ltspice trouble when making a log file.

Dear kees ,

Thanks for sending all the files, but which one do we need to look at?

Best Regards,
Mathias

Problem is with all files, and the more complicated it is, the more problems, Windows can not be used anymore only a hard reset, make a kind of complicated circuit, the problem looks like a memory leak, or the new ltspice needs a lot of memory installed. You can use whever is in the zip file, but how much files you have get?

kees said:

What you mean with options? whjat I have to set to het it right? And also what is that fs/s what happens after 40 us simulation and let stop the software speed, what does not happen in older versions.

If you have changed any Settings o non-default values, and those settings can be set via .options, do so. Maybe you have already done that.

So I ran a 30ms sim without issue on the old mac.

RAWdocument - 26.3 GB

FFT looks different, but not sure what FFT settings you used.

mike

I am busy with testing things, you guys use wine, and a mac, I have ask AI and this is not a good comparison, I have a windows 10 pc so there is a possible ptoblem with the memory manager, I have not set an big page file on a separate computer and test it. Mr MStokowski I see in the fft distortion under the fase also the strange signs as I had, it is still present in the new versions. regards kees

Hi kees ,

kees said:

I have a windows 10 pc so there is a possible ptoblem with the memory manager,

Ah, unfortunately, I believe I've updated all of my Windows machines to 11, as Micro$loth is pretty much forcing upgrades, so I can't test Win 10. Sorry about that. I could try in a Win 10 bottle under Wine, but maybe no point?

kees said:

I am busy with testing things, you guys use wine, and a mac,

True, but I wanted to see if the long sim would run without crash, while producing a very large RAW file. I'll run it on my i5, 12GB Win 11 machine.

kees said:

also the strange signs as I had, it is still present in the new versions

Well, look at that. I'll follow up on this one. I wonder why it was correct in my first run…

mike

Oke, clearly I have no crash, I have a uge slowdown of windows and the apps who are open like LTspice, this happens when I do a long sim, and the sim is done succesfully, but then when the log is made, then and only then the system get on lock, and so slow that I can do nothing anymore, I did manage between the locks to close ltspice but the error, and as such the memory problem did hold on, and go on, the pc did never came out anymore, that is what happens, and I think windows 11 or 10 make not that much difference. 

So now, I go try some things, I have time enough as a pensionate guy to see why, as I did say XVII has no problem whatsoever with the file, making a nice log. the old 24.0.1 did also make a nice log no problems. but had a other issue like unresponsive on buttons when using sometimes, but that is not such a problem, and was gone in later versions, when memory trouble came in.

I go sim some different times with different setups like the page file of windlws making it big on a separate harddisk, and I do try without marching waves etc.

And of course I have a new noard who can do win 11 and will go set it up soon, using this computer with linux, and wine, and try then.

Let you now what the different setups do,  thanks so far for help

I have done a uge sim with marching waves disabled, but it did still slow down windows such that it is not possible to work with it anymore, but much less then first, I have close ltspice and then the harddisc action go on for a while slowing windows, but it does stop after 10 minutes and the windows is normal;ly again, it has to do with or swapping on the harddisc or it has to do with the memory manager from windows. As AI says maybe mine pc with a older cpu and board is the problem, I have a I5 ly down somewhere and have a pc with a I think I7 or I9 in it, do not now exact.

Oke I have done some things, like .save V(feedback I(R29). but can not probe then the circuit except the labels.

Looks like it make now a much smaller raw file. I do wait for the end of the sim.

Mac and wine do work complete different and you have 16 threads while I have four, can go to 8 with the other processor I have ly down somewhere.

Hi Mike. 

I have found the solution, and this is to use .save directive. This did the trick, Buit I do need to make labels for that.  I do not need the data of all these wires so now it sims fine and have no issues anymore with the memory in windows.

In mine case I did .save I(r29) V(out), and can include more labels because now I can not probe the other stuff anymore except when I do set it to do, so what I do is with normal sim to make the .save a text file and switch back when needed, the simulation looks faster also and the log file was there instantly after ltspice was done simulating.

I go tell this also on diyaudio, but I presume you guys did already now this but mine questions  where some way unclear.

Some question? yes, I have, how to interpretere the distortion, why total distortion is higher, I presume it then also take the pwm signals itself also, and the partial is the real audio distortion?

thanks and have a nice day, (or night there).

Hi kees ,

I'm afraid I cannot reproduce this behavior, even on the old Windows machine. I did run a 50ms sim, producing quite a large file. It does take a while to display waveforms.

I'm assuming the second video is 26.0.1? 

mike

Hi mike.

I do now use labels, so the signals of interests are saved and nothing else making a nice small file without any problems and super fast.

I do not deny somthing on the pc is the trouble, for example a other program who has a dll loaded in memory conflicting the ltspice program.

As mine pc is older, I have a much newer one here with ddr4 memory and 7700 cpu I7 I think this will be much better.

The big files do swap, and I think when I let the pc do his job, I think after a hour or so the problem I have autoimatically stop when log creation is ready, maybe LTspice does copy the raw file or swap it in memory I do not now.

But I have a work around, and maybe the real cause will be discovered later, or it was a conflict with other software.

Yes Ihave used the latest LTspice, the only one who have not this behavior is the older 24.0.1 just one or two updates later I did get this problem.

regards kees

Hi kees ,

kees said:

As mine pc is older, I have a much newer one here with ddr4 memory and 7700 cpu I7 I think this will be much better.

My guess is that this should work. Let me know.

kees said:

But I have a work around, and maybe the real cause will be discovered later, or it was a conflict with other software.

That is my guess also. Windows 10 and 11 should be no different, that we know of, so maybe another culprit. That said, keep me informed of anything you discover. All data is useful to helping others.

thanks,

mike

I can manage fine now, the .save directive is of much valeu to limit the raw file so it does run well. Soon I set op a new win 11 abnd look what it does, I can also do load wndons without the rest like programs and look at the results, it can give interestoing data aabout the LTspice program.

I have however a question I do not get the hang of, that is to simulate open loop and closed loop.

When I use nested feedback, like three opamps in example, where do I need to set the injection voltage source to get the real loop data? Nested loops are a problem because these have all a local feedback from the amp output into the integrator and Pi opamps to correct for the pwm errors.

Maybe you can tell me what I need to do, I use behavior siwtches and a delay line to sim the pwm stage non lineair behavior, I need a gain section who is non lineair, I can not use gain blocks here.

regards 

kees

I can manage fine now, the .save directive is of much valeu to limit the raw file so it does run well. Soon I set op a new win 11 abnd look what it does, I can also do load wndons without the rest like programs and look at the results, it can give interestoing data aabout the LTspice program.

I have however a question I do not get the hang of, that is to simulate open loop and closed loop.

When I use nested feedback, like three opamps in example, where do I need to set the injection voltage source to get the real loop data? Nested loops are a problem because these have all a local feedback from the amp output into the integrator and Pi opamps to correct for the pwm errors.

Maybe you can tell me what I need to do, I use behavior siwtches and a delay line to sim the pwm stage non lineair behavior, I need a gain section who is non lineair, I can not use gain blocks here.

regards 

kees