 So, we will start with this important aspect of mechatronic systems, which is a selection of sensors and actuators. We have done the modeling so far, so we need to use this model to now get to some of these estimation of sensor parameters or a sensor you know specifications that we need to get to and actuator specifications ok, we will see some relationship and some interesting ways of computing the sensor or computing or arriving at specification of sensors and actuators. So, I will talk about little bit more detail in the sense we may not have to do entire simulations to kind of get to this, but we can do some kind of you know ballpark estimations of the parameters in some way. So, as we go along we will we will talk so, let me get into the slide show more here ok. Now we first bother about what are the interfaces for sensors and actuators and we have seen already like you know couple of interfaces so, PWM you may have the digital analog converter for actuator or say simple digital output for actuators like solenoid ok, for sensors you may have ADC and a digital conversion interface that we have we are here to see that in diva we may see at some point or you may take a tip in the project digital input outputs we have seen for sensors again we can use that and encoder interface as QA interface we have we have programmed and you may use UART in some cases where the data from sensors is available in UART kind of a form already processed and you get only the numbers in the in the micro controller ok. What are the important considerations for like development of this sensor actuator specifications is first whatever are the system requirements like what is the requirement for your control application or megatronics application that we need to know first or we need to kind of we given abstract problem we arrive at those requirements first and then we move on to sensor requirements ok. So, we map them to the sensor requirements and for this mapping we will need a modeling for the system ok. So, what we have seen in the different forms of models like Lagrangian formulation given of the thing or Newtonian formulation. So, whatever like know these models that we are doing some mathematical development that can be used for this map ok and in fact, that should be used for this map because then we will be able to kind of very clearly get all the you know necessary specifications and we do not over design actuators or sensors ok. So, for that you need this kinematic dynamic analysis to get to do that. Then other thing is like you know you need to make sure that about the noise considerations. Now system specifications will be in terms of some resolution of positioning that is required or resolution of some whatever output that is required and we need to make sure in the presence of noise we get that kind of a requirement satisfied and noise. So, you need to see what is the noise in the environment around it and like know what is what is to be done to suppress the noise and some kind of a filtering that need may need to be done. All those kind of things are to be accounted for will dwell more about noise little later actually. So, you need to keep this in mind when we develop the sensor specifications ok. Then you can have a choice of analog versus digital. So, mostly the recommendation is to go for a digital sensors for making sure that noise consideration is not bothering too much actually and analog you will do only in cases where like you do not have any option for the digital sensor. For example, strain gauge for a to sense you typically will go for analog strain gauges there is no digital output for the strain gauges. So, you use that an amplified output of that then you need to kind of bother about some noise filtering and then use that data from ADC channel to convert. Then you will need to of course match the interfaces and its compatibility in its interface specifications. For example, if you are using encoder interface and you have some range say 50, 60 mm and some resolution for encoder then you get a number of bits. So, the number of bits of the encoder channel should match the range requirement for you. So, number of numbers that are getting represented within that range say 0 to 60 mm I want to kind of go and I have a resolution of maybe 0.01 mm or like that is one encoder count or that is what if it is given then one can estimate how many number of numbers that will get represented from going from 0 to 600 0 to 60 mm and then correspondingly I would have number of bits estimated and like say that encoder channel must have a minimum these number of bits to kind of get the sensing done properly. So, that within that range I do not have to have deal with this number overflow situation scenario. Then of course, use of filters I talked about for noise consideration one has to do that and then filter has another implication for increasing your sampling time because the filter calculations have to happen within the sampling instance. So, again those details will talk little more in the future. There is another thing that you need to think about whether you want analog filter or digital domain filter to be done. Analog domain filter and so, this filter discussion will will reserve for the future part ok. When you go for digital domain filters then only like you will have to do computations in the microcontroller in analog domain filter you may not need to reserve a time in the sampling time for filter calculations. Then of course, cost implications and then you you can do this mathematical modeling and do the preliminary simulations to get to the sensor requirements or actuator requirements and