Spice has numerical noise that often starts oscillators. Here is a Yahoo LTspice post from Helmut that describes it a bit more:
Hello Dieter,
Yes. SPICE doesn't simulate noise in a .tran analysis. It's often only the numerical "noise" in SPICE which starts the oscillation of crystal oscillators.
Maybe you want read a little bit about simulation of crystal oscillators.
There are two ways to get to a negative number. One takes you past the "infinite" value. In truth though. we have a 2D world we can avoid hitting the singular value by going a little reactive.
I use MacroExpress to handle repetitive exercises like that.
...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC\'s and Discrete Systems | manus |
| Phoenix, Arizona 85048 Skype: Contacts Only | |
| Voice:(480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |
I love to cook with wine Sometimes I even put it in the food
It would be hard to make it work with AC or ripple superimposed on a DC voltage. Which one do you want it to refer to?
However, it should be possible to use the arrow to see which way the resistor is oriented, then figure out if positive polarity in the waveform means current is flowing towards the positive end. If it's not the way you want, just flip it around.
ISTR Microcap lets you display the orientation of the components. But that clutters the schematic real fast. The LTspice method leaves a much cleaner schematic and probably ends up giving the same information.
This is a common complaint. There are a number of help documents available, each one has it's own strengths and weaknesses. There is a tremendous amount of information to convey, and finding it can be a major problem. But eventually everything gets sorted out. I started long ago with Intusoft and Microcap on DOS, and went through the trauma of converting everything many times.
LTspice was a bit more difficult than usual, but now that I'm used to it I think it has the fastest schematic entry and the least hassle with convergence of any spice program I've seen. And it certainly gets updated often. If someone mentions a problem on the Yahoo forum, it's usually fixed the same day. None of the other spices I've used can say that. I think the program is invaluable, to the point where you really can't do anything meaningful without it. Mike
I mostly analyze XOs in the frequency/phase domain. I don't have the months to wait for a time domain startup to settle, and there's not a lot of useful information after it does.
Some people are concerned on theological grounds any time the subject goes to that of the infinite. They have near fits when you start comparing the sizes of infinities and their heads explode if the question of whether there is an infinite number of infinitudes. In this case we merely have a 1/0 sort of infinity or a simple pole. This is not to suggest that all poles are simple. I work with one who does quantum physics, but that is way off the topic of this question.
That's the same thing you said the last time. Unfortunately, it is wrong.
You did not read the article.
The most important parameter in a crystal oscillator is the power level.
If it is too high, you can fracture the crystal, or cause the frequency to drift. If it is too low, the crystal may not start.
You cannot set the power level using Frequency analysis. You can only set it in Transient analysis.
The settling time in TA was a chronic problem. It simply took too long.
I solved that problem for you. It now takes only a dozen cycles or so, which is thousands of times faster than the method you were using.
You start by determining the current needed to set the power level to the correct value for the crystal. I give the equations and a solver you can use, or you can use any of the others available on the web.
You initialize the current through the motional inductance to the desired value by setting the voltage across the equivalent series resistance. This is in series with the motional inductance, so the current is the same.
Other attempts to set the current through the inductor failed. The current often rose to giga-amp or tera-amp levels at the start of the analysis. The resistor method solves that problem.
You can now set the maximum time step to give a high resolution, perhaps
0.1% of a cycle or less. You could never do this before since the analysis would be too slow, and you would probably run out of memory or disk space before the analysis was finished.
When the analysis starts, it starts at the point in the cycle where the voltage across the inductor is zero and the current is the level you have set.
There is an initial transient that dies in a cycle or so. After that, you only need a dozen cycles or so instead of tens or hundreds of thousands to get the oscillator to settle so you can measure the operation.
Using the Microcap peak search function, you can quickly see if the amplitude is increasing or decreasing, to parts-per-million resolution. Then adjust the feedback and current into the oscillator as needed to stabilize the amplitude at the desired level.
This also gives you the oscillation frequency to parts-per-million resolution, which is much better than any of the component values. You now can check to see if the trim capacitor has enough range to adjust the frequency to the target value.
This method gives the results you need thousands of times faster than the older method where you had to wait for the oscillator to settle.
Again, you cannot do this in Frequency Analysis. And it is very difficult to measure the power level in the actual circuit using a current probe. The current may be too low to get an accurate reading, or the strays may contribute too much error to be useful. Or the current probe may add enough series impedance to change the power level.
So any oscillator you have shipped using Frequency analysis is at risk, since you have no idea what the power level really is.
But now you can fix the problem. Isn't that great?
ltspice netlists start with a line that says "Version #" (with a number instad of #) and end with a blank line, if there's a 'hook' for extracting binaries, an ltspice netlist handler could be inserted before the existing process.
it'd take me an hour or 2 do do it for slrn (the news readrer I use) probably not worth it, who else is using slrn?
when you point at it you see the clamp probe cursor and the arrow showung the positive current direction (the direction that would give a positive trace)
All devices have each terminal pre-defined as a current input or output, all 2 terminal devices have exactly one of each. (the other devices seem to have all terminals as inputs.)
reversing the device in the circuit will give a trace of opposite polarity asif you had instead reversed the remainder of the circuit.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.