| View previous topic :: View next topic |
| Author |
Message |
Mike Engelhardt
Guest
|
 Posted: Sun Aug 03, 2003 6:11 am Post subject: Re: LTSpice--Why not "perfect" results here? |
 |
|
Your 2nd pulse function gives the warning:
WARING: Specified period is not longer than the sum of Trise,
Tfall, and Ton for v1. Increasing period to 1.01e-005.
This warning can be viewed with the command View=>SPICE
error log.
You have to make sure that the period is longer than what it
says, otherwise it will extend it 1% longer than the minimum.
This comes about as a consequence that LTspice allows pulse
functions with zero on time.
--Mike
"Chris Carlen" <crobc@BOGUSFIELD.earthlink.net> wrote in message
news:bghjd306qa@enews3.newsguy.com...
[quote]Mike Engelhardt wrote:
The duty cycle isn>t exactly 50%. When I change the .tran command to
".tran 0 2.0m 1m 10n" then V(n001) and V(vrc) average to 2.525V and
2.5248V respectively. These are pretty close to the same. Note you
can compute the average/rms of a trace by holding down the control
key and left mouse clicking on the trace label. You might want to
adjust your PWL waveform.
--Mike
Mike, there seems to be some problems with the Pulse source.
If I use this:
Pulse(0 1 0 8u 2u 0 10u)
then I see a triangle wave with a period of exactly 10u like it should
be. All the lower peaks land exactly on 10u increments.
But if I use:
Pulse(0 1 0 9u 1u 0 10u)
then there is a little extra 100n or so delay each cycle before it
starts rising again.
This should definitely not happen. I think it is why the sawtooth
averaged less than it should, causing my PWM to average too much.
What can you say about this?
It seems anytime the fall time is specified very short, then this little
extra delay gets in there making the period *not* what was specified.
Note that there is carefully no discrepancy between the period and sum
of rise+Ton+fall times in these examples.
I>m using 2.04o version, BTW.
Ah-ha! Mike, you>ve got some numeric input bug:
Pulse(0 1 0 9u 1u 0 10u) broken
Pulse(0 1 0 9e-6 1e-6 0 10e-6) broken
Pulse(0 1 0 0.000009 0.000001 0 0.00001) WORKS!
Simulate them for 0.5 to 1ms and watch the horror.
Happy fixin' !
Thanks for assistance.
Good day!
--
_____________________
Christopher R. Carlen
crobc@earthlink.net
Suse 8.1 Linux 2.4.19
[/quote] |
|
| |
|
 Back to top |
David Harmon
Guest
|
| Posted: Sun Aug 03, 2003 2:20 pm Post subject: Re: What>s the difference between signals and nets? |
|
|
On Sat, 2 Aug 2003 23:28:06 +0100 in sci.electronics.cad,
"Kevin Aylward" <kevin@anasoft.co.uk> wrote:
[quote]nws wrote:
In the Eagle system, we have nets and we have signals, what
[/quote]
[chop]
[quote]Indeed I do. I am discussing general terms on schematic captures, I was
not referring specifically to idiosyncrasies of Eagle. It was not
apparent that your issue was Eagle specific,
[/quote]
Oh come on, what do you think that "In the Eagle system" as written
by the original poster and quoted in your reply means then? It
means that the whole thread is Eagle specific. It means that your
reply is completely worthless, in the context of the Eagle system.
The qualification "In the Eagle system" could not possibly have been
more clear. You failed to read the post you were responding to.
[quote]If you know what the asinine way that Eagle defines their own
terminology, why did you post in the first place?
[/quote]
I didn>t post "in the first place". Again you failed to read the
post you are responding to. I was answering the question posed by
the original poster, "nws". |
|
| |
|
| Back to top |
Helmut Sennewald
Guest
|
| Posted: Sun Aug 03, 2003 3:15 pm Post subject: Re: LTSpice--Why not "perfect" results here? |
|
|
"Mike Engelhardt" <pmte@concentric.net> schrieb im Newsbeitrag
news:bghnf8$2gp@dispatch.concentric.net...
[quote]Your 2nd pulse function gives the warning:
WARING: Specified period is not longer than the sum of Trise,
Tfall, and Ton for v1. Increasing period to 1.01e-005.
This warning can be viewed with the command View=>SPICE
error log.
You have to make sure that the period is longer than what it
says, otherwise it will extend it 1% longer than the minimum.
[/quote]
Mike,
this 1% isn>t what a designer would expect in his wildest dreams.
[quote]This comes about as a consequence that LTspice allows pulse
functions with zero on time.
[/quote]
When I use the correct command PULSE(0 1 0 9.98u 20n 0n 10u),
I still have got the error message and the wrong period.
