LtSpice and split supplies :-(

Thread Starter

Hypatia's Protege

Joined Mar 1, 2015
3,228
I have been a Yahoo Groups member forever. I still receive email digests from several groups I track, but haven't explicitly logged in for a year or so. I just tried it, and it asked for the mobile number, which I left blank. It said my login "failed", but yet I can still access the groups I could previously...
In my case any attempt to access the group results in a monition to the effect that 'You must be a member....' --- whereas joining the group requires that I have a Yahoo Acct -- which leads to firm insistence upon a 'mobile number' --- FWIW (likely very little) I blame the legions of Facebook minions whose mindless acquiescence enabled/mainstreamed abolition of online privacy:rolleyes:

Best regards
HP
 

Thread Starter

Hypatia's Protege

Joined Mar 1, 2015
3,228
@MikeML

Attached is a 1V to 1MV sweep of the library model of diode type 1n914 -- I see no difference between this and the "default diode" sweep attached to your post (#57)
In both cases the diode is functioning as a 'straight' conductor -- A fact I find curious inasmuch as it was my understanding that LtSpice did not support reverse breakdown (with the exception of 'on label' Zeners) --AND-- the facts that:

1) The curve, ('zoom' as I might) is "drop dead linear" over the entire sweep
and
2) This diode model (1N914), despite the limitations of its real-world counterpart, functions correctly as a rectifier in various EHT simulations...


Nor is resolution of the paradox to be found in any possibility that the leakage current is 'over ballasted' by R1 & R2 -- Inasmuch as a tenfold reduction in the value of each resistor results in a point-for-point tenfold reduction in V(b).

Ok, obviously, I'm guilty of a glaring (and, I might add, rather embarrassing) oversight --- So... let's hear about it!?:oops::oops::oops:

Best regards and many, many thanks!:):):)
HP

1n914Sweep.png
 

Thread Starter

Hypatia's Protege

Joined Mar 1, 2015
3,228
Addendum to post #62

Further experimentation with the simulation indicates that both the 'default diode' and the 1N914 exhibit a maximum (i.e. unballasted) leakage current of exactly 1uA per MV --- It just gets weirder and weirder:confused::rolleyes:

Best regards
HP
 

eetech00

Joined Jun 8, 2013
3,856
Addendum to post #62

Further experimentation with the simulation indicates that both the 'default diode' and the 1N914 exhibit a maximum (i.e. unballasted) leakage current of exactly 1uA per MV --- It just gets weirder and weirder:confused::rolleyes:

Best regards
HP
Without the BV param specified (and it isn't in the default 1N914 model) it is infinite.
If you add BV=100 it will breakdown at 100v. I copied the model params to the schematic and added the BV param. See attached.
 

Attachments

Aleph(0)

Joined Mar 14, 2015
597
It has been my experience that the default diode behaves quite unexpectedly - even in basic circuits:confused:
HP you are right! I draft half wave rectifier from 100vp-p sine and like you say _default diode_ and 100 ohm resistor for load. Diode conduct full cycle like wire but if use four in bridge they work! I say default diode work like just wire in half wave rectifier and swept circuit because properties say breakdown voltage =0 and it won't let me edit values in properties dialog:mad:! I can't say why default work in bridge or why 1n914 props say breakdown voltage = 70v despite it blocks any voltage approaching to infinity!
Further experimentation with the simulation indicates that both the 'default diode' and the 1N914 exhibit a maximum (i.e. unballasted) leakage current of exactly 1uA per MV --- It just gets weirder and weirder:confused::rolleyes:
HP My result says default diode resistance is 0 ohms both polarity so infinite maximum leakage current but you right about 1n914 it is just 1T ohm resistor when under reverse bias!

HP I say virtual lab has use when saves time and money which it can't do when unpredictable and just wrong! You said need confidence so I say time to throw _land of make believe_ over for a sunny piece of real world! To say your fave expression _enough is enough_:rolleyes:!
 

Aleph(0)

Joined Mar 14, 2015
597
This is a result of inexperience with the LTspice simulator. :rolleyes:
It does have a significant learning curve.
Eetech I mean no disrespect to you but that not excuse for ltspice give wrong results! I agree that HP's ltspice circuit not identical to real world circuit because she not draft transformers and used wrong diodes but results don't reflect on those variables and ltspice also give poor results for basic circuits with everything according to real world plan! I say alerts warnings and error halts are totally ok but results different from real world for exact same setup is just bogus!
Eetech I want you to know I not ripping on you at all! I think its lovely how aac community help each other out:)! I'm just disgusted that we have to make excuses for and put up with crap software cuz nothing else available:mad:
 

MikeML

Joined Oct 2, 2009
5,444
Eetech I mean no disrespect to you but that not excuse for ltspice give wrong results! I agree that HP's ltspice circuit not identical to real world circuit because she not draft transformers and used wrong diodes but results don't reflect on those variables and ltspice also give poor results for basic circuits with everything according to real world plan...
And you would be just plain wrong. Spice produces the results that are implied by the simplified circuit that you create. Spice models what you put in. If you ignore parasitic inductances and capacitances, node leakages, breakdown voltages, etc, then there should be no surprise that the results deviate from a real circuit, which has all of the parsitics, and non-ideal behaviors.

