Still waiting for a Ltspice expert answer.

[Jamie]

Does the wires assume real life situations to some
degree or do we need to add stray inductance, Resistors
and caps to act like a board?

I would of thought it would assume an average lay out of one
persons worse nightmare and the actual real live circuit would
just work better, seeing that this designer would engineer proper
protocols on the board.

It just seems that I don't see many of the small defects the
sim may show, when it is on a real board assembled using common
practices.

Which leads me to my question, does the sim assume a worse case
scenario or maybe the sim just over exaggerates now and then?

This would be LTspice I am referring to. I was thinking of looking at
Pspice to see how that goes.

Jamie

 

[Tim Williams]

I've heard of a few simulators which assumed the designer indicated wire
resistance proportional to the length of the traces on the schematic
drawing. Which often have no relation to actual board traces, and
certainly have no relation to their widths.

As far as I know, most professional environments (SPICE based) assume all
nets are ideal. If the models themselves include parasitics, they will
necessarily be limited to package parasitics. It's always the user's
responsibility to include realistic wiring parasitics.

I don't remember if LTSpice includes package parasitics. I seem to recall
they mostly use MODELs rather than SUBCKTs, but they may've expanded the
MODEL definition to include simple series or parallel elements.

In short, no, SPICE tends to assume ideal conditions. It's a model, not a
reality simulator, and it's your duty to construct a model that has any
useful representation at all.

If you insist on reality simulation, there are suites out there which
attempt to do such things. Ansoft makes such a system, including
electromagnetic (which can do static, AC and transient 3D simulations on
real board layouts, and generate approximate SPICE models for the
circuit), mechanical (stress/strain) and thermal (including fluid flow,
expansion, tempcos to feed back into the circuit). A few others make
similar suites; all are horrifically expensive, only something worthwhile
if you're either making millions of top-quality, moderately complex
products (cars maybe??), or a few very, VERY critical systems (avionics,
aircraft, spacecraft?).

Tim

 

[Jamie]

Do the math... by hand... then use the sim to VERIFY!

...Jim Thompson

Well that just sounds counter productive.

Why waste my time using a sim if I am going to do it by hand in the
first place?

You can haggle all you want Jim, It does not work on me..

I am not one of your goal post. You have JL for that

Jamie

 

[qrk]

Well that just sounds counter productive.

Why waste my time using a sim if I am going to do it by hand in the
first place?

You can haggle all you want Jim, It does not work on me..

I am not one of your goal post. You have JL for that

Jamie

If you don't understand your circuit maths, how do you know if the
simulation is close to reality???? Many times, you can do this in your
head, but sometimes you need to write down a bunch of maths to
understand issues you may run in to. As an excercise, try to
temperature stabilize a simple one transistor amplifier without
understanding the basic transistor equations.

Do you blindly trust your models??? If you do, beware, they are often
wrong. With experience, you will learn a healthy distrust in your
simulator and test equipment.

 

[Jamie]

One persons' nightmare is not "average." WTH?

No, what you suggest -- "average" is whackadoodle. There is no
"average" layout.




There may be standards and usual practice for a variety of tasks --
the more simple the more likely it can be handled by standards.
"Simple" is limited. The harder the job, the more diverse opinion
you'll get regarding what is a proper approach.




"Worst case" and "one persons' nightmare" doesn't mean "average" or
"common." If your wrong views were more coherent, then they would be
easier to pick apart!

You have 1 pF of pad capacitance, and you think it may matter? then
put it in
You care about 0.7 nH of via inductance? then put it in
You care about the inductance of a long trace? then put it in

The models are just *models* -- not the real thing. How good are
they? After all, that is all spice has to work with. Models are
never perfect -- what should spice assume as "average" about them?
Caveat emptor: you need to have some idea of what you're doing in the
first place.

I think your question is really: Can I get SPICE to think for me? No,
you can't.

Well no problem, I'll just continue with how I've been doing it all
these years. Calculator,years of foot notes, engin books, equipment use,
CAD steps and then boards.

I guess spice is good for concept circuit experimenting. Saves on
paper.

I am experimenting with a little tool app I am writing that monitors
the clipboard of a BMP image arriving into it. This tool has FTP code
where I can then hit the default button and send it some where with some
local index to manage them. So far it looks ok. I was thinking to use it
as an uplink tool to send schematics to your web site for the general
public audience instead of dumping spice in the chat areas.

Was thinking of supporting the WMF format too, it would save on
network traffic.


Jamie

 

[Jamie]

I don't think Jamie is even a technician, but he _is_ successfully
becoming a reductive emulation of John Larkin ;-)

...Jim Thompson

Yeah, you are correct as usually Jim. I mean, are you ever wrong?

Yes, I am just a back woods inbreed hill billy. Most of the time I
run around with no clothes on, and when I do have them on, they are
hand me downs, full of holes and have not seen what would represent
laundry detergent or anything close to that. Yes, we can't even afford
making our own soaps out of lard from road kill. etc..

I use a chinese abacus because I have found that I can actually
exceed the math skills that requires more than 10 fingers. Yes, I still
have all my fingers. I have not yet lost any of them while fighting with
my sisters, in the sack.

Nothing like having a hole down, especially when its your lady or
boyfriend, what ever your party maybe.

Jamie

 

[Jamie]

If you don't understand your circuit maths, how do you know if the
simulation is close to reality???? Many times, you can do this in your
head, but sometimes you need to write down a bunch of maths to
understand issues you may run in to. As an excercise, try to
temperature stabilize a simple one transistor amplifier without
understanding the basic transistor equations.

Do you blindly trust your models??? If you do, beware, they are often
wrong. With experience, you will learn a healthy distrust in your
simulator and test equipment.

