LtSpice and split supplies :-(

Thread Starter

Hypatia's Protege

Joined Mar 1, 2015
3,228
Discontinuous investigation of continuous phenom would need infinite data and time!
Can it be? Aleph(0) pondering Aleph(1)?:D:D:D

So lighten up it not end of world:rolleyes:
Who knows? Who cares?
Might I ask what part of "It's good to know which results to trust" don't you understand!
It's easy to minimize everything on the basis that 'nothing really maters' -- howbeit I find said practice both cynical and straight up lazy! - in your aversion to what passes for 'drama' you have renounced intellectual integrity! --- Perhaps I am over-intense at certain junctures --- Your 'brand' of relaxation, however, is more than I can afford!

Respectfully
HP
 
Last edited:

eetech00

Joined Jun 8, 2013
3,858
Hello again HP:)

I was able to get your sim working.:cool:
See attached.

Some comments....

I changed each source voltage so that each output would reflect 25000 VAC RMS (35.352kv peak).:D
You can check this by displaying the source voltage in a plot window, then ctl-rht-click on the source voltage label.
I also changed the diodes to a HV version with an appropriate breakdown voltage. I actually changed the BV value of the diode spice model to a higher breakdown voltage (50kv) to get it to work.:cool:

Unzip the attached file into a folder, then open the .asc in LTspice.

eT
 

Attachments

Last edited:

cmartinez

Joined Jan 17, 2007
8,218
Hello again HP:)

I was able to get your sim working.:cool:
See attached.

Some comments....

I changed each source voltage so that each output would reflect 25000 VAC RMS (35.352v peak).:D
You can check this by displaying the source voltage in a plot window, then ctl-rht-click on the source voltage label.
I also changed the diodes to a HV version with an appropriate breakdown voltage. I actually changed the BV value of the diode spice model to a higher breakdown voltage (50kv) to get it to work.:cool:

Unzip the attached file into a folder, then open the .asc in LTspice.

eT
Pardon the intrusion, but I took the liberty of re-arranging the circuit's components as they were drawn by HP. And although the sim does slow down considerably (so much that I stopped it at 250µs), it doesn't report any errors. Here's my two cents.

Capture.JPG
 

Attachments

Thread Starter

Hypatia's Protege

Joined Mar 1, 2015
3,228
@eetech00 Thanks! Will check it out later this evening when I have a chance!:)

A couple advance questions:

I changed each source voltage so that each output would reflect 25000 VAC RMS (35.352v peak)
It seems this would imply an upper limit of ~ 141kV? -- Component integrity demands a (theoretical) upper limit of 100kV -- Am I to assume your amplitude adjustment (i.e. RMS vs Peak) is merely by way of demonstration? - Or am I missing something (again!:rolleyes:)? :)

I also changed the diodes to a HV version with an appropriate breakdown voltage.
Thank you! -- This is very interesting to me inasmuch as I (incorrectly?) assumed 'maximum stress' parameters were not enforced by LtSpice? -- Was my selection of rectifiers (Re: the simulation) a factor in the the simulation's instability?
FWIW my actual (i.e. physical) circuit employs GE-513s (PIV=45kV 'tho I seem to 'get away with' 50kV;))

I actually changed the BV value of the diode spice model to a higher breakdown voltage (50kv) to get it to work.:cool:
I really must find the time to learn the ways of modeling one's own components! --- A right handy skill, that!:) --- As a 'for instance' how would one go about stipulating recovery time???

Many, many thanks! -- Looking forward to studying the new revision!:):):)

Best regards
HP
 

WBahn

Joined Mar 31, 2012
29,978
Kind friends

Once again LtSpice has me beat:(:rolleyes:

Attached is a schematic file showing two series connected, independently excited, full wave 'voltage' doublers --The problem is that for non-zero values of R[bond] the results (i.e. E plots of 'Ea' and 'Ek') become increasingly erroneous with resistance until at ~ 500kΩ the simulation bombs altogether...

Any insights/workarounds will be greatly appreciated!:):):)

At your mercy
HP
What do you mean by Ea and Ek becoming increasing erroneous? Remember that your plots of each voltage are relative to ground and you only connect to ground at one point. What does a plot of (Ea-Ek) look like?
 

Aleph(0)

Joined Mar 14, 2015
597
It's easy to minimize everything on the basis that 'nothing really maters' -- howbeit I find said practice both cynical and straight up lazy!
Yea well, welcome to my world:(!

in your aversion to what passes for 'drama' you have renounced intellectual integrity!
I didn't know I had any to renounce! HP Plz don't confuse me whith ppl who wuss and turn other cheek by saying they don't like drama! You know how much I detest cowards and non confrontationalism! I just saying if you don't stay within scope then nothing gets done because endless branches of digression lead you away from initial intents and also you should look on hobbies with less passion so can be fun and relaxing instead of stressful! I sorry I angered you:(!

