Says in a paper I have found The root-locus method of Evans is one of the most popular and powerful tools for both analysis and design of single-input single-output (SISO) linear time-invariant (LTI) systems I say hogwash. Maybe amongst academics in the classroom but I have never heard of anybody using root locus for a serious problem. Bode is still by far the most popular method. Hardy
Root Locus
Started by ●January 15, 2013
Reply by ●January 15, 20132013-01-15
On 1/15/13 1:50 PM, HardySpicer wrote:> Says in a paper I have found > > The root-locus method of Evans is one of the most popular and powerful > tools for both analysis and design of single-input single-output > (SISO) linear time-invariant (LTI) systems > > I say hogwash. Maybe amongst academics in the classroom but I have > never heard of anybody using root locus for a serious problem. Bode is > still by far the most popular method. >i've never used it since a controls course in grad school. i don't even use Bode. if it's an IIR (or a TIIR), i just look at how far away from the origin are the pole locations. -- r b-j rbj@audioimagination.com "Imagination is more important than knowledge."
Reply by ●January 15, 20132013-01-15
"HardySpicer" <gyansorova@gmail.com> wrote in message news:98cc43a2-34d4-45a7-86df-dd40cb7fb980@rm7g2000pbc.googlegroups.com...> Says in a paper I have found > > The root-locus method of Evans is one of the most popular and powerful > tools for both analysis and design of single-input single-output > (SISO) linear time-invariant (LTI) systems > > I say hogwash. Maybe amongst academics in the classroom but I have > never heard of anybody using root locus for a serious problem. Bode is > still by far the most popular method.Both Bode and root locus are pre-computer era qualitative methods to get ballpark estimates in simple scenarios. Why would anyone need that today, especially for serious problems? Vladimir Vassilevsky DSP and Mixed Signal Consultant www.abvolt.com
Reply by ●January 15, 20132013-01-15
On Jan 16, 8:10�am, "Vladimir Vassilevsky" <nos...@nowhere.com> wrote:> "HardySpicer" <gyansor...@gmail.com> wrote in message > > news:98cc43a2-34d4-45a7-86df-dd40cb7fb980@rm7g2000pbc.googlegroups.com... > > > Says in a paper I have found > > > The root-locus method of Evans is one of the most popular and powerful > > tools for both analysis and design of single-input single-output > > (SISO) linear time-invariant (LTI) systems > > > I say hogwash. Maybe amongst academics in the classroom but I have > > never heard of anybody using root locus for a serious problem. Bode is > > still by far the most popular method. > > Both Bode and root locus are pre-computer era qualitative methods to get > ballpark estimates in simple scenarios. > Why would anyone need that today, especially for serious problems? > > Vladimir Vassilevsky > DSP and Mixed Signal Consultantwww.abvolt.comBy this you mean you would throw everything on Matlab and simulate it. Well I suppose so, but many folks don't need Matlab for say designing a PLL circuit using lag-lead control. Of course that may not be a serious problem but a skill worth knowing. Are you suggesting that we stop teaching Bode altogether? Hardy
Reply by ●January 15, 20132013-01-15
"HardySpicer" <gyansorova@gmail.com> wrote: "Vladimir Vassilevsky" <nos...@nowhere.com> wrote:> "HardySpicer" <gyansor...@gmail.com> wrote:> > Says in a paper I have found > > > The root-locus method of Evans is one of the most popular and powerful > > tools for both analysis and design of single-input single-output > > (SISO) linear time-invariant (LTI) systems > > > I say hogwash. Maybe amongst academics in the classroom but I have > > never heard of anybody using root locus for a serious problem. Bode is > > still by far the most popular method. > > Both Bode and root locus are pre-computer era qualitative methods to get > ballpark estimates in simple scenarios. > Why would anyone need that today, especially for serious problems? > >By this you mean you would throw everything on Matlab and simulate it.God forbid. This is what I mean when I say matlab for stupidents and matlabi.>Well I suppose so, but many folks don't need Matlab for say designing >a PLL circuit using lag-lead control. Of course that may not be a >serious problem but a skill worth knowing.There is no need to play games with Bode plots and like when you can directly solve for H(s) and X(t) on computer without any sweat.> Are you suggesting that we stop teaching Bode altogether?I say 90% of calculus is about obsolete tricks of pre-computer era. In our days it is an art for the sake of pure art. May be interesing to a handful of specialists, but definitely not for mainstream engineering. Vladimir Vassilevsky DSP and Mixed Signal Consultant www.abvolt.com
Reply by ●January 15, 20132013-01-15
On 1/15/13 2:38 PM, Vladimir Vassilevsky wrote:> "HardySpicer"<gyansorova@gmail.com> wrote: > >> Both Bode and root locus are pre-computer era qualitative methods to get >> ballpark estimates in simple scenarios. >> Why would anyone need that today, especially for serious problems?perhaps to get a solid theoretical result from more general conditions than just handing a few thousand numbers to a computer and asking it, "Is this thing gonna blow up or not?" ...> >> Are you suggesting that we stop teaching Bode altogether? > > I say 90% of calculus is about obsolete tricks of pre-computer era. In our > days it is an art for the sake of pure art. May be interesting to a handful > of specialists, but definitely not for mainstream engineering.i don't do Bode nor Routh-Hurwitz. but i *do* design things with as few fixed constants as possible. both for reasons of elegance and to gleen knowledge (like you get to see the dependence of some sought parameter on some other input parameter, knowing that dependence or lack thereof can elucidate), but also so i can twist the knob later (if i didn't glue the knob to a fixed position). that's why i don't like just integrating with matlab and the simpson's rule or something like that. at best, i like to come up with closed-form equations that relate one thing with another. and, when necessary, i am willing to make a few "assumptions" or approximations to turn a more intractable mathematical problem into an easier one that might lend itself to a closed form solution. (i remember i did this in a paper i co-authored with Duane Wise about using polynomials to interpolate between samples and figuring out how much oversampling is needed to get to a given S/N.) i think 99% of calc is *concepts* that any EE or physicist should have before touching a computer keyboard. what should happen in the limit? what can we glean from these equations? what can we learn about the intrinsic character of this system we analyze? -- r b-j rbj@audioimagination.com "Imagination is more important than knowledge."
