Timeline for How do you simulate voltage noise with LTSpice?
Current License: CC BY-SA 3.0
15 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jan 23, 2013 at 8:50 | comment | added | Oli Glaser | @rawbrawb - yes you're right, it's only modeled in the frequency domain. | |
| Jan 23, 2013 at 8:49 | comment | added | Oli Glaser | @ThePhoton - yes, it would slow things down but it would be nice to have an option to include it, which would be left off normally. The various transistor models do have (optional) noise parameters, so it would be possible to rig something up to make use of these. I may have a closer look when I have time. | |
| Jan 17, 2013 at 4:41 | history | edited | placeholder | CC BY-SA 3.0 |
added 4 characters in body
|
| Jan 17, 2013 at 4:31 | comment | added | placeholder | @OliGlaser .noise requires .ac, from documentation it shows how to make a "noiseless" resistor for certain analysis. To quote "There is no way to shutoff the noise model for the resistor, but sometimes you may want an element ..." this applies to HSPice, PSpice , Spectre. | |
| Jan 17, 2013 at 3:42 | comment | added | The Photon | @OliGlaser, "the basic models do not include noise modelling for e.g. a transient sim". My understanding is, it's not that the device models don't contain enough information, but that the transient simulation doesn't attempt to model noise. A good reason for this is that adding white noise sources would likely drive the transient simulator to use minimal time steps at all times and dramatically slow down the simulation. | |
| Jan 16, 2013 at 22:46 | comment | added | Oli Glaser | Having said all that, I'd love to be proved wrong here, I would find having inbuilt noise transient simulations in a tool very handy (as I'm sure others would) | |
| Jan 16, 2013 at 22:44 | comment | added | Oli Glaser | @rawbrawb - I wasn't suggesting it was a noise source, it's just a reference point as you say - my point was that this mode the only noise modelling I am aware of. I agree it would be nice if the components did have noise modelling built in for the other sims, but I haven't seen any real evidence of it yet in standard SPICE variants. I think they do use them in pretty much the same way, just some have capabilities to use later models of certain components, of e.g. MOSFETS (LTSPice support 7 different types of monolithic model and one vertical model) whereas other may not. | |
| Jan 16, 2013 at 22:33 | comment | added | placeholder | @OliGlaser in the .noise model the V(<out> etc. is NOT a noise source it (according to manuals) is " the total output noise volatge; it may be a single node in which case it is referenced to ground ... This source is not a noise generator; it is a reference for describing the input ..." It is so you can describe what input your noise is refereed to. All spice models SHOULD have the noise modeling built in. While the models may be the same not all variants of SPICE use them in the same way. In noise it may be the case, but I don't know so I'm not making any claims. | |
| Jan 16, 2013 at 21:27 | comment | added | Oli Glaser | @Photon and rawbrawb - since the models for various SPICEs are generally compatible with each other, it would suggest that the issue would lie with the models rather than the SPICE variant. As far as I am aware, the basic models do not include noise modelling for e.g. a transient sim, but will all work with the dedicated noise simulation. For instance if you try the last simulation in my answer as a transient with no input voltage, you will get 0V out (as opposed to the 20-44nV predicted) | |
| Jan 16, 2013 at 19:35 | comment | added | placeholder | @ThePhoton It's called honesty, I don't have LTSPICE running and I'm not about to presume things. You feeding back info on your instance is ideal. Me showing results from my full blown EDA tools may or may not be useful. Every SPICE variant does things slightly differently. | |
| Jan 16, 2013 at 18:02 | comment | added | The Photon | Per the Help file, "This is a frequency domain analysis that computes the noise due to Johnson, shot and flicker noise." So if you have a schematic of your circuit, you're good. But if, for example, you're creating an op-amp model based on datasheet parameters and/or measurements, you often use idealized components like controlled sources, and you want to have a noise source element to fix up the noise characteristics. | |
| Jan 16, 2013 at 17:26 | comment | added | placeholder | @ThePhoton all devices in spice SHOULD be noisy, i.e. not ideal - noise less. So that should already be in the models. I suspect LTSpice should also have that. | |
| Jan 16, 2013 at 17:23 | history | edited | placeholder | CC BY-SA 3.0 |
added 110 characters in body
|
| Jan 16, 2013 at 17:08 | comment | added | The Photon | Just checked. LTSpice does have a noise analsysis available. But I can't find a noise source in the parts catalog --- hopefully someone can come along and say where to find it. | |
| Jan 16, 2013 at 17:03 | history | answered | placeholder | CC BY-SA 3.0 |