Also note the lack of experience with how diodes are actually modeled exhibited in this thread. The "computationally light" linear diode model that I linked to in the help file is appropriated for normal low-voltage, low-frequency simulation. The fact that it is not appropriated to be used at 50kV piv is a manifestation of the lack of experience of the user; not an indictment of Spice modeling..
 

Thread Starter

Hypatia's Protege

Joined Mar 1, 2015
3,228
I say default diode work like just wire in half wave rectifier and swept circuit because properties say breakdown voltage =0
Good observation!:)

I can't say why default work in bridge or why 1n914 props say breakdown voltage = 70v despite it blocks any voltage approaching to infinity!
Indeed! Selective enforcement/modeling of parameters seems to be part of the 'spice' experience:rolleyes:

HP My result says default diode resistance is 0 ohms both polarity so infinite maximum leakage current but you right about 1n914 it is just 1T ohm resistor when under reverse bias!
Oooops! You're correct! A single 'default diode' is merely a (perfect) conductor (PIV=0) --- while two or more connected in series (as in a bridge rectifier arrangment) behave as perfect (would you believe) diodes (PIV=∞):confused::confused::confused: --- I don't make much of this inasmuch as 0V PIV 'junctions' represent an 'indeterminate' concept -- though one wonders why such a 'component' is offered and why, as you point out, its parameters are non-editable in its properties window?:confused:

HP I say virtual lab has use when saves time and money which it can't do when unpredictable and just wrong!
I say alerts warnings and error halts are totally ok but results different from real world for exact same setup is just bogus!
--EMPHASIS ADDED--

Aleph, I agree with you all the way - IF - and this a big 'IF' -- the simulated and 'bread boarded' circuits are indeed parallel! - When acquiring a new 'language' it is quite common for 'utterance' and intent to be far out of sync, as it were... -- Point being - let's understand this prior to making a judgement call...K?:cool:

I say 100 farad capacitor for 100kv not practical for real circuit:confused:!
I believe you are confusing denominations with units -- A forgivable misstep considering the similarity of symbols -- 100fF:)) On the other hand, a 100 Farad capacitor charged to 100kV (i.e. 500 GJ ≈140,000 KWH) -- Would cost ~ $15,000 merely to charge it! Which being but a pittance by comparison with the 'cost' of negligence in regard to same!!!:eek::eek::eek:

All the best
HP:)
 

Thread Starter

Hypatia's Protege

Joined Mar 1, 2015
3,228
And you would be just plain wrong. Spice produces the results that are implied by the simplified circuit that you create. Spice models what you put in. If you ignore parasitic inductances and capacitances, node leakages, breakdown voltages, etc, then there should be no surprise that the results deviate from a real circuit, which has all of the parsitics, and non-ideal behaviors.

Also note the lack of experience with how diodes are actually modeled exhibited in this thread. The "computationally light" linear diode model that I linked to in the help file is appropriated for normal low-voltage, low-frequency simulation. The fact that it is not appropriated to be used at 50kV piv is a manifestation of the lack of experience of the user; not an indictment of Spice modeling..
Whoa! -- Speaking for myself I'm content to merely learn 'spice' such that I may use it with some degree of confidence --- that said, I feel Aleph has a point - if, for instance, a diode specified with a PIV of 70 Volts works 'just dandy' as a rectifier at 100kV - the simulator is 'dropping the ball' - plain and simple! -- unless, that is, fidelity to reality is an inconvenience the developers can't afford and hence rationalize away!:eek: Clearly, the physical properties/behavior of components - as opposed to how spice developers choose to model same -- are properly the concern of the user... --- Granting that a perfect model of reality is an impossibility, I nonetheless, fail to see any need to begrudge prospective users' ignorance of the developer's artificial, and, so it seems, largely arbitrary, construct...

I did not start this thread to debate the merits of spice - though I freely own the blame for the thread taking that direction (with reference to my 'whining' in post #34):oops: -- @Aleph(0) @MikeML @eetech00 -- and all other interested parties -- Please accept my sincere apologies and know that I am genuinely eager to learn!:):):)

Many sincere thanks to all respondents!:)
HP
 
Last edited:

Thread Starter

Hypatia's Protege

Joined Mar 1, 2015
3,228
If you ignore parasitic inductances and capacitances, node leakages, breakdown voltages, etc,
Are 'ballpark' parasitic reactors/reactances included in the models of specific components?

