Hi Guys
I reloaded LTspice the other day, then had the usual problem that the models did not link up and fixed that. Changed the background and element colours as before. Then I tried running a circuit that had previously worked extremely well - low THD, wide bandwidth, all the good stuff. But...
Now the circuits that worked before do not behave properly.
I reloaded LTspice because there were some circuits that it was having trouble finding the DC conditions for, namely class-G and class-H power amps. My simulation computer is off-line so I transfer LTspice using a USB key. When I began the reload on the offline computer, the pop-up window asked if I wanted to over-write - which it recommended - or to update? I let it over-write.
I tried a couple of other previously working circuits and they now don't perform as before.
This casts a doubt as to which result is correct? Were the previous results all flawed? Or are the new ones flawed? How can I believe LTspice if it is this variable?
Has anyone else encountered this? There is basic computer stuff I don't know and maybe I just have to reload it again but select 'update'.
Thanks for any advice.
I reloaded LTspice the other day, then had the usual problem that the models did not link up and fixed that. Changed the background and element colours as before. Then I tried running a circuit that had previously worked extremely well - low THD, wide bandwidth, all the good stuff. But...
Now the circuits that worked before do not behave properly.
I reloaded LTspice because there were some circuits that it was having trouble finding the DC conditions for, namely class-G and class-H power amps. My simulation computer is off-line so I transfer LTspice using a USB key. When I began the reload on the offline computer, the pop-up window asked if I wanted to over-write - which it recommended - or to update? I let it over-write.
I tried a couple of other previously working circuits and they now don't perform as before.
This casts a doubt as to which result is correct? Were the previous results all flawed? Or are the new ones flawed? How can I believe LTspice if it is this variable?
Has anyone else encountered this? There is basic computer stuff I don't know and maybe I just have to reload it again but select 'update'.
Thanks for any advice.
did you add/modify models in the LTspice libs? - I always keep separate libs and use .include
the DC operating point and convergence can be different - did you change setting in Tools>Control Panel>Spice - integration type, tolerances, other sim settings?
and sometimes Mike has tweaked the underlying algorithms for convergence, device modeling
check the changelog.txt for clues to new solver or model algorithms
the DC operating point and convergence can be different - did you change setting in Tools>Control Panel>Spice - integration type, tolerances, other sim settings?
and sometimes Mike has tweaked the underlying algorithms for convergence, device modeling
check the changelog.txt for clues to new solver or model algorithms
Hi Guys
jcx: I simply installed a new version of LTspice, allowing it to over-write the old one, which turned out to be a mistake. It seems that most "updates" are just library models, mostly LT parts but possibly them adding models from others. I don't know the extent of their changes.
When I tried to run a previously working circuit, the models I was using couldn't be found.
I have a separate folder on my desktop that is the models I used. I thought this was truly separate from LTpspice in how the files are handled, but again, my lack of how the computer does certain things may be causing a misinterpretation here.
I went through the my computer path to look at the libraries and changed the names of the models which now had _C instead of simply C after them (using a Cordell BJT model as an example here). This allowed the simulator to recognise the models as they are named on the schematic, and nothing worked as before. It seems like new models of the BJTs were loaded.
The only things I modified in the Tools was the colour choices for the background, lines, parts, unconnected juntcions, and waves.
I'll see if I can find the changelog.txt
I like to have my schematics start off uncluttered, just with the sim operating commands, the schematic name - to which I'll add data notes as I do sims. I prefer to have the simulator recognise the parts "invisibly" without having to list all the models or folders on the schematic. The only schematics I can't get to work that way are ones with tubes. For those I have to add the .include statement.
I think I'll end up having to re-install the models by hand as I did before. My sim computer is offline and I have to transfer the models using a key.
Thanks
jcx: I simply installed a new version of LTspice, allowing it to over-write the old one, which turned out to be a mistake. It seems that most "updates" are just library models, mostly LT parts but possibly them adding models from others. I don't know the extent of their changes.
When I tried to run a previously working circuit, the models I was using couldn't be found.