HP Circuit works fine if ground connection between doublers so like directly ground one side of meter and put mov to ground other side of meter in case meter open or removed. You say inelegant but I say right way cuz then there is solid ground and meter near ground for safety and no corona problem! That's my point why study on why simulator not jibe with real world for other circuit you won't use anyway?
 
Last edited:

WBahn

Joined Mar 31, 2012
29,978
That's my point why study on why simulator not jibe with real world for other circuit you won't use anyway?
There are very good and practical reasons for doing so. The more you know about why a tool behaves the way it does, especially when it appears to not be behaving the way you think it should, the better a user of that tool you will become. It's possible the tool is working fine and the problem is your interpretation of the results. If you figure out why you are interpreting the results incorrectly, you are less likely to make a similar mistake in the future in a situation where the resulting error is not significant enough to raise a warning flag but significant enough to cause a real problem. If the tool IS misbehaving, then understanding why it is misbehaving will put you in a much better position to avoid such situations in the future. Again, the next time it happens you might be unfortunate enough that the errors are significant enough to cause a real problem, but not significant enough to raise a warning flag.
 

eetech00

Joined Jun 8, 2013
3,858
@eetech00 Thanks! Will check it out later this evening when I have a chance!:)

A couple advance questions:

It seems this would imply an upper limit of ~ 141kV? -- Component integrity demands a (theoretical) upper limit of 100kV -- Am I to assume your amplitude adjustment (i.e. RMS vs Peak) is merely by way of demonstration? - Or am I missing something (again!:rolleyes:)? :)
Looking at your circuit and previous posts, you indicated that each half of the circuit is a voltage doubler, each with a resulting output of 50kv.
If that is accurate, then it follows that the output of each voltage source (I believe its actually a transformer coil) would need to be 25kv RMS.
It was initially set to 50kv Peak-Peak which =2*50Kv=100kv for each half, overstressing the already overstressed :Dcomponents. I merely changed each voltage source to output that value. In all honesty though, it would of helped if you posted the actual transformer schematic.:)

thank you! -- This is very interesting to me inasmuch as I (incorrectly?) assumed 'maximum stress' parameters were not enforced by LtSpice? -- Was my selection of rectifiers (Re: the simulation) a factor in the the simulation's instability?
FWIW my actual (i.e. physical) circuit employs GE-513s (PIV=45kV 'tho I seem to 'get away with' 50kV;))
Definitely...1N914's simply aren't designed for such high voltages. Wish you'd stated the diode P/N earlier.
I was scratching my head trying to figure out how this was working on your bencho_O....magic diodes!:D

I really must find the time to learn the ways of modeling one's own components! --- A right handy skill, that!:) --- As a 'for instance' how would one go about stipulating recovery time???

Many, many thanks! -- Looking forward to studying the new revision!:):):)

Best regards
HP
I don't recommend changing spice parameters unless you really understand the effect of doing so.
I didn't re-model the diode, I simply changed the BV value of a diode that was close to what I needed.
But now that I know the diode p/n I could model one if I had a data sheet.

good luck...:)

eT
 
Last edited:

Thread Starter

Hypatia's Protege

Joined Mar 1, 2015
3,228
you indicated that each half of the circuit is a voltage doubler
Correct, specifically a full-wave doubler

each voltage source (I believe its actually a transformer coil)
Correct -- the sources represent the secondaries of separate transformers - the primaries of each being driven by separate resonant Royer power oscillators...

If that is accurate, then it follows that the output of each voltage source (I believe its actually a transformer coil) would need to be 25kv RMS.
Therein lies our 'communications breakdown':D --- The requirement is that the amplitude is 25kV peak hence 50kV peak-to-peak --- Whereas 25kV RMS (sinusoidal) = 25kV*(2^0.5) ≈ 35.35534kV Peak70.71kV peak-to-peak = 70.71kV unloaded output EMF (each doubler) = farewell to the rectifiers and caps!:eek::D:D:D
FWIW A full wave 'doubler' of the topology under discussion always exhibits an (unloaded) output EMF equal to the peak-to-peak value of the input signal...