The problem of the first circuit (first post) can be solved easily. Add the circuit .opt cshunt=100f
Could you expand upon that please? What is the proposed capacitance paralleling?

Without the BV param specified (and it isn't in the default 1N914 model) it is infinite.
If you add BV=100 it will breakdown at 100v. I copied the model params to the schematic and added the BV param. See attached.
Thanks! -- I'll have a look at this and post questions/comments later this evening/Wed morning:)

In spice (and in LTspice) 100f=100*10^-15=1*10^-13=0.1pF
The suffix "f" is "femto".
Please don't judge Aleph too harshly -- She comes by her 'temperament' naturally:oops::)

Many thanks all around!:)
HP
 
Last edited:

MikeML

Joined Oct 2, 2009
5,444
Are 'ballpark' parasitic reactors/reactances included in the models of specific components?
...
For diodes, passive components, simple semiconductor devices:

Resistances of the bulk Silicon are
Junction capacitances are
Inductance of the leads are not
Package capacitance, resistance and inductance are not
Shunt capacitances (layout/PCB traces) from pins to a "ground plane" are not
As we have found out, Breakdown voltages are not.
Static protection on inputs are not


Basically, the characteristics that are primarily modeled are those you would measure in the lab with a curve-tracer or a voltmeter/ammeter...
 

Bordodynov

Joined May 20, 2015
3,177
LTspice may have difficulty in calculating certain schemes (with default settings).
In this case, you must help LTspice (unless you're a masochist). Well help more small capacitors. In reality, there are always the parasitic capacitance in the circuit. If you use the option "cshunt", then you are to each point of the circuit connecting the capacitor to the specified value. The alternative is to be connected to the critical point of the capacitor. For myself, I have developed a library of parasitic elements. They have a simple symbol (point, line). They are almost invisible on the electronic circuit, and they do not clutter up the scheme.
 

Thread Starter

Hypatia's Protege

Joined Mar 1, 2015
3,228
Without the BV param specified (and it isn't in the default 1N914 model) it is infinite.
If you add BV=100 it will breakdown at 100v. I copied the model params to the schematic and added the BV param. See attached.
Thank you for your assistance and patience!!!:)

Having run the simulation (please see the attached image) I have two questions:

--Why is there no drop across R1 subsequent to breakdown? --- Got it!:) There is a drop! - I overlooked the low resistance value:oops:

So -- The outstanding question:
--Why are the 'curves' perfectly linear following breakdown?

If, as I suspect, I'm guilty of a stupid mistake or oversight please don't hesitate to tell me so!:) --- Receipt of epithets is a trivial price for knowledge!:cool:

Many advance thanks!:):):)
HP
 
Last edited:

eetech00

Joined Jun 8, 2013
3,856
@HP

:eek:um...please remove model params....may cause copyright issues.:cool:

Instead, copy them to a file, then use an .include statement on the schematic.
 

Thread Starter

Hypatia's Protege

Joined Mar 1, 2015
3,228
@HP

:eek:um...please remove model params....may cause copyright issues.:cool:

Instead, copy them to a file, then use an .include statement on the schematic.
Done!:) The image was a screen capture -- So I've merely deleted the attachment for the nonce...

.please remove model params....may cause copyright issues
I'll take your word for it -- It didn't look too 'proprietary' to me????:confused:

Instead, copy them to a file, then use an .include statement on the schematic.
What is the procedure for affixing 'statements' to the schematic - or, come to that, editing semiconductor properties? --- I seem to be experiencing the same difficulty as @Aleph(0) in that regard (i.e. the active component's drop-downs are 'read only'):confused:

Best regards
HP:)
 

eetech00

Joined Jun 8, 2013
3,856
Done!:) The image was a screen capture -- So I've merely deleted the attachment for the nonce...
OK...cool.

I'll take your word for it -- It didn't look too 'proprietary' to me????:confused:
I don't wanna go there...;)
But it did have a manufacturer's name on it....that's reason enough for me.

What is the procedure for affixing 'statements' to the schematic - or, come to that, editing semiconductor properties? --- I seem to be experiencing the same difficulty as @Aleph(0) in that regard (i.e. the active component's drop-downs are 'read only'):confused:
Create a text file (example: mydiode.txt) in the same folder as the schematic
In LTspice, highlight the entry in the component dropdown, then copy (ctl-c) and paste (ctl-v) into the text file.
The entry copied into the file should begin with ".model".

Now tell LTspice to "include" the file in the simulation. In LTspice:
Click in the schematic area, then type the letter "t" to begin a text mode dialog.
Select "Spice Directive" radio button, click in the text entry pane, then type the following:

.include mydiode.txt

then click OK.

The include statement should now appear on the schematic.

You can also add plain text in the same manner. Just select the "comment" radio button instead.
Sometime I'll add a directive as plain text, and toggle it between being text or directive as needed.

eT
 
Top