Keyboard Shortcuts
ctrl + shift + ? :
Show all keyboard shortcuts
ctrl + g :
Navigate to a group
ctrl + shift + f :
Find
ctrl + / :
Quick actions
esc to dismiss
Likes
Search
Pole/Zero analysis
|
--- In LTspice@..., Andy <Andrew.Ingraham@...> wrote:
I would like to do a pole/zero analysis of Vout/Vin.I don't think Pole/Zero is supported. I've read that it Hello, Yes, this is what happened with PZ in LTspice. The free SPICE OPUS has .PZ analysis. Best regards, Helmut |
|
On 9/11/2011 11:37 AM, Andy wrote:
Here is what Helmut said in message 44296.I would like to do a pole/zero analysis of Vout/Vin.I don't think Pole/Zero is supported. I've read that it never really "Please search our messages with the following two "secret words". pole zero You will get 285 hits. Please read them all. :-) You can take a short. There is no pole zero analysis in LTspice. To be honest, this pole zero feature in standard SPICE is by far not good enough for real usage." Howard |
|
Ganesan
Why is PZ hard to get to work?
toggle quoted message
Show quoted text
On 9/11/2011 11:48 AM, Helmut wrote:
|
|
--- In LTspice@..., Ganesan <dg1@...> wrote:
Not much point focusing on the WHY PZ analysis is not supported! The long answer is probably that Mike dose not see a need for PZ analysis. This is probably a function of SwitcherCAD having started out life as a SMPS simulator and PZ analysis is a fairly useless in tool for highly non-linear switching circuits. If you use PZ for this type of circuit the results will be highly misleading. -robert |
|
John Woodgate
In message <j4kc7c+m7e7@...>, dated Mon, 12 Sep 2011, RobertTalty <rtalty@...> writes:
Not much point focusing on the WHY PZ analysis is not supported!As I understand it, software PZ analysis as normally implemented doesn't stop at realistic frequencies and pole or zero co-ordinates, because it can't be told what they are. So you get as a result every pole and zero from DC to light, some with extreme co-ordinates, and not all of the irrelevant ones are outside the useful passband of the circuit. -- OOO - Own Opinions Only. Try www.jmwa.demon.co.uk and www.isce.org.uk John Woodgate, J M Woodgate and Associates, Rayleigh, Essex UK When I point to a star, please look at the star, not my finger. The star will be more interesting. |
|
Ganesan
PZ analysis is an important tool in Linear System identification..
toggle quoted message
Show quoted text
The dc to daylight problem can be easily solved by the user putting in "freq min" and "freq max". I t can be further solved by the user putting in a few poles and zeroes, he or she can easily identify as initial guesses (invariant or weakly variant). Further, the user can specify a limit on the total number of "poles np" and total number of "zeroes nz". Also while the default can be a mean square error criterion for convergence, user specified regions of accuracy (some pass bands, transition bands,delay over pass bands, placement of notches, 3db bandwidth, etc will greatly improve the accuracy. of the answers.) I have used some system simulators in the past with some of these type of features and they always produced great results.. (As usual remember Garbage in ---> Garbage out. A good straw man is to work with is a RLCladder with widely varying Rs, Ls, and Cs) (as we go to higher speeds, the movement of parasitic poles and zeroes with the chosen gain is becoming more critical and impedance matching has always been a sticky issue) Cheers AG On 9/12/2011 2:26 AM, RobertTalty wrote:
|
|
Out of curiosity what program has all of the pole-zero analysis capability you describe.