Definitely...1N914's simply aren't designed for such high voltages. Wish you'd stated the diode P/N earlier.
So LtSpice now enforces standard semiconductor junction PIV (in addition to 'on-label' Zenners)? --- In the past I've attempted to model reverse breakdown of standard diodes without success - thus it seems they've improved their library 'models'? --- Good deal!:)


I don't recommend changing spice parameters unless you really understand the effect of doing so.
Sounds like good advice! -- as regards my LtSpice proficiency - seems I know just enough to be 'dangerous':oops::eek::D

Many thanks!:)
HP
 
Last edited:

Thread Starter

Hypatia's Protege

Joined Mar 1, 2015
3,228
What do you mean by Ea and Ek becoming increasing erroneous? Remember that your plots of each voltage are relative to ground and you only connect to ground at one point. What does a plot of (Ea-Ek) look like?
It varies run-to-run but the below attached screen capture is typical -- FWIW: After ~20 seconds the simulation often (but not always) aborts with one of several error messages, the following being most common:
Lt Spice error message said:
Singular matrix: Check nodes d2#int1 and d4#int1
Iteration #9
Here's a typical 'erroneous' plot: (Note: This post continues below the image:))
BadSim.png



When the 'center point' is connected to ground via low or zero resistance the result 'jibes' with reality
To wit: (continued)
GoodSim.png

Close inspection correctly shows the ripple to be ~ 300v (each rail) and, as expected, twice the excitation frequency:)

Note: For your convenience I have attached the files for the properly simulated schematic (LtSpiceWorks.asc) and the erroneously simulated schematic (LtSpiceFails.asc)

Many thanks for your interest/assistance!:):):)

Best regards
HP
 

Attachments

Last edited:

Thread Starter

Hypatia's Protege

Joined Mar 1, 2015
3,228
Pardon the intrusion, but I took the liberty of re-arranging the circuit's components as they were drawn by HP. And although the sim does slow down considerably (so much that I stopped it at 250µs), it doesn't report any errors. Here's my two cents.

View attachment 94218
Looks as if LtSpice got hold of some 'bad hooch"!:eek::eek::eek: The anode EMF (Ea) starts at -20MV then haltingly soars to 130MV (fully 1,300 times the circuit's unloaded upper limit!) on the other hand the cathode EMF (Ek) is basically 'flat lined' with occasional meteoric 'spikes'!:eek:o_O

In actual operation Ea and Ek swiftly and cleanly attain their loaded operating EMFs (a scant +50kV and -50kV respectively) thence exhibit a highly favorable (and expected) ~ 300V, 60kHz sawtooth ripple...

While I'm certain there's a perfectly reasonable explanation of LtSpice's Um... 'deviance' - I must say I'm eager to hear it!:D

FWIW -- the ".asc" files attached to post #32 provide examples of functional and dysfunctional simulations...:)

Many thanks for your time, effort and assistance!!!:):):)

Best regards
HP
 
Last edited:

Thread Starter

Hypatia's Protege

Joined Mar 1, 2015
3,228
Hello again HP:)

I was able to get your sim working.:cool:
See attached.

Some comments....

I changed each source voltage so that each output would reflect 25000 VAC RMS (35.352kv peak).:D
You can check this by displaying the source voltage in a plot window, then ctl-rht-click on the source voltage label.
I also changed the diodes to a HV version with an appropriate breakdown voltage. I actually changed the BV value of the diode spice model to a higher breakdown voltage (50kv) to get it to work.:cool:

Unzip the attached file into a folder, then open the .asc in LTspice.
It seems the display format 'rounds' any value of Ea and Ek having an absolute value greater than ~ 50Kv to +/-50Kv (respectively) please note in the image below that an input EMF of 20kV peak correctly displays an output of ~40kV (loaded):
(Post continued below image)
plot2.png





Below I have stipulated a peak input value of 100kV and with bogus results:rolleyes: (Continued)

plot5.png
This is really too bad of Linear Technologies!:mad: -- It would be bad enough if it merely truncated the display - but it actually adjusts the values as per truncation (as evinced by the V[Ea,Ek] field)!!!:eek::mad: --- I have to consider whether LtSpice is worthy of "addition to my toolbox" -- Guess it's a matter of 'Ya git whatcha pay for' --- trite but true!!!:mad::mad::mad:

Maintaining an open (if jaded) mind...

Sincere thanks for your time, effort and assistance!!!:)
HP
 
Last edited:

Thread Starter

Hypatia's Protege

Joined Mar 1, 2015
3,228
Yea well, welcome to my world:(!
Are we eating worms now?:rolleyes:

I just saying if you don't stay within scope then nothing gets done because endless branches of digression lead you away from initial intents
Agreed - as far as it goes -- that said, learning to use (and, hence, trust) one's tools is hardly unreasonable digression...

you should look on hobbies with less passion so can be fun and relaxing instead of stressful!
Good deal! --- If you know of a means to that end (sans 'chemical assistance') please advise!:):cool:

You say inelegant but I say right way cuz then there is solid ground and meter near ground for safety and no corona problem!
I said it was inelegant - not 'wrong' --- IOW I agree with the objective merits of your assessment...:)

That's my point why study on why simulator not jibe with real world for other circuit you won't use anyway?
If I'm to seriously use LtSpice I must aquatint myself with its proper use/limitations - this includes resolving/understanding unexpected performance...