The only way to overcome this problem is something like
PULSE(0 1 0 9.98u 19.999n 0n 10u) .
It is now obvious that it is a simple problem when calculating with
floating point numbers. Floating pointer numbers are only exact
for integers and the numbers which exavtly fit the value of
mantissa*2^exponent.
So I have to use one very odd number in the PULSE command to avoid
this error. I recommend to use some correction(rounding) in LTSPICE to
overcome this problem.
I assume the chek is something like this:
// t_r(rise), t_f(fall), t_on(width), t_p(period)
May be your code look like this:
if ( (t_r + t_f + t_on) <= t_p )
{
ok;
}
else
{
fprintf(Error message);
t_on = 0.01 * (t_r + t_f); // The bad assumption
t_p = t_r + t_f + t_on;
}
Mike, here is a ready to use C-program to correct your pulse source check.
Hope that helps. It corrects as expected from an engineer.
Best Regards
Helmut
// Ltspice_pulse.cpp : Defines the entry point for the console application.
//
// LTSPICE PULSE source check
//
// Algorithm: Adjust first t_w until 0 if t_r + t_f_+ t_w > t_p .
// If that is not sufficient, then adjust t_r and t_f until
// t_p can be met. Never chnage t_p.
// If no t_p is defined, then set t_p = t_r + t_f_+ t_w .
//
// ___________ __________
// / \ /
// / \ /
// / \ /
// / \ /
// / \ /
// __________/ \________/
//
// | t_r |<-- t_w ->| t_f | |
// | |
// |<------------ t_p ----------->|
//
#include "stdafx.h"
int main(int argc, char* argv[])
{
double t_r, t_f, t_w, t_p, t_w_max;
t_r = 9.98e-6;
t_f = 15e-9;
t_w = 20e-9;
t_p = 10e-6;
// The algorithm starts here !
if (t_r == 0) // Some other param checks
first.
t_r = 1e-12; // LTSPICE uses 5% of t_p
here.
if (t_f == 0) // Most users didn>t
understand this 5%.
t_f = 1e-12;
if (t_p == 0)
t_p = t_r + t_f + t_w;
if ( (t_r + t_f + t_w) <= t_p )
printf("ok\n");
else
{
t_w_max = t_p - (t_r + t_f); // Max. possible t_w time
if ( (t_w_max - t_w) / t_p < -1e-6 ) // Use a limit for the error
message
printf("Possibly bigger correction\n"); // to avoid the need of x99999
numbers.
else
printf("Possibly smaller correction\n");
if (t_w_max >= 0) // t_r + t_f <= t_p
{
if (t_w > t_w_max) // Adjust t_w to t_w_max
t_w = t_w_max;
}
else // t_r + t_f > t_p
{
t_w = 0; // Set t_w to 0
t_r = t_r + t_r/(t_r + t_f) * t_w_max;
t_f = t_f + t_f/(t_r + t_f) * t_w_max;
}
}
printf("t_r t_f t_w t_p t_w_max\n %e %e %e %e %e\n",
t_r, t_f, t_w, t_p, t_w_max);
return 0;
} |
|
| |
|
| Back to top |
Kevin Aylward
Guest
|
| Posted: Sun Aug 03, 2003 3:16 pm Post subject: Re: What>s the difference between signals and nets? |
|
|
David Harmon wrote:
[quote]On Sat, 2 Aug 2003 23:28:06 +0100 in sci.electronics.cad,
"Kevin Aylward" <kevin@anasoft.co.uk> wrote:
nws wrote:
In the Eagle system, we have nets and we have signals, what
[chop]
Indeed I do. I am discussing general terms on schematic captures, I
was not referring specifically to idiosyncrasies of Eagle. It was not
apparent that your issue was Eagle specific,
Oh come on, what do you think that "In the Eagle system" as written
by the original poster and quoted in your reply means then? It
means that the whole thread is Eagle specific.
[/quote]
It means the the poster *believes* that the question is specific to a
product.
[quote]It means that your
reply is completely worthless, in the context of the Eagle system.
The qualification "In the Eagle system" could not possibly have been
more clear.
[/quote]
Not relevant. It is very, very common for people to ask questions,
believing that it is specific to some particular product, but in fact,
has nothing particular to do with that product. One only has to note how
often people request PSpice information, when in fact, its generic Spice
information they are after. Go and read some posts, and look at the
replies.
[quote]You failed to read the post you were responding to.
[/quote]
Nope. I took an assumption that the poster might be looking for generic
information. Any fool, can read "Eagle". One tries to read between the
lines when people post. Its very rare that what people request is
actually what they want.
