Re: Simulation runs very slowly: test.asc
Thanks Tony. Yes, there’s the LT3080 and its ilk as well as many others. Appreciate your forthcoming effort in checking what you have and uploading it. ?
toggle quoted message
Show quoted text
From: [email protected] <[email protected]> On Behalf Of Tony Casey via groups.io
Sent: Friday, April 4, 2025 6:46 PM
To: [email protected]
Subject: Re: [LTspice] Simulation runs very slowly: test.asc? On 04/04/2025 23:51, Christopher Paul via groups.io wrote: There are at least two unencrypted models I saw today: LM317_TRANS (No note on PSRR) and a more extensive model LM317A_TRANS, which states: ? * Model Usage Notes: * * A. Features have been modelled *????????? 1. Startup Response *????????? 2. Peak Current Limit *????????? 3. Dropout Voltage vs Output Current? *????????? 4. Line & Load Transinet (sic) Response * * B. Features haven't been modelled *????????? 1. Quiescent Current vs Input Voltage *????????? 2. Temperature dependent characteristics *????????? 3. PSRR Response ? If the more extensive model says no to PSRR, I don’t hold out hope for the less extensive one.
The other ones I have are decent enough. I'll check them against the datasheet, before uploading anything.
TBH, you can do a lot better than the LM317 these days. ?
|
Re: Simulation runs very slowly: test.asc
Agree with the “precise” correction. ? To make a current source, a positive supply is connected to the LM317’s IN pin, a resistor R is connected between OUT and ADJ, and the ADJ pin is connected to a grounded load. Here R = 1.25V / ILoad, where ILoad is the desired load current. This two-component current source circuit has only two connections to the outside world. Excite the source at IN and measure the load current, or put an excited voltage source in series with the load and measure the current that flows when the ADJ source is a simple DC voltage (accounting for the load impedance); PSRR (ohms) is the same as the impedance seen by the load. ? For my simulation, I used?LM317A current_source.zip??Current source circuit, using TI's own LM317A model which I found in the User Group files along with others. I could go into detail about why PSRR is not working properly if you wish. ? At some point I’m going try the more promising transistor model?LM317_Test.zip?Working test schematic of LM317, symbol and model originally uploaded by Jason Vanryan. Performance not verified against datasheet. ? Chris ? ? ?
toggle quoted message
Show quoted text
From: [email protected] <[email protected]> On Behalf Of Andy I via groups.io
Sent: Friday, April 4, 2025 6:28 PM
To: [email protected]
Subject: Re: [LTspice] Simulation runs very slowly: test.asc? On Fri, Apr 4, 2025 at 05:46 PM, Christopher Paul wrote: The problem is that the notes in the T.I. model specifically state that it is NOT applicable to PSRR.
To be more precise, it says that PSRR Response "hasn't been modelled", whatever that means.? It might mean only that the effect may be there but nobody has calibrated or checked it yet for conformance with the specs.? But I'm guessing that it is more than that. The lack of mention of PSRR in the LM317 (not -A) model doesn't imply that PSRR is correctly modeled there.? Unfortunately, it doesn't mean anything.??No way to tell without trying it. ... And since the standard LM317 one-resistor current source is essentially a two-terminal device, PSRR (characterized as supply voltage divided by load current) is identical to Zout, so that’s out the window too.
|
Certainly room for improvement, but FAR better than squinting at the screen to read the net names, or worse, hovering the probe, because I haven’t assigned all of them yet!
?
toggle quoted message
Show quoted text
From: [email protected] <[email protected]>
On Behalf Of Andy I via groups.io
Sent: Saturday, April 05, 2025 9:57 AM
To: [email protected]
Subject: EXTERNAL: Re: [LTspice] .MEAS scripting
?
I usually ignore that box in the top right.
Ctrl-F > Type text > Text becomes highlighted if it is found.
That's all I needed to see.
