When the width of the line is minus one, andhere the extent of the line is plus one. Now, I want to ask you issues and questions. This Y-axishere represents the variation in S22; how many dB alters are there in S2 2, in the response? In this case, once we have the resistor R1, the X-axis, all the way to plus one, thatchange in S2 2 is about one, two, three, four, four and a half dB. Okay? So, the interactionis giving four and a half dB extra included variation to our S22, merely due to the interaction betweenthe resistor, R1, and the extent of the lines; four and a half dB variation. But , noticeon this side, if I go to the resistor low-grade, to minus one, that alteration in S22 due tothe interaction is reduced to less than 2 dB. Now, as a designer, where would you liketo be? Would you like to be in this area where you have an added four and a half dB? WhenI computed, because you have other modifications from other factors, this is only the interactionsthat are causing this variation.Would you like to be in this four and a half dB variationin S22, or would you would like to be in this area where its less than 2 dB variation? I would choose this area. I cannot control the width of the lines due to process variation, but what I can see is the resistor. I can control the resistor variation, and Ican also reduce the value of that resistor. What “its certainly true it is” trying to tell me is if I reducethe value of this resistor from 20 -ohms to 18 -ohms, for example, and on top of this ifI control the resistor variation by making it a fat resistor instead of skinny, to reducethe sensitivity, I to further reduce that S22 variation and I can be here.Okay, this is a very simple example of theinteraction scheme. As I have just mentioned, some patterns you might find that the interactionsbetween two elements is huge, it is causing major difficulties in the specific characteristics. This is whythe interaction planneds are useful and very beneficial for the designers, so theycan specify their design before proceeding. Now, let me go back to my designing, and letme open another design. This is another Design of Experiments example that I would like toshare with you.Notice now , now, in this experiment, I am racing a Design of Experimenton matching networks. And the style I did this, all the components in the input according network, as shown here, all the components in this matching network I represent them, or I steerthem, in the same direction, leant some variation to it, with the variable X. So, I have X, Y, Z. The X represents the input matching network, Y represents the middle stage, andZ represents the output joining network. So, this DOE experiment is really done noton three variables, it is done on three coinciding networks. So, if you have a design that hasall kinds of equal networks, you know, like, gives say, a four-stage amplifier, then you can run a DOE experiment, Design of Experiments, on just the matching networks.And Design of Experiments will tell you, Hey, the inter-stage matching network between FETnumber two and FET number three is whats compelling the problem in your blueprint. Then, you can go ahead and alteration that inter-stage matching network and solve the problem.So, this is really useful on twinned systems , not on factors simply. So, if I go ahead, right now, and affected simulate, it will, again, led the venture on the three pairing networks , not the three components.And notice, because I have recently been three components, I extend eight ventures. Two to the thirdis eight experiments. Okay, let me open the data display. Okay, and heres the datadisplay for the three parts I have.Lets study them. Lets first look at the Paretochart for the advantage. This is telling me that 96% of the gain discrepancy is caused by themiddle section. You remember I have three pairing networks, the input followed by theFET structure followed by the output fit system. The FET structure is the Y. You rememberX, Y, Z? So, the Y, the FET structure, which has the resistor, R1, if you recollect, iscausing 95% of my amplification variant. And the input parallelling network, the whole input matchingnetwork, is causing exclusively motiving one or two percent of the amplification variation.So I learnedsomething now, that my middle-of-the-road part, my centre joining system that has the resistor, R1, is causing the income variation now. And again, I want to see how much increase variationis that? And I look here and I found out that the resistor is causing about 0.3 dB variationin the gain. Okay? Now, tells look at this noise figure spec.This Pareto chart here for the interference chassis is telling me that, believe it or not, youmight expect that the noise person discrepancy is coming from the input pairing network.Right? The input matching system, in general, is the main sensitive section for racket figure.But, look at this.This is telling me that 95% of my sound representation discrepancy is begun, again, by the middle section, by that FET and the resistor, R1 , not by the input matchingnetwork. The input according network is causing the sound anatomy to change merely, like, 2 %. Lets assure the effect areas is how much is that change. And again, we see that itis very, very small. It is 0.01, or 0.015 dB change due to that resistor, R1. Okay , now well look at the S22 spec, andwe was noted that the Z, which is the output harmonize network, is causing 90% of the variationin S22. Which is anticipated, right? Output matching network, output S22, you are familiar with, the S22? So, 90% of the variation in S2 2 is coming from that Z section, which is the output matchingnetwork. We notice here it exits from -2 to -8, which is about 6 dB change in S2 2 dueto that output matching structure. So, in this example, mostly, I depicted you how matchingnetworks, DOE, can be used on matching systems; this and this and this. And again, if youhave an amplifier, or any designing that has, like, perhaps, 20 matching networks different, perhaps you have a macro-cell that has an amplifier followed by a mixer followed by another amplifier, an LO amp, you are familiar with, to drive the mixer, “youve had” countless pairing networks.You could runa Design of Experiments on all of the join networks. Let say you have ten matchingnetworks, “youre running” two to the tenth ventures, and ADS does them really fast for you. So, you find out which according network is causing the yield trouble. And once you find thisyield coinciding the relent trouble is caused by, makes say, one parallelling structure, man, its just like an x-ray machine. You can focus and you can focus on that matchingnetwork and fix it, and the yield trouble will be eliminated. All the difference willbe eliminated. This is a great tool to use.Its awfully, very helpful. Thank you for watching this Design of Experimentspresentation, and we do have in ADS, we do have an example in the lessons index, microwave routes. We have a MMIC example that has all these blueprint records and all thesesimulation set-ups and outcomes. I support you to go and open this example file and justhit simulate, and watch the data display and learn more how to use it. Its a veryuseful tool. Thank you for watching ..
Free Discount Prescription Drug Coupons