I have a separate folder on my desktop that is the models I used. I thought this was truly separate from LTpspice in how the files are handled, but again, my lack of how the computer does certain things may be causing a misinterpretation here.
I went through the my computer path to look at the libraries and changed the names of the models which now had _C instead of simply C after them (using a Cordell BJT model as an example here). This allowed the simulator to recognise the models as they are named on the schematic, and nothing worked as before. It seems like new models of the BJTs were loaded.
The only things I modified in the Tools was the colour choices for the background, lines, parts, unconnected juntcions, and waves.
I'll see if I can find the changelog.txt
I like to have my schematics start off uncluttered, just with the sim operating commands, the schematic name - to which I'll add data notes as I do sims. I prefer to have the simulator recognise the parts "invisibly" without having to list all the models or folders on the schematic. The only schematics I can't get to work that way are ones with tubes. For those I have to add the .include statement.
I think I'll end up having to re-install the models by hand as I did before. My sim computer is offline and I have to transfer the models using a key.
Thanks
Hi Guys
"did you add/modify models in the LTspice libs? - I always keep separate libs and use .include"
A year ago when I first installed LTspice I installed models and modified one to match the devices I actually use. All of that worked well until a month ago when I started simulating class-H amps, then it had trouble finding the DC operating point. That is why I tried to reload LTspice.
"the DC operating point and convergence can be different - did you change setting in Tools>Control Panel>Spice - integration type, tolerances, other sim settings?"
Did not change these settings as I don't know what they are for.
"and sometimes Mike has tweaked the underlying algorithms for convergence, device modeling
check the changelog.txt for clues to new solver or model algorithm"
Tried to find the changelog but no luck. Like I said, there is a lot of basic computer stuff I don't know.
Last night I tried reloading the models since running a sim results in dialog boxes saying "Cannot find so-and-so". Eventually got it to find the models but things still do not sim as they used to.
Back when I put LTspice on this computer a year ago, I also put it on another off-line computer I have. This other computer is slower but it has all the models I've been using. So, I copied a schematic from the problem computer and ran it on the slow computer with results consistent to how things used to be. This suggests to me there is nothing wrong with the models or the schematics and that I can hopefully trust the results from the past year.
In turn, it suggests maybe there is some other fault in my problem computer? Possibly not related to LTspice? The problem computer has two versions of XP on it, where the slow computer that runs fine has only the older version of XP. I'm not sure if they went from 32-bit to 64-bit or something like that during the life cycle of XP? Anyway, the slow computer would have the older form and runs the circuits properly. I haven't tried the class-H sim in it but will shortly. I think if it runs it better than the problem computer did that points to some other faut than LTspice? Not sure.
Thanks for any advice!
"did you add/modify models in the LTspice libs? - I always keep separate libs and use .include"
A year ago when I first installed LTspice I installed models and modified one to match the devices I actually use. All of that worked well until a month ago when I started simulating class-H amps, then it had trouble finding the DC operating point. That is why I tried to reload LTspice.
"the DC operating point and convergence can be different - did you change setting in Tools>Control Panel>Spice - integration type, tolerances, other sim settings?"
Did not change these settings as I don't know what they are for.
"and sometimes Mike has tweaked the underlying algorithms for convergence, device modeling
check the changelog.txt for clues to new solver or model algorithm"
Tried to find the changelog but no luck. Like I said, there is a lot of basic computer stuff I don't know.
Last night I tried reloading the models since running a sim results in dialog boxes saying "Cannot find so-and-so". Eventually got it to find the models but things still do not sim as they used to.
Back when I put LTspice on this computer a year ago, I also put it on another off-line computer I have. This other computer is slower but it has all the models I've been using. So, I copied a schematic from the problem computer and ran it on the slow computer with results consistent to how things used to be. This suggests to me there is nothing wrong with the models or the schematics and that I can hopefully trust the results from the past year.
In turn, it suggests maybe there is some other fault in my problem computer? Possibly not related to LTspice? The problem computer has two versions of XP on it, where the slow computer that runs fine has only the older version of XP. I'm not sure if they went from 32-bit to 64-bit or something like that during the life cycle of XP? Anyway, the slow computer would have the older form and runs the circuits properly. I haven't tried the class-H sim in it but will shortly. I think if it runs it better than the problem computer did that points to some other faut than LTspice? Not sure.
