SPICE tarnsient analysis of oscillators - sampling time

If i understand the fundamental problem correctly, y our having an issue simulating at realistic freq. with LTspice? Well let me say this first, your p rogram isn't the only front end with that issue. well - i'd recommend looking at either another program for that specific metric(might as well see if it's unanimous or not too) or adjusting your design. Have you reduced or eliminated anything s uperfluous to this signal path? or the other way can you encapsulate this and use a PLL to eit her maintain freq under load, OR and this still makes my head spacy - you can use a PLL to r un your sim as you were/had it - and clean up / or reduce proportionally the magnitude of your results.

i've seen a lot of strange things happening behind the scenes for a couple programs...I wrote one of them! hahahh...but i wrote it for me and my very specific use. LTspice is no lone wolf...Do you work with Labview? Cadence? Matlab? they're m ore or less the best options im aware of....

sorry i couldnt be more specific - precise operati ons have too many variables

Jasen Betts

Jasen Betts

Lark> >>>>>

Jasen Betts

ev> >>>>>>>>

the frequency error is unmeasurable

Let it run for 1,000 cycles. After

overlap exactly. The frequency

the time step down to get a useful

ulation times silly. There is

be set. When precision matters, I

e V source. Run that for 5 seconds

chunky line segments. Change the

lot better.

e using LTSpice with compression on.

Its a solution looking for a problem.

above, is that the LT Spice

sually works fine for things like

tants are nasty.

en I specify a pulse generator

driving a slow RC or something.

ckets inside a transmission

an unlimited amount of

d signal at the other end of the

. it doesn't need separate storage for

is only saved at time-step

ll along the line, so can't

wonder how many effective

whether the time step matters

nd poke a pulse into one

s of reflections, maybe of

forth. There's no "black

lot of internal nodes.

not conducted out of the

of that end node as an

replayed back to the source

lates circuits; it doesn't

appears at the second end

hat doesn't flow into the

the second end and

hm

. It needs internal nodes.

hash of it.

It may use an integral equation for that bit. Chopping the TL up into short sections isn't an efficient solving method.

Cheers

Phil Hobbs

I do that with real coax a fair amount. It's good for generating odd pulse trains and so on. with quite decent timing.

Cheers

Phil Hobbs