then of course, one has to see the manufacturers catalog to do the proper choices because you may not find exactly of all the specifications you have drawn. The sensors may not be available, but there will be some conservative like you know the the the you will have some sensors which are beyond satisfying all these specifications and having some more kind of a you know additional features we can go for that ok. So, we need to have this minimum estimation done and then choose something which is better than that that is how we choose from the manufacturers catalog ok. So, let us take an example of linear motion stage and then work about how do we get going for the specifications here. So, you have say motor stage we considered like you know lead screw configuration which with spring loaded elements for. So, when we have such a kind of a system motion stage with a lead screw and we have given this problem ok. So, we would like to have this closed loop control implementation ok for such a stage such that the drive accuracy is about 5 micron and speed of operation to be achieved is about 20 mm per second and range is given to be 60 mm. So, what should be sensor and actuator specifications? So, this is like a typical problem that could be appearing. So, some of the data may be given, some of the data may be missing and you need to kind of make assumption or make a reasonable assumption for the application that is required. So, this is now we are given this already. So, this may not be in this form some people may give ok oh look I want to use this stage for biological slide scanning purpose ok. Now, only with this kind of a specification given you need to ask some more questions to the people for whom you are kind of developing this stage and get to know like know what are the more detailed specifications that they require in terms of design ok. So, these design specifications arriving at is based on the what is the requirement for a user of the application. Then we will need to get for more detailed kind of a now sensors. So, here you based on this data given one can kind of now come up with say ok if I mount this sensor on my motor side and then there is a lead screw. The lead screw will have some specifications of like know lead for one rotation is say 1 mm, 2 mm some kind of a specification. Then one can say for more motion of 5 micron I need to have some kind of a kinematics estimated to see how this 5 micron on the lead screw side or on the motion side like know linear motion side translate to the rotary motion side that you can do very easily by using some kind of a you know simple kinematics ok. So, that kind of a estimation needs to be done. So, to get to actual sensor specifications. Then other thing like for example, to get to actual specifications one has to kind of do some kind of a think about some kind of a dynamics ok. So, we will get to that what are the moving masses in the system and things like that ok. So, I will give you some kind of a broad outline of the process that can be followed. So, where to start I mean know. So, this again all these things will be little bit of a common sense if you keep it active like things will be falling in place. So, how do we go about doing this what you think? Just think about like ok suppose you are given these specifications and now you want to get to the sensor and actual specifications as we have seen like know considering this lead screw one can get of get to little bit of kinematics. So, I will just give you some broad outline of the process that you will follow. I mean it is just a common sense I mean if you think there are some additional steps that can be done this is all up to you ok. So, see first like you know you can do this physical construction of system in sketch and observe like know what are the kinematic connections in relationships ok. This is very simple for this linear motion state, but it may get into little bit more complexity for more complicated kind of problems. So, we will see we will say similar kind of a thing one more example of two R manipulator to get to this kind of discussion. So, then now you can get sensor resolution and range specification based on these kinematic relationships as I said for like you know the specification phi micron on X side motion how much it translates to the rotary motion for the sensor. And typically if phi micron is a resolution and corresponding to that you have estimated resolution on the motor theta side you will need to be better than that from the from the digital control perspective. Because see if phi micron is so let us say corresponding to phi micron let us say it is some alpha angle on the motor side. Now, if you select resensors which is sensor which is exactly of this alpha resolution ok. Then the problem is that like your control is happening every sampling instance and then there are many different kind of other elements that are there for example, some of the noise considerations. Noise in this case would it would be like you know it is a digital sensor if you select then noise from the sensor may not be there, but there be some kind of a friction in the in the system which will affect the positioning resolution ok. So, there are the errors that are there in the positioning that will happen based on some kind of a dynamics of a system that need to get accounted for in the in the resolution ok. So, we need a better resolution than alpha and it is typically like one fourth or one fifth of alpha that we can go for. So, it is very hard to get single count of so say see. So, when we say resolution is alpha it is a single count of the sensor which is alpha. So, one count