[quote]
If you know what the asinine way that Eagle defines their own
terminology, why did you post in the first place?
I didn>t post "in the first place". Again you failed to read the
post you are responding to. I was answering the question posed by
the original poster, "nws".
[/quote]
My apologise, I missed this.
Kevin Aylward
salesEXTRACT@anasoft.co.uk
http://www.anasoft.co.uk
SuperSpice, a very affordable Mixed-Mode
Windows Simulator with Schematic Capture,
Waveform Display, FFT>s and Filter Design. |
|
| |
|
| Back to top |
Kevin Aylward
Guest
|
| Posted: Sun Aug 03, 2003 3:16 pm Post subject: Re: What>s the difference between signals and nets? |
|
|
David Harmon wrote:
[quote]On Sat, 2 Aug 2003 23:28:06 +0100 in sci.electronics.cad,
"Kevin Aylward" <kevin@anasoft.co.uk> wrote:
nws wrote:
In the Eagle system, we have nets and we have signals, what
[chop]
Indeed I do. I am discussing general terms on schematic captures, I
was not referring specifically to idiosyncrasies of Eagle. It was not
apparent that your issue was Eagle specific,
Oh come on, what do you think that "In the Eagle system" as written
by the original poster and quoted in your reply means then? It
means that the whole thread is Eagle specific.
[/quote]
It means the the poster *believes* that the question is specific to a
product.
[quote]It means that your
reply is completely worthless, in the context of the Eagle system.
The qualification "In the Eagle system" could not possibly have been
more clear.
[/quote]
Not relevant. It is very, very common for people to ask questions,
believing that it is specific to some particular product, but in fact,
has nothing particular to do with that product. One only has to note how
often people request PSpice information, when in fact, its generic Spice
information they are after. Go and read some posts, and look at the
replies.
[quote]You failed to read the post you were responding to.
[/quote]
Nope. I took an assumption that the poster might be looking for generic
information. Any fool, can read "Eagle". One tries to read between the
lines when people post. Its very rare that what people request is
actually what they want.
[quote]
If you know what the asinine way that Eagle defines their own
terminology, why did you post in the first place?
I didn>t post "in the first place". Again you failed to read the
post you are responding to. I was answering the question posed by
the original poster, "nws".
[/quote]
My apologise, I missed this.
Kevin Aylward
salesEXTRACT@anasoft.co.uk
http://www.anasoft.co.uk
SuperSpice, a very affordable Mixed-Mode
Windows Simulator with Schematic Capture,
Waveform Display, FFT>s and Filter Design. |
|
| |
|
| Back to top |
Ulrich Strate
Guest
|
| Posted: Mon Aug 04, 2003 12:18 am Post subject: Re: LTSpice--Why not "perfect" results here? |
|
|
Hello,
Helmut Sennewald wrote:
[quote]When I use the correct command PULSE(0 1 0 9.98u 20n 0n 10u),
I still have got the error message and the wrong period.
The only way to overcome this problem is something like
PULSE(0 1 0 9.98u 19.999n 0n 10u) .
[/quote]
I tried n>s instead of u>s and it worked:
PULSE(0 1 0 9.98u 20n 0n 10000n)
regards
Ulrich |
|
| |
|
| Back to top |
Helmut Sennewald
Guest
|
| Posted: Mon Aug 04, 2003 12:34 am Post subject: Re: LTSpice--Why not "perfect" results here? |
|
|
"Ulrich Strate" <df4kv@web.de> schrieb im Newsbeitrag
news:bgjn6f$p6cq8$1@ID-125754.news.uni-berlin.de...
[quote]Hello,
Helmut Sennewald wrote:
When I use the correct command PULSE(0 1 0 9.98u 20n 0n 10u),
I still have got the error message and the wrong period.
The only way to overcome this problem is something like
PULSE(0 1 0 9.98u 19.999n 0n 10u) .
I tried n>s instead of u>s and it worked:
PULSE(0 1 0 9.98u 20n 0n 10000n)
[/quote]
Hello Ulrich,
thanks for the tipp.
I am shure your workaround peforms by luck and nobody can guarantee it
will work with most other numbers. The whole case is simply a floating
point number problem which needs special treatment.
I hope Mike implements something similar to the proposal in my previous
posting.
Best Regards
Helmut |
|
| |
|
| Back to top |
Ulrich Strate
Guest
|
| Posted: Mon Aug 04, 2003 1:46 am Post subject: Re: LTSpice--Why not "perfect" results here? |
|
|
Helmut Sennewald wrote:
[quote]"Ulrich Strate" <df4kv@web.de> schrieb im Newsbeitrag
news:bgjn6f$p6cq8$1@ID-125754.news.uni-berlin.de...