LTspice always was a bit picky about what it found.? e.g., it only finds text that starts with what you type for it to find.? No wildcards, and no finding strings embedded within longer strings, except when it starts the longer string.
|
I usually ignore that box in the top right.
?
Ctrl-F > Type text > Text becomes highlighted if it is found.
?
That's all I needed to see.
?
LTspice always was a bit picky about what it found.? e.g., it only finds text that starts with what you type for it to find.? No wildcards, and no finding strings embedded within longer strings, except when it starts the longer string.
?
Andy
|
Thank you.
On 2025-04-05 17:36, Bell, Dave via
groups.io wrote:
I love “rural
charm of LTspice”!
Wild critters and
cow pies…
?
?
All
part of the fascinating rural charm of LTspice.?
On 2025-04-05 17:28, Bell, Dave via
groups.io wrote:
I felt
very stupid a while after that email:
The search
box seems always to be there, but only becomes active
after Ctrl-F.
I just
never saw anything pop up or change, so ASSumed it didn’t
work!
The box is
not labelled, just a blank area on the toolbar, and after
Ctrl-F, a slightly smaller input box outline appears in
the larger? blank area. Unless one is expecting it or
looking directly at the normally inactive area, . . .
?
Dave
?
?
On 04/04/2025 21:34, Bell, Dave via
groups.io wrote:
Related, there apparently used
to be a Ctrl-F search for net nets by name.
I am running XVII, and that
does nothing.
Also, found a hint that maybe
the search function was hiding under Tools, but can’t
find it there.
AND, is there (still?) a
Highlight Nets function?
Find
<NETNAME> does work in XVII. Ctrl-F brings up a search
box at the top right in the icon bar. Enter any text and it
will be highlighted if it exists, and if it includes a net
name, the net connections will be highlighted too. But the
search is strictly text. So make sure to name your nets with
that in mind.
?
--
OOO - Own Opinions only If something is true: * as far as we
know - it's science *for certain - it's mathematics
*unquestionably - it's religion
--
OOO - Own Opinions only
If something is true:
* as far as we know - it's science
*for certain - it's mathematics
*unquestionably - it's religion
|
I love “rural charm of LTspice”!
Wild critters and cow pies…
?
toggle quoted message
Show quoted text
From: [email protected] <[email protected]>
On Behalf Of John Woodgate
Sent: Saturday, April 05, 2025 9:32 AM
To: [email protected]
Subject: EXTERNAL: Re: [LTspice] .MEAS scripting
?
All part of the fascinating rural charm of LTspice.?
On 2025-04-05 17:28, Bell, Dave via groups.io wrote:
I felt very stupid a while after that email:
The search box seems always to be there, but only becomes active after Ctrl-F.
I just never saw anything pop up or change, so ASSumed it didn’t work!
The box is not labelled, just a blank area on the toolbar, and after Ctrl-F, a slightly smaller input box outline appears in the larger? blank area. Unless one is expecting it or looking directly at the normally
inactive area, . . .
?
Dave
?
?
On 04/04/2025 21:34, Bell, Dave via groups.io wrote:
Related, there apparently used to be a Ctrl-F search for net nets by name.
I am running XVII, and that does nothing.
Also, found a hint that maybe the search function was hiding under Tools, but can’t find it there.
AND, is there (still?) a Highlight Nets function?
Find <NETNAME> does work in XVII. Ctrl-F brings up a search box at the top right in the icon bar. Enter any text and it will be highlighted if it exists, and if it includes a net name, the net connections will
be highlighted too. But the search is strictly text. So make sure to name your nets with that in mind.
?
--
OOO - Own Opinions only If something is true: * as far as we know - it's science *for certain - it's mathematics *unquestionably - it's religion
|
All part of the fascinating rural charm of
LTspice.?
On 2025-04-05 17:28, Bell, Dave via
groups.io wrote:
I felt very
stupid a while after that email:
The search
box seems always to be there, but only becomes active after
Ctrl-F.
