages of the digital recording over the other recording is that the noise floor being very very low we can have signals from devices like thermocouples as you know thermocouples give a very very low millivolts signal as opposed to may be an axial armature may be 10 millivolts and if I was to for a good compare this are basically low frequency signals and these are relatively high frequency signals. Now, having a signal less than or around 0.1 millivolts it would be very difficult to measure either by FM or direct recording but, you have a digital audio tape you could record such low signals in low frequency as well as high values and at high frequencies because they have a high dynamic range and then low noise floor. But, all these three types of tape recorders are available in the market I mean of course, you know it will be very difficult for you to find a digital sorry a direct recorder or an FM tape recorder with spool of the cassette nowadays. But, even finding a direct recorder could be difficult at times because you know nowadays people are using the DVD recorders or totally recording in a flash car or a compact media or even writing it straight to the hard disk or the device. But, the advantages of these recorders are they are portable low power consumption some of these recorders can be even powered from the 12 volt DC adapter given in your car. So, the mobility of such recorders is excellent because why do we have this data recorders in the first place? We have this data recorders because we cannot possibly go to the plant may be the plant is a coal mine a very dusty environment. We cannot go and sit next to a crusher or a grinding mill and do the vibration analysis has to find out the remaining useful life of this crusher or the grinding mill rather have the transducers mounted around the grinding mill or the crusher and then record the signal on a tape recorder and then this tape recorder obviously has to be portable has to be lightweight could be powered by any source and then once we have recorded it in a particular media we can bring this tape recorder back to our laboratory wherein we can do the signal analysis and try to find out the condition of this machine. It is not that the tape recorder is fixed or tape recorder is always there in the lab no this is not the objective of our recording here or recording in condition based maintenance or monitoring is to be have a big steel plant for that matter next to you sitting in your analysis office. Obviously, I cannot bring a steel plant to my office, but if I have recorded the signals in a tape recorder which reproduces the signals as they were recorded or as they were measured right at the steel plant my job is done and with that objective actually we are recording this signals and then we will see what are the advantages of using a portable recorder with high dynamic range where the data can be stored for long long of time. So, right in the beginning of this class I told you about the importance of two factors one is the sampling frequency and other is the number of data points. If you will recall because I need to do signal analysis that means if I have a signal like this some signal this has to be sampled at a fixed interval this has to be fixed at the sampling interval and the total time of this data total time is t is equal to n times delta t this t is equal to n times delta t where n is the number of data points. In the previous diagram all those crosses I had done they were actually the locations of the sampled digital data. So, in terms of the storage memory requirement because this digital data has to be stored in some memory location in the digital recording. Now, if every data requires 16 bits bits or 2 bytes of memory space suppose I am sampling at f s is equal to may be 100 kilohertz that means delta t is equal to 1 by 100 kilohertz. So, 10 to the power minus 5 seconds is my sampling interval and if I have to find out the total time is equal to n times delta t n is equal to the total time by delta t. Now, if I want to record 10 minutes of data at a sampling frequency of 100 kilohertz my number of data which has to be stored is 10 minutes 10 into 60 seconds into 1 by delta t which is 1 by 10 to the power minus 5 that is 600 into 10 to the power 5 data points. Now, if every data point requires 2 bytes I require the memory required to store 10 minutes of data sampled at 100 kilohertz in a 16 bit recorder would be 2 into 600 into 10 to the power 5 bytes. I would know 1 kilo byte is equal to 1024 bytes if I approximate it as you know may be for the sake of discussions may be 10 to the power 3 bytes the memory required 2 into 600 into 10 to the power 5 that is equal to about 1220 into 10 to the power 6 approximately 120 megabytes m b. Though I have used an approximate that 1 kilo byte is 1000 byte that is not quite right, but just for the sake of this discussion actually 1024 bytes. So, approximately they to summarize to store 10 minutes of data sample at 100 kilohertz in a 16 bit digital recorder would require approximately 100 120 megabyte of digital memory. Now, you can now multiply now if you want to instead of 10 minutes you want to make it 100 minutes another extra 0 coming here. If you want to sample at a still higher frequencies you know here the because n is equal to t by t by delta t or t times f s. If you sampling frequency is increased n would increase if your time is increased n would increased. So, depending on that you can do a calculation and then try to understand what is the amount of memory required, but in digital recording the most limiting factor is the sampling frequency. In digital recorders depending on our frequency of interest or depending on the frequency of interest we can have this sampling frequency decided from the Shannon sampling theorem. Now, suppose we are talking about a mechanical signals basically in the audio range from 0 to 20 kilohertz may be we sample at sampling frequency f s is equal to 44.1 kilohertz and this is a set sampling frequency and then we can record as much as you want provided we have the memory space, but nowadays there are very very high sampling devices which can sample at about 100 mega samples per second and this is the present state of the art is we can have a to d systems a to d card which can sample at about 100 mega samples per second and then this high samplings are actually used for studying ultrasonics because ultrasonics occur at frequency below or greater than 20 kilohertz or you want to study about acoustic emission which occur from anywhere from 2 to 5 megahertz. So, if I was to record any ultrasonic signals or capture an acoustic emission phenomena I have to be sampling at very very high frequencies could be you know 100 megahertz and so on and these a to d cards are very costly very difficult to find at times. Now, the less nowadays systems are coming in where such sampling devices are there and then we eventually we have to store this data n data recorder data points n data points and from the previous example you just saw that by playing around with the number of data points sorry by playing around with the sampling frequency and the time data time which you require to record we can increase the number of data points. Now, so how do such digital data recorder replace the conventional you know you recall the analog data recording people are doing by hand how