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
- LTspice
- Messages
Search
|
Re: Freqeucny Dependent resistor
Tony,
toggle quoted message
Show quoted text
I missed this reply of yours earlier. I hadn't thought of the skin effect, though given that is the result of penetration of fields, it must have some reactive element, surely? I did actually look into this in some depth about a year ago in relation to inductor design but more has flowed out than remained. (Though I do remember nothing significant in phase angle within the audio band. But to say it was near zero required about three assumptions to be true which might have been a stretch.) You also succinctly cover the radiation question that John does. ka = 2 is considered the usual limit, incidentally, though on what grounds don't know. ka = 1 might be flat and ka = 2 might give you a bandwidth. I have usually regarded it as an experimental result - although the equations aren't that difficult IIRC, there may be losses that leave the practical results a little awry. But, as you say, plenty of reactive element in there. Christian On 15 September 2011 20:43, Tony Casey <tony@...> wrote:
** |
|
Re: Freqeucny Dependent resistor
Hi again Tony,
toggle quoted message
Show quoted text
My starting question was whether radiation impedance actually held up as only real - his example of a possibility. John then slightly moved the goalposts in response by saying that they were by definition only real; look no further than the power response (and we know that nothing can be dissipated reactively). This was not so damning an argument as it looked, I said, because power is not a vector quantity. I don't care where the power goes. I just wanted to know if there was a direction associated with the amplitude. To keep this on track - perhaps Nigel can tell us whether this was what he had in mind as a VCVS type f dependent resistor. I think we would know about it if it could be done. Christian On 15 September 2011 23:14, Tony Casey <tony@...> wrote:
** |
|
Re: Ideal Swich Model missing
--- In LTspice@..., Michael Stuarts wrote:
The model for SW appears to missing. If anyone can help me createWhy not? Do you have a reading comprehension problem? This is a serious question because Help both clearly explains that you must provide a model statement to define your switch and gives you a link to an example schematic that you may run if you are too unfamiliar with the correlation between the netlist notation example provided and how it derives from the schematic. Perhaps English is not your native language? Or perhaps you really didn't read Help? I am curious because I really would like to understand why Help (which seems perfectly fine to me) does not seem to work for so many users (so there is no wrong answer on your part other than a less than honest one). Regards -- analogspiceman PS: Here is the relevant section from Help. ----------------------------------------------------------- S. Voltage Controlled Switch Symbol Names: SW Syntax: Sxxx n1 n2 nc+ nc- <model> [on,off] Example: S1 out 0 in 0 MySwitch .model MySwitch SW(Ron=.1 Roff=1Meg Vt=0 Vh=-.5 Lser=10n Vser=.6) The voltage between nodes nc+ and nc- controls the switch's impedance between nodes n1 and n2. A model card is required to define the behavior of the switch. See the schematic file .\examples\Educational\Vswitch.asc to see an example of a model card placed directly on a schematic as a SPICE directive. ----------------------------------------------------------- |
|
Re: Freqeucny Dependent resistor
Tony Casey
--- In LTspice@..., Christian Thomas <ct.waveform@...> wrote:
Christian, I don't think John was stating that there is no reactive part, only that power cannot be dissipated in it. Regards, Tony |
|
Re: Ideal Swich Model missing
Tony Casey
--- In LTspice@..., "michaelstuarts" <michaelstuarts@...> wrote:
Hello Michael, From the Help: S1 out 0 in 0 MySwitch .model MySwitch SW(Ron=.1 Roff=1Meg Vt=0 Vh=-.5 Lser=10n Vser=.6) You don't say what your switch will be used for, but I'm betting that if you deleted the last three terms in the model above, it would do what you wanted. If you're having trouble relating that to your schematic, do this: 1. Place a sw from the parts chooser, and change its value to MySwitch. 2. Add a SPICE directive, and paste in the .model text. If you're trying to control it from something like TTL or CMOS, change Vt to 1.8V, or something between the logic levels. Regards, Tony |
|
Re: Freqeucny Dependent resistor
Tony Casey
<snip>
I believe it is impossible to have a physically realizable resistor that is frequency dependent and has no reactive component. A transmission line comes close in that its resistance is almost constant for a range of frequencies, but as you know at very low frequencies, neglecting any series resistive components, its reactance is largely capacitive.</snip> Hello Hubert, You are of course correct: the only resistor possible that had no reactive component would also have zero size. Any conductor of finite size inevitably has both inductance and capacitance, even if it has no resistance, like a superconductor. I think the salient point of the discussion really is whether we can legitimately model part of a system, in which the reactive part doesn't significantly change as a consequence - with a notional frequency-dependent resistor. I think there are converse examples too, where the reactive part changes without a consequential change to the resistive part: a capacitor where the real part of permittivity reduces with frequency without significant change to the imaginary part. A good example of this is an FR4 PCB trace. It's capacitance reduces with frequency, whilst its resistance increases, albeit for a completely unconnected reason. I'm sure there are many more. (I detect the red pen of the moderator twitching, so had better stop here.) Regards, Tony |
|
Re: Freqeucny Dependent resistor
That's a pretty damning point!