Thanks for any advice!
Last edited:
Hi Guys
Today I tried running the class-H circuits that were being problematic in my fast computer in my slow computer and they ran fine. Where the fast computer couldn't find the DC operating points easily, the slow computer has no issue and does it fast.
Just for a reminder: Both computers are off-line and had LTspice loaded into them about a year ago. At the same time I loaded models and everything worked okay - just the fast computer was faster... hehe
About a month ago, I was simulating class-H amp circuits on the fast computer and it had trouble finding the DC operating point. Other circuits seemed to still sim okay. I thought that reloading an updated LTspice might help, but that has done nothing for the fast computer. It would not recognise the modesl and had over-written many of them. Reloading the models and fixing the recognition issue still left the system not running any of the amp circuits properly - everything comes up unstable.
Meanwhile, loading all the schematics from the fast computer onto the slow computer and running them there gets the historic results of stability and good performance.
So... I am left thinking there is something amisss with the fast computer that may or may not be a spice issue? The only other thing I do on that computer is word processing and that seems unaffected.
Have fun
Today I tried running the class-H circuits that were being problematic in my fast computer in my slow computer and they ran fine. Where the fast computer couldn't find the DC operating points easily, the slow computer has no issue and does it fast.
Just for a reminder: Both computers are off-line and had LTspice loaded into them about a year ago. At the same time I loaded models and everything worked okay - just the fast computer was faster... hehe
About a month ago, I was simulating class-H amp circuits on the fast computer and it had trouble finding the DC operating point. Other circuits seemed to still sim okay. I thought that reloading an updated LTspice might help, but that has done nothing for the fast computer. It would not recognise the modesl and had over-written many of them. Reloading the models and fixing the recognition issue still left the system not running any of the amp circuits properly - everything comes up unstable.
Meanwhile, loading all the schematics from the fast computer onto the slow computer and running them there gets the historic results of stability and good performance.
So... I am left thinking there is something amisss with the fast computer that may or may not be a spice issue? The only other thing I do on that computer is word processing and that seems unaffected.
Have fun
Something must have changed on the fast computer to make it malfunction... The quick way to verify it, is to copy the LTSpice directory from the slow computer to it and try again. I think it will work if you just overwrite all the files, but you can also try putting them in another directory to be safe.
The default location of the installation is at C:\Program Files\LTC, so copy the whole thing to a thumbdrive from the slow computer and paste it back into the fast computer. Good luck!
The default location of the installation is at C:\Program Files\LTC, so copy the whole thing to a thumbdrive from the slow computer and paste it back into the fast computer. Good luck!
Last edited:
Hi Guys
jazbo80: By "thumbdrive" I assume you mean a USB key? That is how I have to get spice and other things to these computers anyway...
I'll have my wife look at the "copy the directory" idea since she is better versed with computer software than me.
At this point I'm happy that the slow computer is running the files as they used to run, nominally verifying that that work is "good".
Thanks.
jazbo80: By "thumbdrive" I assume you mean a USB key? That is how I have to get spice and other things to these computers anyway...
I'll have my wife look at the "copy the directory" idea since she is better versed with computer software than me.
At this point I'm happy that the slow computer is running the files as they used to run, nominally verifying that that work is "good".
Thanks.
Hi Guys
Thanks for the guidance, jazbo8. I talked to my wife about that and she thought it would be good as a backup after trying simply reloading the old version of LTspice that I still have on the USB key. That worked. I copied the models from the folder on the desktop and the fast computer is running spice as it used to !
Everything is working now. Thanks for your help.
Have fun
Thanks for the guidance, jazbo8. I talked to my wife about that and she thought it would be good as a backup after trying simply reloading the old version of LTspice that I still have on the USB key. That worked. I copied the models from the folder on the desktop and the fast computer is running spice as it used to !
Everything is working now. Thanks for your help.
Have fun
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.