Hello,
Helmut Sennewald wrote:
When I use the correct command PULSE(0 1 0 9.98u 20n 0n 10u),
I still have got the error message and the wrong period.
The only way to overcome this problem is something like
PULSE(0 1 0 9.98u 19.999n 0n 10u) .
I tried n>s instead of u>s and it worked:
PULSE(0 1 0 9.98u 20n 0n 10000n)
Hello Ulrich,
thanks for the tipp.
I am shure your workaround peforms by luck and nobody can guarantee it
will work with most other numbers. The whole case is simply a floating
point number problem which needs special treatment.
[/quote]
Yes, of course I agree. Your suggested cure looks reasonable, rather
adjusting waveforms than clock frequencies which often need to be
xtal-controlled in real life..
I encountered that problem (not considering it as a "bug" then) a few times
before, so I got used to check the error log for timing overlaps followed
by "tuning".
regards
Ulrich
p.s.: THX for your numerous very helpful hints concerning import of models
etc.! |
|
| |
|
| Back to top |
Mike Berger
Guest
|
| Posted: Mon Aug 04, 2003 10:10 pm Post subject: Re: Update... |
|
|
Are hebrew resistors marked from right to left?
"TheResistor.Net" wrote:
[quote]Additionally we have fixed all bugs on the site that were on the
resistor color codes calculators. It is now available working in our
new design in 8 languages - French, English, Norwegian, Hebrew,
German, Portuguese, Spanish & Italian! We will be adding more as time
goes by.[/quote] |
|
| |
|
| Back to top |
Jim Thompson
Guest
|
| Posted: Tue Aug 05, 2003 12:24 am Post subject: Re: LTSpice--Why not "perfect" results here? |
|
|
On 04 Aug 2003 17:45:18 GMT, "Mike Engelhardt" <pmte@concentric.net>
wrote:
[quote]This really just deals with the fact the LTspice has been extended
to allow the pulse function to work as a triangular wave. However,
it could have trouble with both on time and off times being zero,
so it made an adjustment if it didn>t numerically work out so that
it could reliably give you the triangular wave and issue the warning
in the .log file.
To get some perspective, run the following deck in PSpice and see
how a standard simulator completely corrupts the wave form:
* triangular wave
V1 N001 0 pulse(0 1 0 .5m .5m 0 1m)
[snip][/quote]
V1 N001 0 pulse(0 1 0 .5m .5m 1p 1m) ; Works Just Ducky ;-)
...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC>s and Discrete Systems | manus |
| Phoenix, Arizona Voice:(480)460-2350 | |
| Jim-T@analog_innovations.com Fax:(480)460-2142 | Brass Rat |
| http://www.analog-innovations.com | 1962 |
For proper E-mail replies SWAP "-" and "_"
Why is it that Democrats can>t debate politely?
And are only rude and interruptive.
Lack of mental capacity? |
|
| |
|
| Back to top |
Mike Engelhardt
Guest
|
| Posted: Tue Aug 05, 2003 7:59 am Post subject: Re: LTSpice--Why not "perfect" results here? |
|
|
Chris,
LTspice uses it>s own string to double routines so different ways
of writing the same number can lead to a different value in the
equivalent of the 16th digit when converted to binary floating
point.
They all simulate identically under the current version, 2.04t.
--Mike
"Chris Carlen" <crobc@BOGUSFIELD.earthlink.net> wrote in message
news:bgn3o30tma@enews4.newsguy.com...
[quote]Mike Engelhardt wrote:
This really just deals with the fact the LTspice has been extended
to allow the pulse function to work as a triangular wave. However,
it could have trouble with both on time and off times being zero,
so it made an adjustment if it didn>t numerically work out so that
it could reliably give you the triangular wave and issue the warning
in the .log file.
To get some perspective, run the following deck in PSpice and see
how a standard simulator completely corrupts the wave form:
* triangular wave
V1 N001 0 pulse(0 1 0 .5m .5m 0 1m)
R1 N001 0 1K
.tran 3m 3m
.probe
.end
Anyway, the core trouble with Ton=Toff=0 has been resolved in
version 2.04p so that it no longer has to extend the period
numerically significantly to reliably execute a triangular wave.
However, if the period is specified and is less then
Trise+Tfall+Ton, then it is still extended to Trise+Tfall+Ton
and the same warning is issued.
Chris>s deck will now run as he expected.
--Mike
Hi Mike, thanks for the update which I>ll download in a minute.