Fucking beginners

What a gas!

Jamie

 

[John S]

Yeah, you are correct as usually Jim. I mean, are you ever wrong?

Yes, I am just a back woods inbreed hill billy. Most of the time I
run around with no clothes on, and when I do have them on, they are
hand me downs, full of holes and have not seen what would represent
laundry detergent or anything close to that. Yes, we can't even afford
making our own soaps out of lard from road kill. etc..

I use a chinese abacus because I have found that I can actually exceed
the math skills that requires more than 10 fingers. Yes, I still
have all my fingers. I have not yet lost any of them while fighting with
my sisters, in the sack.

Nothing like having a hole down, especially when its your lady or
boyfriend, what ever your party maybe.

Jamie

So your ladies and boyfriends have their holes down? I guess that would
promote the inbreeding. Try to resist.

 

[Jamie]

Don't forget to add, "delay" All these simulators keep forgetting
about that pesky quirk of reality, "It takes time for a signal to go
from here to there." And, oh, yes, don't forget to add that even
peskier term, "radiation loss" That way when you simulate charging a
cap from another cap, you can reconcile the apparent loss of energy.

Yes, the sim is good for concept sketching but not very good for the
refinements. I find the bench model+Test gear tells the story.

Like I've said before, I never used a Sim until the last few months
and only here basically for general chatter.

Jamie

 

[John S]

Does the wires assume real life situations to some
degree or do we need to add stray inductance, Resistors
and caps to act like a board?

I would of thought it would assume an average lay out of one
persons worse nightmare and the actual real live circuit would
just work better, seeing that this designer would engineer proper
protocols on the board.

It just seems that I don't see many of the small defects the
sim may show, when it is on a real board assembled using common
practices.

Which leads me to my question, does the sim assume a worse case
scenario or maybe the sim just over exaggerates now and then?

This would be LTspice I am referring to. I was thinking of looking at
Pspice to see how that goes.

Jamie

I believe that this is one of the stupidest questions I've ever seen on
this group. Got any more?

 

[[email protected]]

Wires in LTSpice (and PSpice, as far as I know), are made of Magic
Stuff. A node is a node is a node (in Spice-land) and has no voltage
variations across it.

That's true (with LTSpice) but all connections on a wire are not equal. You
can measure the current in a wire at any place along the wire. LTSpice looks
left and right from that point to get the current at that point. The topology
matters.

Jim's right, although I would state it differently: the simulation is
help for those who already know what's up; try something that the
simulation isn't prepared to give a correct answer for, and you'll get a
wrong answer.

Like transmission line speed > C. ;-)

It _is_ handy for complex, nonlinear circuits (like switchers), but you
still have to pose questions that it can answer sensibly.

It's also handy for statistical or worst-case analysis on complex designs.

 

[Jamie]

What a sick puppy, but I'm not surprised, since you seem to have
chosen to follow Larkin as your mentor.

...Jim Thompson

Actually, I think Larkin is a smart Dude! I take my hat to him for his
offerings he has made here, despite all the interference.

Jamie

 

[[email protected]]

Well, that's what I hoped he'd get -- wires in LTSpice are made of magic
stuff, wires in this humdrum mundane world we live in are made of real
stuff.

If you want to simulate the real stuff with the magic stuff, you need to
add detail to your simulation.


True, although even there you can often go much farther with pencil &
paper methods than most folks go.

Perhaps, but doing statistics on 100,000 cases isn't much fun by hand. ;-)

 

[Fred Abse]

And, oh, yes, don't forget to add that even
peskier term, "radiation loss" That way when you simulate charging a
cap from another cap, you can reconcile the apparent loss of energy.

Please, pretty please, don' open that can of worms again. It was talked to
death some time ago ;-)

 

[Fred Abse]

The model is the RLCG model for transmission lines.

R = resistance
L = inductance
C = capacitance
G = conductance

The R and L are in series with the line. The c and G are in parallel.
These parameters are on a per-meter basis, multiply by the length dz to
obtain the total element value in ohms, henries, farads and siemens.

Most, if not all Spice do not fully implement an RLCG model. The lossy
line model, LTRA, only implements R, L, and C. Nonzero G not supported
(yet).

The parameters are on a *per-unit* basis, not per-meter, allowing the use
of feet, inches, millimeters, cubits, parsecs, whatever.

Examples:

..model RG223 LTRA (LEN=1 R=11.3m L=0.077u C=30.8p)

..model RG59 LTRA (LEN=1 R=51.6m L=0.115u C=20.5p)

Units in this case are per foot.

 

[Fred Abse]

PSpice has a full RLCG model... "TLOSSY" in the parts index.

Didn't know that. LTspice *appears* to have, then you try to use G, and it
tells you to go away.

It appears to be code lifted straight out of Berkeley Spice 3.

 

[Jamie]

Jamie/Maynard/WTF is just one banana peel short of insanity... and yet
he's smug about it.

He has no clue of the technical competency of 'qrk' who properly
advised, "Do you blindly trust your models??? If you do, beware, they
are often wrong. With experience, you will learn a healthy distrust in
your simulator and test equipment."

I happen to know 'qrk' quite well. His company was a past chip design
customer of mine. 'qrk' really knows his stuff.

Yet Jamie/Maynard/WTF smart-mouths his superiors. Time to shun him
into non-existence.

...Jim Thompson

I am puzzled by you Jim, I really thought you had more common sense
than that.

If seems your above statement is completely contradictory to what I
stated by Ltspice not being reliable, spice in general i guess and bench
equipment is the only truth.

Oh well, I'll just past it off as age on your part. I have an aging
father in-law and I see the resemblance.

Just goes to show how some people are not able to make proper
assessments of others. Most of those generally do not become management.

Jamie