Bifilar Wound Balun Transformer

"John Larkin" <jlarkin@highlandtechnology.com> wrote in message 
news:6tej98pr9faetrfr04ju36v3n8jpirtgj7@4ax.com...
> Spice is all double floats, no? 1e6 isn't a very big number. A current
> sampler is open-loop so won't have convergence problems.

What the numbers are doesn't matter, the solver uses iterative algorithms 
that save inverting the whole N x N matrix.  Matrix operations are O(N^3), 
so a modest circuit with 100 nodes takes about 10^6 more operations than a 
basic battery and lightbulb circuit.  And that's if it guessed the correct 
timestep the first time, which very rarely occurs.  SPICE throws away a 
*lot* of computation.

Obviously, you've never simulated anything of significance, or you'd know 
this.  The degree of these approximations is controlled by the simulation 
parameters, which experienced users understand must be optimized for a 
given model to avoid slowing to a crawl (with default settings, LTSpice 
likes to wander off into the picoseconds, eating glue) or generating 
errors (most often "timestep too small").

Tim

-- 
Deep Friar: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms 

On Tue, 6 Nov 2012 20:24:28 -0600, "Tim Williams"
<tmoranwms@charter.net> wrote:

>"John Larkin" <jlarkin@highlandtechnology.com> wrote in message 
>news:6tej98pr9faetrfr04ju36v3n8jpirtgj7@4ax.com...
>> Spice is all double floats, no? 1e6 isn't a very big number. A current
>> sampler is open-loop so won't have convergence problems.
>
>What the numbers are doesn't matter, the solver uses iterative algorithms 
>that save inverting the whole N x N matrix.  Matrix operations are O(N^3), 
>so a modest circuit with 100 nodes takes about 10^6 more operations than a 
>basic battery and lightbulb circuit.  And that's if it guessed the correct 
>timestep the first time, which very rarely occurs.  SPICE throws away a 
>*lot* of computation.
>
>Obviously, you've never simulated anything of significance,


Only a couple of hundred million dollars worth.

My first simulation, when I was about 20, was of a 32,000 horsepower
steamship propulsion system, all the way from the my control system to
the hull moving through the water. I didn't use Spice.

It worked just fine.


-- 

John Larkin                  Highland Technology Inc
www.highlandtechnology.com   jlarkin at highlandtechnology dot com   

Precision electronic instrumentation
Picosecond-resolution Digital Delay and Pulse generators
Custom timing and laser controllers
Photonics and fiberoptic TTL data links
VME  analog, thermocouple, LVDT, synchro, tachometer
Multichannel arbitrary waveform generators

On Nov 7, 11:14=A0am, Jamie
<jamie_ka1lpa_not_valid_after_ka1l...@charter.net> wrote:
> BillSlomanwrote:
> > On Nov 7, 8:23 am, John Larkin <jlar...@highlandtechnology.com> wrote:
>
> >>On Sun, 04 Nov 2012 19:49:14 +0100, Fred Bartoli <" "> wrote:
>
> >>>John Larkin a crit :
>
> >>>>On Sun, 04 Nov 2012 18:57:27 +0100, Fred Bartoli <" "> wrote:
>
> >>>>>Fred Abse a crit :
>
> >>>>>>On Sat, 03 Nov 2012 18:50:54 -0500, Tim Williams wrote:
>
> >>>>>>>The important thing about transmission line transformers is to for=
get
> >>>>>>>about using them as transformers. =A0Use them as transmission line=
s!
>
> >>>>>>Anybody know how to accurately model a transmission line transforme=
r in
> >>>>>>Spice, taking into account core properties?
>
> >>>>>For a simple one, just as it is:
> >>>>>use a TLine/RLC tline and between the 2 ""shield/ref plane" connecti=
ons
> >>>>>you just tie the magnetizing inductance, with maybe your core model
> >>>>>(losses, non linearities, hysteresis,...)
>
> >>>>What's interesting about the LT Spice transmission lines is that they
> >>>>have no common-mode DC continuity between ends. They act as if there
> >>>>is an ideal 1:1 isolation transformer in the circuit.
>
> >>>>That sort of makes sense, since, say, the outer conductor of a coax
> >>>>has its own complex impedance against the universe, and Spice elects
> >>>>to not model that. Still, it can throw you if you don't know about it=
,
> >>>>and make baluns seem to work much better than they will in real life.
>
> >>>If you want to accurately model a coax cable you need two TLines. One
> >>>modeling the center/shield transmission line, and a second one to mode=
l
> >>>the shield WRT to "space".
>
> >>>>So a txline can make a handy 1:1 ideal transformer. Unlike a 1:1 VCVS=
,
> >>>>it's bidirectional and the output loads the input.
>
> >>>The "standard" perfect transformer is composed of a vcvs to transport
> >>>voltage to the secondary and a CCCS to reflect the secondary current
> >>>back to the primary, and a 0 voltage source to probe it.
> >>>It is much less computationally demanding than the Tline which has to
> >>>maintain history.
>
> >>I guess there's nothing stopping one from making a couple of gigahenry
> >>inductors coupled with K=3D1.
>
> > That's the joy of simulation. You can test ideas that would cost a
> > mint in superconducting wire and liquid helium if you wanted to try
> > them out on the bench.
>
> > The are applications where that kind of expenditure on real parts
> > might be justifiable.
>
> I get the idea that a simulator is the only source for any inclinations
> you come up with..

