Installing and using LTspice IV (now including LTXVII), From beginner to advanced

[FdW]

But it is not, it is only different from what one (the modern human) expects.

The F(ahrad) is implicit, so it is always missing (e.g. never included) [the same is true for R(esistance), L, …]

Thus 1p(F), 1n(F), 1(F) etc.

The confusing part is (maybe) that the parser ignores all that is does need and/or understands. This also removes the (F) [implicit or explicit] from the parser output as it is 'extra' [not needed and/or understand]

Thus 1p(F)xxxx<EOL>

This works not universally on everything and everywhere, it seems that the LTspice implementation does not use the same parser for all data.

In this case 1p as expected, the F is implicit, xxxx is any characters up to the end of line, <EOL> is end of line

Also the SPICE parser is ignoring case, thus 'p' is equivalent to 'P' etc.

 

[jan.didden]

OK, I should have said: 'it is inconsistent'. Accepting both 1 and 1V as one volt, but not accepting 1 and 1F as both 1 Farad, is inconsistent.
Accepting 1f as 1 Farad is plain nonsense. How would you then specify 1 femto Farad.

Jan

 

[FdW]

As above (in post #1361) sorry I was a bit slow in patching up the message.

In your example the 'V' is an 'extra' (e.g. not needed and not understood, thus not included in the parser result).

The 'V' is thus NOT accepted

 

[jan.didden]

[snip] sorry I was a bit slow in patching up the message.

It's still early

 

[jan.didden]

Frans, Karl, mystery solved.
The issue is the freq range analyzed versus the period for the waveform.

Someone at the Yahoo LTspice forum suggested this:

The following delay.net file:

~~~
* delay
B1 out 0 V=V(src) laplace=exp(-s*period)
V1 src 0 AC 1
.param period=50us
.ac dec 1000 2k 30k
.end
~~~

... produces a perfect circle when you 
look at V(out) in Nyquist view with linear axes.

But if I increase the freq range for the .ac, or the period in the param statement, the circle progressively becomes a 'ball of wound wool', filling up from outside to inside! See attachments.

Which I believe can be taken care of by making the period a specific fraction of the stop frequency.

The other ways of drawing the circle probably had a similar issue.

Jan

 

[keantoken]

Try increasing the number of steps in the .ac line, you are seeing interpolation between datapoints.

Also you are plotting multiple-turn circles, so the datapoints are a lot farther apart than they need to be.

 

[jan.didden]

Try increasing the number of steps in the .ac line, you are seeing interpolation between datapoints.

Also you are plotting multiple-turn circles, so the datapoints are a lot farther apart than they need to be.

Anthony, that does not seem the case. Less points leads to a circle circumference that starts to show line segments instead of smooth curvature, but does not influence the '# of circles' so to say.
20 points vs. 2000 points.

Jan

 

[keantoken]

Jan, I played with it and figured out how to limit the circle to 1 turn and not waste data points. You have to let the Laplace transform know over what BW it needs to plot the circle, otherwise the circumference will change depending on the AC settings.

EDIT: sorry, that one was bad. Image updated.

To get an even smoother circle, set the AC analysis to linear. With logarithmic frequency the points concentrate towards the high frequencies.

 

[jan.didden]

Excellent!
I guess the 20k in the Laplace is related to the 50uS period.
Can that also be parametrized and made dependent on other parameters?
I mean, period = 1/20k which is the frequency you put in the Laplace. Why 20k?

Edit: if you make that 'test freq' = 1, and the period 1/1 = 1, it also works fine.

So the final then becomes:

V=V(src) laplace=exp(-s*(stopfreq-startfreq)**-1)

Thanks,

Jan

 

[keantoken]

Yes, I just realized that. I thought the period was a Laplace variable. Ah, what will I do with myself. :/

It could be simplified to laplace=(-s*ACstop**-1) if you can tolerate a small gap in the circle.

 

[jan.didden]

Ah, what will I do with myself. :/
.

From where I sit, you'r doing fine

Jan

 

[jan.didden]

So to wrap up so far: the blue curve is the 6dB gain curve. If I understand the theory correctly (not 100% sure yet) this shows that at some frequency the gain margin is minimum at 6dB. That freq being about 1.3MHz.

Thanks guys, this was a very nice exercise, learned some new tricks thanks to you.

Jan

 

[Mooly]

Frans, Karl, mystery solved.

I am reading it all, even if a lot of it is over my paygrade

 

[kees52]

Little strange I get no mail anymore from diyaudio if there are posts. just suddenly I get nothing, now for more then a week.

 

[keantoken]

I've just gotten a flood of maybe 20 notifications. I've also been getting notifications a week late! They have gotten really unreliable.

 

[jan.didden]

Simplest circle so far (courtesy Helmut Sennewald over at the Yahoo LTspice user forum):

Behavioral source (B): V=delay(V(src),1/(stopfreq-startfreq))

The floodgates are open: he gave at least three new ways, in various forms. If you really know LTspice (I don't yet) the sky is the limit!

Jan

 

[jan.didden]

Little strange I get no mail anymore from diyaudio if there are posts. just suddenly I get nothing, now for more then a week.

I am getting the notifications for this thread daily, seems still to work.
But I know Jason is working on a complete overhaul/replacement of the forum so maybe sometimes pieces break and are then fixed.

BTW I probably will visit my sister in Sprang-Capelle this afternoon. I'll wave to you ;-)

Jan

 

[jan.didden]

Another small contribution I found which may be useful:

How to Use a Chip Vendor Op-Amp Model in LTSpice

Jan

 

[FdW]

That would be nice but in all simulators I have seen a number alone always works, like 1e-15 is always a femto-Farad. Differentiation between upper and lower case letters is always problematic especially with software that has a long legacy. Case in point email addresses.

But for email there is a good reason Baudot code - Wikipedia

Note that SPICE may claim the same reason, but it then is not a good reason, it was (most likely) done for easy(er) parsing and compatibility with (then) older terminals (even Baudot teletypes).

Anyway, one days solution is an other days problem. I think SPICE needs a workover

 

[kees52]

I am getting the notifications for this thread daily, seems still to work.
But I know Jason is working on a complete overhaul/replacement of the forum so maybe sometimes pieces break and are then fixed.

BTW I probably will visit my sister in Sprang-Capelle this afternoon. I'll wave to you ;-)

Jan

Your hand was not big enough, I did not see it, or you came not through sprang-capelle, I do ride a recumbent bicycle, ligfiets, en die is inmiddels beroemd en berucht doordat ze me hier zwart maken tot aan pedo aan toe, en zijn veroordeeld, en 500 euro schadevergoeding gehad, maar wordt wel bedreigt zo nu en dan, wel afnemend...Misschien heeft zelfs je zus wel eens wat gehoord over de ligfietser uit schrevelstraat, staan zelfs huizen leeg hier, tja een pedo, Leuk sprang-capelle, maar ben hier wel geboren, ken je Casper uit kaatsheuvel? zijn vader is Ingenieur. So far a little dutch, because it was not technical meaning, sorry for this off tread...

Today I did get again mail, maybe it is because i did not visit here some time, busy with court, myine nabure want the wood heater out of mine house, so I am busy coming time.

I did not now you had a sister here, the world is pretty small.

Someone do now superspice?, I did came buy of that and installed it..

regards