within one count you get to the accuracy it is very hard problem to solve from you know digital control perspective. So, it may be better off by solving up to two or three times it is comfortable to solve like you know. So, your resolution is alpha by three for the sensor and what you get after like you know the control is alpha that is scenario. And why that happens we can see some some part in the simulation see there are many different error inducing that happens along the different elements in the whole of the path of control ok. So, one of them is like say you can see the sensor quantization. Quantization aspect will have some kind of effect on the dynamics. Like that there are the noise I have talked about then like in the friction I have talked about. So, from this perspective like you know you need a better kind of a resolution than your estimate of what is required from the you know given the user specification or application specifications ok. So, this is an important point to note here and one has to kind of this if you to get to that exactly it is a hard problem ok. So, one can work with this ballpark kind of a estimates that you know base comes based on mainly kind of experience that you are able to kind of position within like know. So, when we are within say 3 to 4 times like you know the resolution of the sensor actually is there. So, that is why like you know your sensor resolution needs to be little better than what you get by just this you know kinematic relationships estimate ok. And that typical number is maybe 3 or 4 times better. Then you can note this other step that you follow is like you know noting like moving elements in the system and give some rough estimates of what is their mass and other stuff. Then so, then they can be kept as a variables sometimes depend upon like you know what is the application that you are dealing with. Then you can determine the equivalent inertia or load inertia and load on motor that is coming and then this will help to get the motor specifications done ok. Now the motor specifications would come not by like you know you can one way you can go about it is millet entire kind of a process of whatever you want to achieve little cube or some, but you can get simple ballpark estimates by using like this you know force into velocity of power relationships like you know force into velocity or torque into angular velocity this kind of a power relationships you can use. So, you can go directly like you know say you are given this linear stage the force on the output side is estimated is friction force and some inertial load force that is estimated to be some value and then you have some velocity maximum velocity that is coming up in the system. So, you can use maximum force and maximum velocity to get like you know the conservative estimate of total power required by the system or one can kind of translate that into torque and angular velocity on the motor side and then do that ok. So, wherever you do power calculations like you say you do say power calculations on the load side if you have done by this linear state for example, for this linear stage then while translating to the motor side one can use some kind of a efficiency of power transmission ok depending upon like you know friction say for example, in this case you have a spring loaded element that is coming up right. So, that is you remember this stage assembly you have this spring loaded element which is loading onto the screw. So, as to kind of prevent a backlash now with this spring loading will your friction will be very high in this case. So, your transmission efficiency will go down because of the friction and you need to kind of account for that while building your actuator. So, right now we are thinking of using some kind of a servo motor as an actuator in this case and this will start affecting the power required for the servo motor ok. So, one can start with the estimates many times when you do not have a actual kind of a thing and then. So, this design process or like this is little bit an iterative you start off with getting some kind of a details and then use those details to get to the manufacturer catalog to select a proper sensor and once you select a sensor you see other more details information and then see whether that will suffice what you have you know made a choice ok. So, and if it is not then again iterate like you know next time to kind of select other kind of specifications ok. So, that is how the process will typically go on and this process you may probably need to use it in in your project now. When you do not have like no real choice in these sensors and actuators we may do that ok oh look I have estimated, but I do not have whatever sensor I have is this only because of the resource constraint then you can say ok how this suppose you select that kind of a sensor how it will translate to the you know resolution or accuracy that you will achieve for your final task ok. So, that translation that that we would like to see in your you know project presentations and anything like that ok. So, we will select for example, here for linear encoder we will select for these if you are really interested in the output side positioning and we know that there are some kind of a backlash elements in the system which will affect linear to rotary part, but if we say ok we do not have much of a backlash or backlash is taken care of by the spring loading then rotary encoder can also be sufficient because we know that there is no uncertainty between the input and output ok. So, then so, rotary encoder if you want to select then it will be much what do you say coarser resolution