I just never
saw anything pop up or change, so ASSumed it didn’t work!
The box is
not labelled, just a blank area on the toolbar, and after
Ctrl-F, a slightly smaller input box outline appears in the
larger? blank area. Unless one is expecting it or looking
directly at the normally inactive area, . . .
?
Dave
?
?
On 04/04/2025 21:34, Bell, Dave via
groups.io wrote:
Related, there apparently used to
be a Ctrl-F search for net nets by name.
I am running XVII, and that does
nothing.
Also, found a hint that maybe the
search function was hiding under Tools, but can’t find it
there.
AND, is there (still?) a
Highlight Nets function?
Find
<NETNAME> does work in XVII. Ctrl-F brings up a search
box at the top right in the icon bar. Enter any text and it
will be highlighted if it exists, and if it includes a net
name, the net connections will be highlighted too. But the
search is strictly text. So make sure to name your nets with
that in mind.
?
--
OOO - Own Opinions only
If something is true:
* as far as we know - it's science
*for certain - it's mathematics
*unquestionably - it's religion
|
I felt very stupid a while after that email:
The search box seems always to be there, but only becomes active after Ctrl-F.
I just never saw anything pop up or change, so ASSumed it didn’t work!
The box is not labelled, just a blank area on the toolbar, and after Ctrl-F, a slightly smaller input box outline appears in the larger? blank area. Unless one is expecting it or looking directly at the normally
inactive area, . . .
?
Dave
?
toggle quoted message
Show quoted text
From: [email protected] <[email protected]>
On Behalf Of Tony Casey
Sent: Saturday, April 05, 2025 2:19 AM
To: [email protected]
Subject: EXTERNAL: Re: [LTspice] .MEAS scripting
?
On 04/04/2025 21:34, Bell, Dave via groups.io wrote:
Related, there apparently used to be a Ctrl-F search for net nets by name.
I am running XVII, and that does nothing.
Also, found a hint that maybe the search function was hiding under Tools, but can’t find it there.
AND, is there (still?) a Highlight Nets function?
Find <NETNAME> does work in XVII. Ctrl-F brings up a search box at the top right in the icon bar. Enter any text and it will be highlighted if it exists, and if it includes a net name, the net connections will
be highlighted too. But the search is strictly text. So make sure to name your nets with that in mind.
?
|
What I meant is that the search context is just text. You can't specify what type of object you're searching for. -- Regards, Tony On 5 April 2025 15:01:18 CEST, "Andy I via groups.io" <AI.egrps+io@...> wrote: On Sat, Apr 5, 2025 at 05:18 AM, Tony Casey wrote:
... But the search is strictly text. So make sure to name your nets with
that in mind.
Can a net name (or any other schematic text) be anything other than text?
Andy
--<br>Regards,<br>Tony<br>
|
'text' in this context may mean a specific
character set, such as 'Extended ASCII'. I've not found a way to
put Greek letters in any LTspice text, but some people seem to
be able to. Apart form '?', that is.
On 2025-04-05 14:01, Andy I via
groups.io wrote:
On Sat, Apr 5, 2025 at 05:18 AM, Tony Casey wrote:
... But the search is strictly text. So make sure to
name your nets with that in mind.
Can a net name (or any other schematic text) be anything other
than text?
?
Andy
?
--
OOO - Own Opinions only
If something is true:
* as far as we know - it's science
*for certain - it's mathematics
*unquestionably - it's religion
|
On Sat, Apr 5, 2025 at 05:18 AM, Tony Casey wrote:
... But the search is strictly text. So make sure to name your nets with that in mind.
Can a net name (or any other schematic text) be anything other than text?
?
Andy
?
|
On 04/04/2025 21:34, Bell, Dave via
groups.io wrote:
Related, there
apparently used to be a Ctrl-F search for net nets by name.
I am running XVII,
and that does nothing.
Also, found a hint
that maybe the search function was hiding under Tools, but
can’t find it there.