does it replace how does it replace analog recording. So, in this data logger which we will call I can decide frequency of storage may be continuously and imagine I have a signal which is low frequency content and another which has high frequency content like the case of the thermocouple and the actual number. So, depending on my interest low frequency content could be sample at a low frequency high frequency content could be sample at a higher frequency the in the same data logger the number of input channels is also a very important characteristics. For example, you know there are 16 channel recorders even there are 128 channel recorders and so on that depends on the interest of course, we all the previous examples we have done with one channel if simultaneously I would not record 16 channels of data or 128 channels of data I need to also record them and multiply the number of data points by the small n and small n is the number of channels. And this data points has to be stored in a memory which is RAM memory may be which could be erased and then if you want to have a permanent record you can write to CD DVD or even put it in your hard disk or may be also in a memory card or a device. Nowadays you know you get cameras you know flash cards with you know 32 GB or 64 GB storage capacity. And these are traditionally you know compared to the data recording facilities we had in the 1960s where in a typical spool or a cassette tape would at most have you know 60 minutes or 90 or 120 minutes of data could be stored. This was the duration of the spool cassette type 60, 90, 120 obviously we cannot do continuous data recording over months in such recording because of the limitation of the storage space. But nowadays with such compact memory cards with capacity of 32 or 64 GB you know 64 GB would last you quite a may be you know just showing example in the data recorder how this can be stored. So, come back here this is a typical data channel recorder which we have in our laboratory this is a this is where the digital cassette tape goes in. And if you will see here there are 8 channels of input and 8 output channels and then you can record about 120 minutes of data in each tape at a maximum frequency of 20 kilo hertz in each of the 8 channels. So, the number of channels times the frequency bandwidth is a constant in this kind of data recorders. So, it is 8 channels at 20 kilo hertz and then you could be if you want to make more channels if you want to make 16 channels it has to be sampled at 10 kilo hertz and so on. So, this option can be there are 32 channels at 5 kilo hertz, but the limitations of such data recorders are I obviously cannot record for months together recorder is not possible and then I obviously cannot record at very very high frequencies may be in mega hertz. And then of course, again this has to be stored in data tapes which could be contaminated. So, of recently quite you know in the last may be 5 6 years people are using this digital data recorders like you see one which we have in our laboratory here. You can have 16 channels of record and then you can give in signals and for such a recorder we can have very very high sampling frequencies and then we can record lot of data points. Just give an example for this digital data recording with about 2 giga point of memory space there is a provision of changing the sampling rate 100 mega samples per second or 10 mega samples per second with single channel I can be recording at 20 seconds of data 100 mega samples you know mega hertz per second. So, very very high sampling rate whereas you know even you know we were happy with 5 kilo samples per second. Even if you for a mechanical signal if you sample at 100 kilo samples per second with single channel you can record 5 hours of data and if you are running at very very slow phenomena like monitoring temperatures 200 samples per second that itself is very very high while monitoring temperatures you could be recording with single device for about days without worrying about your memory space about the reproducibility of such signals and so on. A very important point comes in this recorders is the how do you determine the frequency response of the data recorder you could also try this with the digital recorder a simple way to do this is suppose we have a data recorder all you have to do is record a random noise signal at the input and then play back the signal output signal and then simultaneously plot this as a function of frequency play back signal by the input signal. And you will see a frequency because if the play back has to be of the same level as the random noise signal and if you have recorded a random noise of a particular band width you will see and in a dB scale this would be 0 dB this value because in log of 1 is equal to 0. So, if play back is of the same level as the input signal it will be 1. So, it will be 0 decibels. So, this is the useful frequency limit of the signal. Research it actually does this, but it is a good idea to check it out if you have an FFT analyzer you can also see the if you have an FFT analyzer you can also see the phase response. So, the phase of the transfer function between the play back and the input and this phase angle should also not vary and this usually this phase angles typically are the of the order of 2 to 3 degrees. So, then you can see the frequency limit of the signal. So, the upper frequency limit is decided by doing such a small test using a random noise in a data recorder and you could use that same technique in a digital recorder as well and there should not be any crosstalk between between the channels. Now, once we have recorded this signal we also need to transfer the signals and usually the transmission rate of the signal or signal transmission rate depends on the number of channels multiplied by the acquisition rate and the resolution. So, the data transfer rate even if it is given in bits per second and acquisition rate in samples per second resolution in bits per second. So, the data rate the data transfer rate transfer or transmission rate is the number of channels multiplied by the acquisition rate multiplied by the resolution and acquisition rate it is the unit is samples per second and resolution is bits per sample and the units of data transmission rate is bits per sample sorry bits per second. And usually when we use wireless for data transmission rate we have about the maximum level of 10 mbps or you know 1 gb gb. So, we can decide what would be the sampling rate and this is the limitation of the present day data transmission by signal by wireless transmission as to the maximum data transfer rate available to us and which is about 10 mbps or 1 gbps and then this depends on the number of channels of data you are transferring the acquisition rate and resolution. Typically the resolutions nowadays people use is about 16 samples per second or 24 samples per second and the sampling rate for a low frequency signal it is easier to transfer by wireless, but when we are talking about high frequency signals you know you want to sample or send and 10 channels of vibration data over the wireless the data transmission rate of the present wireless technology is limiting these levels of data transfer. And so that would conclude our discussions on the signal recording in the analog digital domain and the signal transfer by in the wireless. Thank you.