Congratulations on coming up with an idea. You don't do it often. For
extra credit come up with an idea that makes some kind of sense - you
haven't yet, but while there's life there's hope.

--
Bill Sloman, Sydney

On Nov 7, 12:42=A0pm, John Larkin <jlar...@highlandtechnology.com>
wrote:
> On Wed, 07 Nov 2012 02:30:01 +0100, Fred Bartoli <" "> wrote:
>
>
>
>
>
>
>
>
>
> >John Larkin a crit :
> >> On Tue, 6 Nov 2012 15:29:35 -0800 (PST),BillSloman
> >> <bill.slo...@ieee.org> wrote:
>
> >>> On Nov 7, 8:23 am, John Larkin <jlar...@highlandtechnology.com> wrote=
:
> >>>> On Sun, 04 Nov 2012 19:49:14 +0100, Fred Bartoli <" "> wrote:
>
> >>>>> John Larkin a crit :
> >>>>>> On Sun, 04 Nov 2012 18:57:27 +0100, Fred Bartoli <" "> wrote:
> >>>>>>> Fred Abse a crit :
> >>>>>>>> On Sat, 03 Nov 2012 18:50:54 -0500, Tim Williams wrote:
> >>>>>>>>> The important thing about transmission line transformers is to =
forget
> >>>>>>>>> about using them as transformers. =A0Use them as transmission l=
ines!
> >>>>>>>> Anybody know how to accurately model a transmission line transfo=
rmer in
> >>>>>>>> Spice, taking into account core properties?
> >>>>>>> For a simple one, just as it is:
> >>>>>>> use a TLine/RLC tline and between the 2 ""shield/ref plane" conne=
ctions
> >>>>>>> you just tie the magnetizing inductance, with maybe your core mod=
el
> >>>>>>> (losses, non linearities, hysteresis,...)
> >>>>>> What's interesting about the LT Spice transmission lines is that t=
hey
> >>>>>> have no common-mode DC continuity between ends. They act as if the=
re
> >>>>>> is an ideal 1:1 isolation transformer in the circuit.
> >>>>>> That sort of makes sense, since, say, the outer conductor of a coa=
x
> >>>>>> has its own complex impedance against the universe, and Spice elec=
ts
> >>>>>> to not model that. Still, it can throw you if you don't know about=
 it,
> >>>>>> and make baluns seem to work much better than they will in real li=
fe.
> >>>>> If you want to accurately model a coax cable you need two TLines. O=
ne
> >>>>> modeling the center/shield transmission line, and a second one to m=
odel
> >>>>> the shield WRT to "space".
> >>>>>> So a txline can make a handy 1:1 ideal transformer. Unlike a 1:1 V=
CVS,
> >>>>>> it's bidirectional and the output loads the input.
> >>>>> The "standard" perfect transformer is composed of a vcvs to transpo=
rt
> >>>>> voltage to the secondary and a CCCS to reflect the secondary curren=
t
> >>>>> back to the primary, and a 0 voltage source to probe it.
> >>>>> It is much less computationally demanding than the Tline which has =
to
> >>>>> maintain history.
> >>>> I guess there's nothing stopping one from making a couple of gigahen=
ry
> >>>> inductors coupled with K=3D1.
> >>> That's the joy of simulation. You can test ideas that would cost a
> >>> mint in superconducting wire and liquid helium if you wanted to try
> >>> them out on the bench.
>
> >>> The are applications where that kind of expenditure on real parts
> >>> might be justifiable.
>
> >> It's cool to stick a micro-ohm resistor somewhere in a circuit to
> >> sample current, and follow that with a 1e6 gain amp with infinite
> >> CMRR.
>
> >> I like to build analog error computers into my circuits too, so I can
> >> graph error or goodness on the same plot as actual signals.
>
> >> I do tend to worry about power dissipation and real-world stuff in
> >> sims, where it just doesn't matter. A couple of 1 ohm resistors make a
> >> good divider, but I worry about the current, so I use 1Ks.
>
> >On complex designs (well simulations) it's better to worry about matrix
> >conditioning and in this regards a 0V voltage source and a CCVS in lieu
> >of your 1 ohm and 1E6 gain VCVS is much better.
> >I have some applications where convergence is sometimes hard to obtain
> >and I sure wouldn't want to make the situation worse than it needs to be=
.
>
> Spice is all double floats, no? 1e6 isn't a very big number. A current
> sampler is open-loop so won't have convergence problems.
>
> LT Spice sometimes does weird things with time steps, especially if a
> circuit has radically different time constants here and there, or
> mysterious stuff inside a part model. Real transistors seem to stress
> it, too. Somebody should write a book or article about all that.

It's been done. I think I referenced a text on the subject in my Ph.D.
thesis back in 1969.

--
Bill Sloman, Sydney