AND, is there
(still?) a Highlight Nets function?
Find <NETNAME> does work in XVII. Ctrl-F brings up a search
box at the top right in the icon bar. Enter any text and it will be
highlighted if it exists, and if it includes a net name, the net
connections will be highlighted too. But the search is strictly
text. So make sure to name your nets with that in mind.
--
Regards,
Tony
|
Re: Switching Speed of a BJT
I realize the difficulty.
The direct language used by Jonathan King doesn't bother me; it's what moves us forward, provided it's objective.
|
Re: Switching Speed of a BJT
All the other responders make good points.? ?
?
First, understand how a BJT works.? After that, improve your understanding on modeling a BJT in LTSPICE.? Apologies for being blunt, but there is not a good alternative to understanding.? Blundering your way to a good simulation result isn't knowledge.
?
?
|
Re: Simulation runs very slowly: test.asc
On 04/04/2025 23:51, Christopher Paul
via groups.io wrote:
There are
at least two unencrypted models I saw today: LM317_TRANS (No
note on PSRR) and a more extensive model LM317A_TRANS, which
states:
?
* Model
Usage Notes:
*
* A.
Features have been modelled
*?????????
1. Startup Response
*?????????
2. Peak Current Limit
*?????????
3. Dropout Voltage vs Output Current?
*?????????
4. Line & Load Transinet (sic) Response
*
* B.
Features haven't been modelled
*?????????
1. Quiescent Current vs Input Voltage
*?????????
2. Temperature dependent characteristics
*?????????
3. PSRR Response
?
If the more
extensive model says no to PSRR, I don’t hold out hope for the
less extensive one.
The other ones I have are decent enough. I'll check them against the
datasheet, before uploading anything.
TBH, you can do a lot better than the LM317 these days.
--
Regards,
Tony
|
Re: Simulation runs very slowly: test.asc
On 04/04/2025 23:22, Andy I via
groups.io wrote:
To further
complicate this, the non-A SPICE model has 4 pins whereas the
physical device has only 3, if you don't count the physical tab,
which is supposed
to connect to the OUT pin.? In the SPICE model, the extra pin is
named as if it was a second output, but in reality it goes
nowhere!? It is a mistake, and has gone undetected for 10+ years
at T.I.
That's one of the changes I had to make. The model is non-workable
as downloaded, as the pin OUT_1 doesn't connect to anything. I just
deleted it. I guess no one has actually checked it. Sad.
--
Regards,
Tony
|
Re: Simulation runs very slowly: test.asc
On Fri, Apr 4, 2025 at 05:46 PM, Christopher Paul wrote:
The problem is that the notes in the T.I. model specifically state that it is NOT applicable to PSRR.
To be more precise, it says that PSRR Response "hasn't been modelled", whatever that means.? It might mean only that the effect may be there but nobody has calibrated or checked it yet for conformance with the specs.? But I'm guessing that it is more than that.
?
The lack of mention of PSRR in the LM317 (not -A) model doesn't imply that PSRR is correctly modeled there.? Unfortunately, it doesn't mean anything.??No way to tell without trying it.
?
... And since the standard LM317 one-resistor current source is essentially a two-terminal device, PSRR (characterized as supply voltage divided by load current) is identical to Zout, so that’s out the window too.
I don't follow.
?
Andy
?
|
Re: Simulation runs very slowly: test.asc
There are at least two unencrypted models I saw today: LM317_TRANS (No note on PSRR) and a more extensive model LM317A_TRANS, which states: ? * Model Usage Notes: * * A. Features have been modelled *????????? 1. Startup Response *????????? 2. Peak Current Limit *????????? 3. Dropout Voltage vs Output Current? *????????? 4. Line & Load Transinet (sic) Response * * B. Features haven't been modelled *????????? 1. Quiescent Current vs Input Voltage *????????? 2. Temperature dependent characteristics *????????? 3. PSRR Response ? If the more extensive model says no to PSRR, I don’t hold out hope for the less extensive one. ?