But how do you explain this (in ver. 2.04o):
* Y:\spice\LT-SWcad\solenoid.drive\pulse-bug.asc
V1 N001 0 PULSE(0 1 0 9u 1u 0 10u)
V2 N002 0 PULSE(0 1 0 9E-6 1E-6 0 10E-6)
V3 N003 0 PULSE(0 1 0 0.000009 0.000001 0 0.00001)
.tran 0 1000u 0 50n
.backanno
.end
The warnings are issued for V1 and V2, but *not* V3 which simulates as I
expected.
???
Good day!
--
_____________________
Christopher R. Carlen
crobc@earthlink.net
Suse 8.1 Linux 2.4.19
[/quote] |
|
| |
|
| Back to top |
Brad Velander
Guest
|
| Posted: Thu Aug 07, 2003 11:50 am Post subject: Re: Protel DXP - Global changes |
|
|
Gone bye, bye!
Try the Altium website to find tutorials and or PDF documents outlining
the new methodogies.
--
Sincerely,
Brad Velander
"EmSinc" <EmSinc@hotmail.com> wrote in message
news:ScdYa.53205$PD3.4550821@nnrp1.uunet.ca...
[quote]Hi,
I>ve just started my first PCB with DXP version and I>m looking for Global
button in the components properties (from 99SE ). Was it removed or
changed
the location only?
Thanks in advance for any help.
Adam
[/quote] |
|
| |
|
| Back to top |
EmSinc
Guest
|
| Posted: Thu Aug 07, 2003 5:14 pm Post subject: Re: Protel DXP - Global changes |
|
|
I found it. The name of the function/button is gone but you can apply the
global changes with Find Similar Objects function from right-click menu in
SCH or PCB.
Adam
"Brad Velander" <spam_this@nowhere.com> wrote in message
news:cFmYa.45289$LD6.993074@news0.telusplanet.net...
[quote]Gone bye, bye!
Try the Altium website to find tutorials and or PDF documents
outlining
the new methodogies.
--
Sincerely,
Brad Velander
"EmSinc" <EmSinc@hotmail.com> wrote in message
news:ScdYa.53205$PD3.4550821@nnrp1.uunet.ca...
Hi,
I>ve just started my first PCB with DXP version and I>m looking for
Global
button in the components properties (from 99SE ). Was it removed or
changed
the location only?
Thanks in advance for any help.
Adam
[/quote] |
|
| |
|
| Back to top |
Spehro Pefhany
Guest
|
| Posted: Thu Aug 07, 2003 9:51 pm Post subject: Re: Post PCB Layout Problem |
|
|
On 7 Aug 2003 09:29:49 -0700, the renowned priyal1977@yahoo.com
(Priyal) wrote:
[quote]Hi all,
I have a PCB that is already fabricated and ready to use. There are
several ICs on this board. I just discovered that the pin-assignment
of one of the ICs differs a lot from the pin-out of its datasheet.
I do NOT want to re-spin the board as it will cost me too much.
I am planning to layout another board (a small one) on which i will
put this IC.
My question is, is there any way i can connect these 2 boards to
communicate with each other? Will these be a feasible solution?
[/quote]
Maybe. What kind of package is involved.. is it a 500+ pin BGA or a
TSSOP-48 or a DIP-40? If it>s a DIP package and you can buy a smaller
SMT part you might be able to make a tiny board that fits overtop the
original footprint.
If it>s just a couple of pins, you might be able to fix it by lifting
legs and using magnet wire and perhaps a drip or two of epoxy to tack
the wires.
Best regards,
Spehro Pefhany
--
"it>s the network..." "The Journey is the reward"
speff@interlog.com Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog Info for designers: http://www.speff.com |
|
| |
|
| Back to top |
EmSinc
Guest
|
| Posted: Fri Aug 08, 2003 1:07 am Post subject: Re: Protel DXP - Global changes |
|
|
I found it. The name of the function/button is gone but you can apply the
global changes with Find Similar Objects function from right-click menu in
SCH or PCB.
Adam
"Brad Velander" <spam_this@nowhere.com> wrote in message
news:cFmYa.45289$LD6.993074@news0.telusplanet.net...
[quote]Gone bye, bye!
Try the Altium website to find tutorials and or PDF documents
outlining
the new methodogies.
--
Sincerely,
Brad Velander
"EmSinc" <EmSinc@hotmail.com> wrote in message
news:ScdYa.53205$PD3.4550821@nnrp1.uunet.ca...
Hi,
I>ve just started my first PCB with DXP version and I>m looking for
Global
button in the components properties (from 99SE ). Was it removed or
changed
the location only?
Thanks in advance for any help.
Adam
[/quote] |
|
| |
|
| Back to top |
|