than a linear encoder because there is some kind of a lead screw connection between the linear and rotary part and for a for a entire rotation of 360 degree on encoder side or or motor side you will have a displacement only of 1 mm on say linear side ok. So, that that is corresponding to the lead of the screw and that is how like know your requirement of coarser or like finer resolution will change based on where you are mounting the sensor. So, you need to think what is the place that you are going to mount your sensor on and according to that the resolutions have will change and one has to be one needs to be little smart to choose this properly. So, that not to harm the applications also and at the same time you do not increase the cost unnecessarily ok. So, that is how like know this choice of mounting the sensor at appropriate place would have some kind of you know thinking required required. Then as I said earlier this we decide like know diameter of the lead screw and lead of the lead screw. Now, this all like know the choice of the lead screw and the sensor you know there now we can see that they are coupled here ok. So, once you choose the lead screw a particular thing for the for the same application the sensor requirements gets frozen or if you think ok know I want to kind of like know have some some kind of a better possibility that I do not want to use to costly sensor here I want to have coarser sensor used and I can choose a lead screw which is having lesser lead than you know previously a previous choice ok. So, like that one can think about and this is this is all I think the the based on some kind of a simple common sense understanding of you know these concepts of kinematic relationships ok and making use of them to to think and get get to what we want finitely to achieve ok. There will be these considerations of manufacturing and tolerances which will have some kind of a cost implications. So, we need to make sure that these tolerances on manufacturing are not like know directly seen on the on the you know resolution that we want to achieve for the for the sensor. Say for example, if I want a 5 micron resolution in the motion my manufacturing of the lead screw should have a better kind of a surface undulations than 5 micron ok otherwise the surface undulations will start moving the stage there and then like know I may not get a perfect linearity that that I would look I am looking for ok. So, those kind of considerations will be will be there in addition. But this is about surface finish surface finish would be most most of the times like no better or what happens is many times as we do the operation I mean more and more you use use screw then like know it will get more and more smoothened and it may get better. You might have seen you know your car also has this kind of a brand new purchased car the piston and cylinder you know there is these tolerances are such that their friction is little higher. But as you kind of start using like running the car in first few months like you know they get automatically better by where little bit of where that is happening there. So, rotary and this we talked about now based on the speed required ok the dynamic response also is important to see. So, based on speed that is required you will have a good dynamic response estimated like know on the encoder side. So, say speed is given on the load side say lead screw x speed we saw 20 m per second what it translates to on the on the motor side ok on the rotary side which is how many pulses per second are coming. And we need to have a sensor which can give you output which is resolving those pulses nicely ok. So, if the sensor inside is such that like the pulse has some kind of a finite rise time and like the fall time you may not be able to kind of get those pulses nicely for very extremely high speed ok. So, you may observe that maybe we can see you may not you do not have oscilloscope to observe, but we can try doing this experiment where we start running motor pretty high speed and see that you know you will find your encoder may start giving some spurious data ok. The data will not be completely matching what is happening there ok. So, that kind of a scenario would happen and at that time if you kind of look at the signal in the oscilloscope the pulses will not be very sharp like square pulse, but it may be more like a like you know half way to rise and then again fall ok. Because the rise fall rise time and fall times there kind of start now merging without any flat on the complete rise may not happen there ok. And then you will get the resolution based on the resolution you get the resolution and range you will get a bit accuracy for the channel and then this dynamics. So, there are some kind of a dynamic consideration for the sensor, but they are typically not so stringent actually. Many times sensors will be responding much faster than your application requirement, but application requirement especially if the speeds are very high ok. Now, these are relative terms very high you get to know what is very high only after like the estimation of these and like you know calculation and seeing ok oh these are the number of pulses coming in say one revolution and that is why one pulse time will be some something like that and then you estimate that pulse per pulse time seeing that ok. Now my rise time and fall time if they add up to the pulse time then you will find that ok oh no that encoder is now on the verge of getting worst kind of a scenario ok. So, that needs to kind of be given some consideration ok, but most sensors you will find that this requirement is not very stringent ok.