toggle quoted message
Show quoted text
Howard On 9/12/2011 5:27 AM, Ganesan wrote:
PZ analysis is an important tool in Linear System identification.. |
|
Ganesan
Boeing, Lockheed,Nothrop Grumman, IBM and Bell Lab's Capecod to name a
toggle quoted message
Show quoted text
few... Control systems guys eat this for breakfast , lunch and dinner...Large system identification programs like earthquake modelling, economic forecasting are some other one that come int o mind. Beyond that I will be getting into customer proprietary stuff and that is not good for business... Cheers AG On 9/12/2011 4:21 PM, Howard Hansen wrote:
|
|
AG,
toggle quoted message
Show quoted text
sorry if you understood my response to be questioning the value of PZ analysis. Let me state unequivocally that PZ analysis is a VERY a valuable systems analysis tool especially for feedback control systems. Unfortunately LTspice does not support PZ analysis. It is a big failing, however, given the choice, I would much rather they simply not support a tool than support a misleading tool. Old style spice PZ analysis was useful only in linear control systems, however in today's world of sampling the data right at the source, we really need PZ analysis that effectively works with system level Z-domains and S-domains descriptions. Ltspice is a useful Z-domain simulator BUT Matlab is much better so I do PZ analysis in Matlab (simulink). regards Robert --- In LTspice@..., Ganesan <dg1@...> wrote:
|
|
Ganesan
Thanks .. Points noted..... But Matlab is not for free....lol
toggle quoted message
Show quoted text
cheers AG On 9/12/2011 9:09 PM, rjtalty wrote:
|
|
Ganesan
How do you propose to understand the effect of parasitic poles and
toggle quoted message
Show quoted text
zeroes on impedance matching and on loop gains as we start pushing the bandwidth limits...? Right now I have a 1.2 GHZ single stage amp.. which has at least 5 parasitics which in the past I ignored, but cannt do so anymore? And it is a pain in the butt to go back and forth between any Spice simulator and Matlab...( they just dont talk the same way) cheers ag On 9/12/2011 9:34 PM, Ganesan wrote:
|
|
John
Is that really true? I don't really work in the s plane usually, or in any practical way, but it could be very useful when one has inadvertently introduced a zero to know where it is. Also, being able to pinpoint poles would make no end of difference - it would save lots of calculator time between screen shows. It also strikes me as an advance in general. If what one saw could tempt users to view things in a more basic way they would understand the working 'tant mieux'. It would advance the 'feeling' about a circuit. Many may not agree, but I think 'feeling' is good in design. If something can put in front of you all the toys that you are playing with, then how can that be worse? And you often can't see on screen, in the tenths of a degree of phase shift, if anything is going on whatever, when you are looking within your bandwidth. I have a delicious example of this. Why should an amplifier, driving well designed speakers, sound better when it has a 0R22 resistor on its output? It does, and there is a reason, and it has nothing to do with damping factors. The resistor is nominally there to aid stability, but it in fact characterises the whole amplifier. Looking at the result of high pass and low pass sections of a speaker is the way to find out what's happening here (both of which get better by the way), but wouldn't it be lovely if LTS flagged the zero that had been introduced. "Hoy mates, brand new zero. " Don't know what you've done, but here it is right under my finger!" That has to be worthwhile, doesn't it? I brace myself for Woodgate's searing reply as he cauterises my ignorance. CT , On 12 September 2011 08:53, John Woodgate <jmw@...> wrote: ** [Non-text portions of this message have been removed] |
|
Robert
Are the outputs of the switching circuits really that non-linear? Every bit of work i have done on what I assume are power factor corrected circuits has presumed that I can draw as I please. Is that not the case? If PFC isn't getting the phases directly in line, then what is it? Is it a very weak relationship with some sort of impedance associated? Obviously there must be more than I know because it is not my area, but I would then have to ask where the power is being dissipated? Do the power supplies get stupidly hot? This isn't necessarily an LTSpice modelling problem, but if you have something I should know, then please send me the model of how it works in real life. Non-linear is a very subjective term unless there are manifest discontinuities.Do they really not happen the other way around? I would, quite seriously, like to be given examples of non-linear behaviour. It is not that I don't believe they exist in the electrical world, though they are few and far between (let's put harmonic distortion in the sub 1% category to one side) but in mechanics they seem to be taken for granted while I reckon they hardly exist. (On that score I just see different impedances). CT On 12 September 2011 08:26, RobertTalty <rtalty@...> wrote: ** [Non-text portions of this message have been removed] |