But does it hold? You've squared up a vector quantity to get something that by definition has no direction. Telling me that power is a scalar is surely a starting point, not a proof that nothing reactive is there. CT On 15 September 2011 21:46, John Woodgate <jmw@...> wrote: ** [Non-text portions of this message have been removed] |
|
Re: About impedance
--- In LTspice@..., Philip Bellingham <rmhc78a@...> wrote:
All, By definition s-parameters are small signal linear AC parameters, and in most cases the parameters are a function of frequency. Usually, one is interested in only a single frequency or a mall range of frequencies. The work fine for their intended purpose. Rick |
|
Re: Freqeucny Dependent resistor
John Woodgate
In message <CANj54jz_C3C0wTdO5kkCRbVh3qDQsrda9qatksP5h1bX+ohtqA@...>, dated Thu, 15 Sep 2011, Christian Thomas <ct.waveform@...> writes:
But are you sure that radiation resistances are only real?By definition: they are notionally responsible for the real power that is radiated. Reactive elements cannot be responsible. -- 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. |
|
Re: About impedance
Tony,
All valid points. I agree that creating SPICE models for s-parameter characterized passive parts using only native SPICE passives (C, L, and R) is an art and the native models are often inadequate even with the provision for adding parasitics. The same is true for modeling active components. Therefore, I agree that creating SPICE models that can be used for all types of analysis (TRAN, AC, etc.) is a daunting task when the only data you have is s-parameters. My experience has been that frequency domain characterization and simulation is often sufficient. To that end, the utility, in our group files, that converts an s-parameter file to a set frequency dependent controlled sources in SPICE, creates models that are quite acceptable when performing AC network analysis.?This does limit you to AC analysis, and for active devices you are constrained to the bias conditions present?when the s-parameters were measured. Regards, ?? - Philip ________________________________ From: Tony Casey <tony@...> To: LTspice@... Sent: Thu, September 15, 2011 12:42:09 PM Subject: [LTspice] Re: About impedance ? --- In LTspice@..., Andy <Andrew.Ingraham@...> wrote: Indeed. And for passive products, such as SAW filters or isolators, S-parametersNaive question. Why would you want to do that? If you have s-parameterOften you have s-parameters for a component (a transistor or MMIC), are the only data you're ever likely to get from vendors. Also for capacitors and inductors, vendors such as Murata and ATC also provide S-parameter data for use at frequencies at which the simple equivalent circuits we're used to in SPICE are hopelessly inadequate to describe the performance. Regards, Tony [Non-text portions of this message have been removed] |
|
Re: Freqeucny Dependent resistor
John
I put my caveat in for good reason ... The way I thought about it was to think what I would do to design one. I would need a reactive component somewhere, to do it easily, and then I would cancel out that component's phase shift with an all-pass. I am fairly certain that can't be done perfectly (though for the moment I can't quite think why not for a first order original. I suppose because it has to add, so you can never get a perfect time delay.) But are you sure that radiation resistances are only real? I'm fairly certain that acoustic radiation impedances aren't naturally real and that it's only because drivers are mass controlled that they appear so. (I suspect that means masses in the f = ma sense rather than an equivalent C.) Christian On 15 September 2011 20:32, John Woodgate <jmw@...> wrote: ** [Non-text portions of this message have been removed] |
|
Re: Freqeucny Dependent resistor
I believe it is impossible to have a physically realizable resistor that is frequency dependent and has no reactive component. A transmission line comes close in that its resistance is almost constant for a range of frequencies, but as you know at very low frequencies, neglecting any series resistive components, its reactance is largely capacitive.