toggle quoted message
Show quoted text
From: [email protected] <[email protected]> On Behalf Of Tony Casey via groups.io
Sent: Friday, April 4, 2025 2:08 PM
To: [email protected]
Subject: Re: [LTspice] Simulation runs very slowly: test.asc? On 04/04/2025 18:14, Christopher Paul via groups.io wrote: Yes, and I’m not impressed with other aspects of their model, either. The notes for the most advanced T.I. LM317A model I can find specifically state that the model does not address PSRR. What other small signal characteristics does it not address? I also noticed that ADJ terminal’s DC current is 0 rather than the datasheet’s nominal 50uA.
I downloaded an LM317 unencrypted behavioural model from TI's website today, and in the notes it makes no mention that PSRR is not modelled. Looking through the code, I can see that it is modelled. However, the PSRR isn't modelled at all versus output current or output capacitor value. On that basis, I wouldn't use it. (I also had to make a couple of changes to the model to get it to work at all.)
I also have several other LM317 models, including one from Motorola, which are transistor level models, and those definitely do model PSRR, although there are variations between them.
They all show that you should use a minimum of about 47μF for the output capacitor - preferably greater. The PSRR also improves considerably if you if you add a smaller cap from "ADJ" to GND.
--
Regards,
Tony
|
Re: Simulation runs very slowly: test.asc
Yes, a full transistor model would be great! ? The problem is that the notes in the T.I. model specifically state that it is NOT applicable to PSRR. And the results I get when I do simulate for PSRR are wildly different from some bench test results I’ve gotten. And since the standard LM317 one-resistor current source is essentially a two-terminal device, PSRR (characterized as supply voltage divided by load current) is identical to Zout, so that’s out the window too. ? ?
toggle quoted message
Show quoted text
From: [email protected] <[email protected]> On Behalf Of Andy I via groups.io
Sent: Friday, April 4, 2025 1:25 PM
To: [email protected]
Subject: Re: [LTspice] Simulation runs very slowly: test.asc? On Fri, Apr 4, 2025 at 12:14 PM, Christopher Paul wrote: ... By any chance do you know of models that can be relied upon to assess PSRR and output impedance?
I don't know how well this works, but....? In theory, if you can find a full transistor-level model for one, it ought to be reasonable for PSRR and Zout.? However, finding one could be a challenge. Also, there are issues between the design or intended circuit, and the actual circuit including parasitics.? Not just talking about capacitance.? If I remember correctly, Hans Camenzind (creator of the 555) wrote pretty good of the differences between intended circuit and what you get when you put it in an IC wafer.
|
Re: Simulation runs very slowly: test.asc
On Fri, Apr 4, 2025 at 02:08 PM, Tony Casey wrote:
I downloaded an LM317 unencrypted behavioural model from TI's website today, and in the notes it makes no mention that PSRR is not modelled. ...
It is noteworthy that the two parts (LM317 and LM317A) have such different SPICE models - and the fact that they have different datasheets and do not even share the same webpage.? At first it seems like they could be the same part but where the one with the -A suffix is just a more accurate version of the other.? Maybe not.
?
It makes one wonder if they were designed differently.? Or perhaps by different design teams at T.I.
?
It is odd.
?
If they were variants of the same part, then their SPICE models ought to be interchangeable.? It is uncommon for a SPICE model to include inaccuracy.? If that were the only difference, then both SPICE models would be the same; but clearly they are not.
?
To further complicate this, the non-A SPICE model has 4 pins whereas the physical device has only 3, if you don't count the physical tab, which is supposed to connect to the OUT pin.? In the SPICE model, the extra pin is named as if it was a second output, but in reality it goes nowhere!? It is a mistake, and has gone undetected for 10+ years at T.I.
?
Andy
?
|
Bell, Dave
John Woodgate
Tony Casey
jacfev