|
John Woodgate
In message <CANj54jw=OCHDA5zXdS_8=3LGOxBDgPMjQGtncXrsFWtb=bdyVg@...>, dated Tue, 13 Sep 2011, Christian Thomas <ct.waveform@...> writes:
"Hoy mates, brand new zero. " Don't know what you've done, but here it is right under my finger!"It's at 13.5 GHz and has a Q of 0.01.(;-) I leave any searing to analogspiceman. I'm reporting my experience with a Spice that does have pole-zero analysis, and I understand that others have the same fascinating features. Whether it's possible to introduce a sense of realism to PZ analysis **in the context of Spice**, I have no idea. No doubt it *is* possible to write a stand-alone program whose settings can be adjusted so as to ignore extreme frequencies (at both ends) and extreme singularity co-ordinates. -- OOO - Own Opinions Only. Try www.jmwa.demon.co.uk and www.isce.org.uk John Woodgate, J M Woodgate and Associates, Rayleigh, Essex UK When I point to a star, please look at the star, not my finger. The star will be more interesting. |
|
Tony Casey
<snip>
I have a delicious example of this. Why should an amplifier, driving well</snip> Hello Christian, Doubtless, I will be reminded this is OT, but... If it is *really* so that the insertion of a 0R22 resistor improves the sound of an amplifier/speaker, then regardless of what you say, one or both of the them clearly has a design shortcoming. This is a nasty kludge, and normal users shouldn't be expected to have to do this kind of thing to get the best from their systems. Speakers should be designed to be driven from a voltage source; amplifiers should be stable into any load. If the resistor really is necessary, then it should already be incorporated into the speaker. Regards, Tony |
|
John Woodgate
In message <j4n0hu+hrlt@...>, dated Tue, 13 Sep 2011, Tony Casey <tony@...> writes:
If the resistor really is necessary, then it should already be incorporated into the speaker.You mean, in series with the 6 ohms or so (when it's cold!) of voice-coil resistance?(;-) -- OOO - Own Opinions Only. Try www.jmwa.demon.co.uk and www.isce.org.uk John Woodgate, J M Woodgate and Associates, Rayleigh, Essex UK When I point to a star, please look at the star, not my finger. The star will be more interesting. |
|
Tony Casey
--- In LTspice@..., John Woodgate <jmw@...> wrote:
Hello John, Of course you are correct in the case of a single MC driver, but due to poor (crossover) design, some systems exhibit ridiculously low impedance at some places in the frequency band. In cases like this, it is entirely possible that the resistor improves things, but only insofar as it disguises a design failing. Regards, Tony |
|
John Woodgate
In message <j4o6mi+dnim@...>, dated Tue, 13 Sep 2011, Tony Casey <tony@...> writes:
Of course you are correct in the case of a single MC driver, but due to poor (crossover) design, some systems exhibit ridiculously low impedance at some places in the frequency band.Some might; I know one well-known manufacturer had crossovers burn up about 40 years ago due to a 'parasitic' series-tuned circuit in the crossover than no-one had noticed. However, there is also a dynamic effect which causes high peak current demands even though the steady-state impedance is well-behaved. This effect was pointed out by Matti Otala (he of TIM ?fame?), with fairly complex examples. But Peter Baxandall pointed out that a simple RC circuit does it with a low-frequency square wave voltage applied. As, of course, LTspice will willingly demonstrate (must get back on-topic somehow!). In cases like this, it is entirely possible that the resistor improves things, but only insofar as it disguises a design failing.But 0.22 ohms is really too little, in most cases, to fix the problem. There is a long story to this, featuring another of P Bax's invaluable contributions, but it really is OT here. -- OOO - Own Opinions Only. Try www.jmwa.demon.co.uk and www.isce.org.uk John Woodgate, J M Woodgate and Associates, Rayleigh, Essex UK When I point to a star, please look at the star, not my finger. The star will be more interesting. |
to navigate to use esc to dismiss