toggle quoted message
Show quoted text
Hubert ----- Original Message -----
From: Tony Casey To: LTspice@... Sent: Thursday, September 15, 2011 12:43 PM Subject: [LTspice] Re: Freqeucny Dependent resistor <snip> --- In LTspice@..., Christian Thomas <ct.waveform@...> wrote: > > Well, that's a question AG. > > Might we not be looking at a naive question here? Ie. Can I please have a > resistor that changes with frequency but with none of those nasty reactive > elements? If that's the case then looking in the s-plane is not the answer > being sought. > > In which case the answer needed is "No, you can't. Or at least you can't > have a full solution. (I think that must be right). But we do have some > useful reactive components that perform that function, and that's what > everyone else uses. C and L in LTSpice; and their s-plane behaviour is > built in." > > CT </snip> Hello Christian, I'm sure you ask the question tongue-in-cheek, because you surely must be aware of instances where the real part of an impedance changes with frequency without significant change in the reactance. What about the resistance of straight length of wire? This increases due to the skin effect, whereby as the frequency rises more and more of the current travels closer to the outer (indeed for circular cross-section, the only) surface of the wire, so in effect reducing the cross-sectional area of the wire. In the limit, there is also a change in the inductance per unit length too, but it is not significant compared to the change in resistance. And although not strictly a "component", there is the free space acoustic radiation resistance of a diaphragm, which also rises with frequency up to the frequency where the circumference is approximately equal to the wavelength. I will concede in this example that the reactive part of the impedance also changes at a fair rate of knots over the same frequency interval. I'm sure you already knew all that. But it does illustrate why it is perfect legitimate to seek frequency-dependent resistance models. Regards, Tony |
|
Re: Freqeucny Dependent resistor
Ganesan
There is a whole field of active filter synthesis based on FDNR
toggle quoted message
Show quoted text
(Frequency Dependent Negative Resistance).... LTspice makes it easier to synthesize... <> cheers A. Ganesan ================================================================================= On 9/15/2011 2:13 PM, Christian Thomas wrote:
Well, that's a question AG. |
|
Re: Freqeucny Dependent resistor
Tony Casey
<snip>
--- In LTspice@..., Christian Thomas <ct.waveform@...> wrote: </snip> Hello Christian, I'm sure you ask the question tongue-in-cheek, because you surely must be aware of instances where the real part of an impedance changes with frequency without significant change in the reactance. What about the resistance of straight length of wire? This increases due to the skin effect, whereby as the frequency rises more and more of the current travels closer to the outer (indeed for circular cross-section, the only) surface of the wire, so in effect reducing the cross-sectional area of the wire. In the limit, there is also a change in the inductance per unit length too, but it is not significant compared to the change in resistance. And although not strictly a "component", there is the free space acoustic radiation resistance of a diaphragm, which also rises with frequency up to the frequency where the circumference is approximately equal to the wavelength. I will concede in this example that the reactive part of the impedance also changes at a fair rate of knots over the same frequency interval. I'm sure you already knew all that. But it does illustrate why it is perfect legitimate to seek frequency-dependent resistance models. Regards, Tony |
|
Re: Solid-state relay model
I have made some. They work perfect, except when I put them in a circuit, then they behave unbelievably weird. I'm working the issue.
toggle quoted message
Show quoted text
--- In LTspice@..., "supertjhok" <supertjhok@...> wrote:
|
|
Re: Freqeucny Dependent resistor
John Woodgate
In message <CANj54jwDGEMhNQA5E=DD=cMmYJTwDN=mMCrX2FaMfB4YgiLxuw@...>, dated Thu, 15 Sep 2011, Christian Thomas <ct.waveform@...> writes:
In which case the answer needed is "No, you can't. Or at least you can't have a full solution. (I think that must be right).I'm not sure. We have real resistances that change with frequency - radiation resistances for example, There are definitely resistances - at any fixed frequency they are consistent with Ohm's and Joule's Laws. So why can't I define a resistor R(f), such that R(f) = kf, or K*F(f)? k and K have dimensions, of course, but that doesn't seem to be a problem. -- 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. |
|
Re: Freqeucny Dependent resistor
Well, that's a question AG.
Might we not be looking at a naive question here? Ie. Can I please have a resistor that changes with frequency but with none of those nasty reactive elements? If that's the case then looking in the s-plane is not the answer being sought. In which case the answer needed is "No, you can't. Or at least you can't have a full solution. (I think that must be right). But we do have some useful reactive components that perform that function, and that's what everyone else uses. C and L in LTSpice; and their s-plane behaviour is built in." CT On 15 September 2011 19:28, Ganesan <dg1@...> wrote: ** [Non-text portions of this message have been removed] |
|
Re: Freqeucny Dependent resistor
Ganesan
Should Laplace be turning in his grave?
toggle quoted message
Show quoted text
On 9/15/2011 12:51 PM, Michael Peter Kiwanuka wrote:
Hi Shivesh, |
|
Re: Freqeucny Dependent resistor
Hi Shivesh,
I have not a great deal of understanding of Laplace transforms but I have a niggling feeling of the transform of an increasing ramp is (1/s**2). Which definition of L-tranform are you usiing ? Best regards Michael To: LTspice@... From: shivesh_sl@... Date: Wed, 14 Sep 2011 20:01:53 +0000 Subject: [LTspice] Re: Freqeucny Dependent resistor Hi George, Laplace transform for ramp in increasing values is (1/s). Yuo search we for this. Since we have decreasing function hence inverse of previous case, you get (s). since s is complex no. with value of containing (w)=2*PI()*freq hence for R_eq you multiply 2*PI() to compensate for one coming in from (w). I hope this explanation helps. I know it is not great but probably you get gits of it. Regards, Shivesh --- In LTspice@..., "George Evans" <george.evans@...> wrote:
[Non-text portions of this message have been removed] |
to navigate to use esc to dismiss