All the best
HP

PS Thanks for reminding me of the verb 'Jibe' --- 'Informals' are handy for their verbal 'flexibility':cool::cool::cool:
 
Last edited:

Thread Starter

Hypatia's Protege

Joined Mar 1, 2015
3,228
There are very good and practical reasons for doing so. The more you know about why a tool behaves the way it does, especially when it appears to not be behaving the way you think it should, the better a user of that tool you will become. It's possible the tool is working fine and the problem is your interpretation of the results. If you figure out why you are interpreting the results incorrectly, you are less likely to make a similar mistake in the future in a situation where the resulting error is not significant enough to raise a warning flag but significant enough to cause a real problem. If the tool IS misbehaving, then understanding why it is misbehaving will put you in a much better position to avoid such situations in the future. Again, the next time it happens you might be unfortunate enough that the errors are significant enough to cause a real problem, but not significant enough to raise a warning flag.
+!:cool::cool::cool:
 

eetech00

Joined Jun 8, 2013
3,858
It seems the display format 'rounds' any value of Ea and Ek having an absolute value greater than ~ 50Kv to +/-50Kv (respectively) please note in the image below that an input EMF of 20kV peak correctly displays an output of ~40kV (loaded):
Just "seems" that way.:)
You can change the x/y ranges in the plot window by clicking on the grid value along the edge of the grid. You can also zoom in to the area of interest and move the cursor along the horizontal axis for approximate measurements.

For precise measurements, you should learn to use .meas statements that can provide "point in time", and other measurements that are displayed in the "error log" (log not really just for errors).

I can't emphasize enough how valuable a tool LTspice really is.:cool:
If you have more specific questions, join the LTspice Group at yahoo.com. Its an extremely valuable resource.

BTW,
If your looking for simulation results that reflect actual bench test results, then you need device models that closely reflect the behavior of the actual devices used in the test. The devices used in your simulation do not.:(
For the simulation "The quality of output will be determined by the quality of input".:D

eT
 

eetech00

Joined Jun 8, 2013
3,858
This is really too bad of Linear Technologies!:mad: -- It would be bad enough if it merely truncated the display - but it actually adjusts the values as per truncation (as evinced by the V[Ea,Ek] field)!!!:eek::mad: --- I have to consider whether LtSpice is worthy of "addition to my toolbox" -- Guess it's a matter of 'Ya git whatcha pay for' --- trite but true!!!:mad::mad::mad:

Maintaining an open (if jaded) mind...

Sincere thanks for your time, effort and assistance!!!:)
HP
:eek:This is a result of inexperience with the LTspice simulator. :rolleyes:
It does have a significant learning curve.

Try the LTspice Group at yahoo.com. Its an extremely valuable resource.
 

cmartinez

Joined Jan 17, 2007
8,218
While I'm certain there's a perfectly reasonable explanation of LtSpice's Um... 'deviance' - I must say I'm eager to hear it!:D
Well, I'm no LTspice expert, nor am I one of the people that helped code it. But in my limited experience, the sort of results you've been getting are usually the consequence of imperfect numerical analysis techniques. They're caused, for instance, by iterations in which divisions are performed by a number that is too close to zero, or by the use of floating point variables in the code that do not have enough resolution to obtain an accurate result.

It might be, for instance (and most probably), that the software is internally using 32 bit numbers throughout the code, and that just won't cut it when performing calculations in extreme situations, like your circuit seems to pose.
Maybe (and this is mere speculation from my part) if the program were to be re-compiled to run in a 64 bit environment (with its internal variables also updated to make use of the whole spectrum of the hardware available to the OS) the result of your simulation would be very different.
 

Thread Starter

Hypatia's Protege

Joined Mar 1, 2015
3,228
I can't emphasize enough how valuable a tool LTspice really is.:cool:
For the simulation "The quality of output will be determined by the quality of input".:D
This is a result of inexperience with the LTspice simulator. :rolleyes:
It does have a significant learning curve.
Fair enough!:) I won't dismiss LtSpice based upon the outcome of my 'dabbling':oops: -- It obviously has a lot going for it to have garnered the professional following it enjoys:cool::)

Again, many thanks for your assistance!:)

Best regards
HP
 
Top