Reply by ●January 15, 20132013-01-15
Vladimir Vassilevsky <nospam@nowhere.com> wrote: (snip)>> Are you suggesting that we stop teaching Bode altogether?> I say 90% of calculus is about obsolete tricks of pre-computer era. In our > days it is an art for the sake of pure art. May be interesing to a handful > of specialists, but definitely not for mainstream engineering.In recent years, the way calculus is taught has been changing. Instruction is going more numerical, where graphing calculators can compute the answer that previously would have been solved analytically. Now, they still teach things like integration by parts, but we used to use integral tables after we learned how to actually do it, and now there are calculators that will do symbolic integration. It used to be that numerical math was all left for numerical analysis class, and only symbolic math (partial fractions, integration by parts, trig. substitution, etc.) were taught in math class. More numerical math, earlier, is probably good, but there might be some loss in not learning the analytical methods quite as well as we used to. Should students be expected to supply analytical solutions to physics and engineering problems, or are numerical answers good enough? -- glen
Reply by ●January 15, 20132013-01-15
On Tue, 15 Jan 2013 13:10:51 -0600, Vladimir Vassilevsky wrote:> "HardySpicer" <gyansorova@gmail.com> wrote in message > news:98cc43a2-34d4-45a7-86df-dd40cb7fb980@rm7g2000pbc.googlegroups.com...>> Says in a paper I have found >> >> The root-locus method of Evans is one of the most popular and powerful >> tools for both analysis and design of single-input single-output (SISO) >> linear time-invariant (LTI) systems >> >> I say hogwash. Maybe amongst academics in the classroom but I have >> never heard of anybody using root locus for a serious problem. Bode is >> still by far the most popular method. > > Both Bode and root locus are pre-computer era qualitative methods to get > ballpark estimates in simple scenarios. Why would anyone need that > today, especially for serious problems?For serious problems involving plants whose models can't be adequately extracted into an s-domain model, for one. (Well, for lots and lots, actually -- I almost never get a satisfactory s-domain or z-domain model from a plant, and I very often get very good measured plant frequency responses). -- My liberal friends think I'm a conservative kook. My conservative friends think I'm a liberal kook. Why am I not happy that they have found common ground? Tim Wescott, Communications, Control, Circuits & Software http://www.wescottdesign.com
Reply by ●January 15, 20132013-01-15
On Tue, 15 Jan 2013 10:50:13 -0800, HardySpicer wrote:> Says in a paper I have found > > The root-locus method of Evans is one of the most popular and powerful > tools for both analysis and design of single-input single-output (SISO) > linear time-invariant (LTI) systems > > I say hogwash. Maybe amongst academics in the classroom but I have never > heard of anybody using root locus for a serious problem. Bode is still > by far the most popular method.The Evans root locus was used by a lot of practitioners in the 50's through the 70's (AFAIK, from talking to old practitioners). I still find it handy for sketching out how the poles of a plant of some assumed type will respond to tuning. They're also, in my opinion, a good teaching tool. But I generally use a computer to make 'em, and I'm not sure there's much value in teaching all the gory details of drawing them up by hand when you can make them easily enough. Bode plots are probably the most popular method among circuit designers. For problems where you can get good plant response measurements by doing a swept-sine measurement but can't get a good s- or z-domain model from _any_ measurement, Bode plot design is a pretty darned good. -- My liberal friends think I'm a conservative kook. My conservative friends think I'm a liberal kook. Why am I not happy that they have found common ground? Tim Wescott, Communications, Control, Circuits & Software http://www.wescottdesign.com
Reply by ●January 15, 20132013-01-15
On Tue, 15 Jan 2013 14:04:39 -0500, robert bristow-johnson wrote:> On 1/15/13 1:50 PM, HardySpicer wrote: >> Says in a paper I have found >> >> The root-locus method of Evans is one of the most popular and powerful >> tools for both analysis and design of single-input single-output (SISO) >> linear time-invariant (LTI) systems >> >> I say hogwash. Maybe amongst academics in the classroom but I have >> never heard of anybody using root locus for a serious problem. Bode is >> still by far the most popular method. >> >> > i've never used it since a controls course in grad school. > > i don't even use Bode. if it's an IIR (or a TIIR), i just look at how > far away from the origin are the pole locations.That works well if you're making filters that live entirely in software. But if you're controlling something real things can break down badly at the step where you go from measured data to a Laplace or z-domain model. Bode plots let you side-step that by using measured data directly. -- My liberal friends think I'm a conservative kook. My conservative friends think I'm a liberal kook. Why am I not happy that they have found common ground? Tim Wescott, Communications, Control, Circuits & Software http://www.wescottdesign.com
