The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Analog Design >> Harmonic Oscillator with ideal Opamp models
https://designers-guide.org/forum/YaBB.pl?num=1322576364

Message started by buddypoor on Nov 29^th, 2011, 6:19am

Title: Harmonic Oscillator with ideal Opamp models
Post by buddypoor on Nov 29^th, 2011, 6:19am

Hello, I have a question to all who are involved in electronic circuits.

However, the question is not techical but more or less linguistic and concerns the general speech comprehension.

Problem description: There are some configurations (I avoid the term „electronic circuits“) that exhibit oscillatory properties during transient simulations (in fact: they show self-sustained sinusoidal oscillations). However, only if IDEAL opamps are used. Therefore, only simulations can reveal threse properties.

As soon as real opamp models are used (at least one pole in the open-loop frequency response) there are no oscillations.
Instead, the amplifier output saturates immediately, because the mentioned amplifier pole is shifted to the right half of the s- plane (RHP).

The problem touches the validity of the well-known oscillation condition (Barkhausen) - and the question is simply:

Based on the general speech comprehension - should such a special configuration be called „circuit that is able to oscillate"? Even, if this is the case for ideal (artificial) models only?

Thank you
buddypoor (LvW)

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by loose-electron on Nov 29^th, 2011, 3:50pm

Could you better explain or illustrate the question?

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by buddypoor on Nov 30^th, 2011, 1:39am

loose-electron wrote on Nov 29^th, 2011, 3:50pm:

Could you better explain or illustrate the question?

I am afraid that I have expressed myself not clear enough.
Therefore, I will reformulate my question:

1.) Assume that there is an electronic circuit (hardware) that is not able to oscillate - in spite of unity loop gain at one frequency f=fo only. This is neither a surprise nor a failure of Barkhausen’s criterion because this rule is only a necessary one.
2.) Surprisingly, if this circuit is transferred to a simulation program and if an ideal opamp model (gain not frequency dependent) is used, the output shows a sinusoidal signal having a frequency fo. (By the way, I can explain this fact - but that is not my question).

Now the question:
How should I describe the property of that specific combination of passive and active parts with respect to the oscillation condition?

Is that circuit an oscillator in accordance with Barkhausen’s criterion: Yes or no?

Thank you

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by loose-electron on Nov 30^th, 2011, 11:10am

If I understand the question correctly:

Oscillation is conditional on two things:

Additive gain of a loop.
Phase/frequency characteristics of that loop.

You can do a behavioral model defined in those terms.

Does that get you an answer?

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by loose-electron on Nov 30^th, 2011, 11:14am

You may want to look at this:
http://web.mit.edu/klund/www/weblatex/node4.html

To me, this all goes back to control systems and Blacks Law.

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by Alexandar on Dec 1^st, 2011, 1:31am

The question is not very clear to me either, but I'll try to write something related to it.

If you talk about Barkhausen' criterion, it implies you're talking about harmonic oscillators. Not first order oscillators.

In 'steady state' condition, its poles don't move a lot, but in startup it does. This nonlinear behavior provides the startup. If you have ideal elements, which do not give any nonlinearity, they might have a problem oscillating, or startup vice verse. Please note that by 'movement of poles' I am not talking about (phase) noise, but deterministic movement of the poles.

The moving poles 'stuff' becomes even more clear when you talk about first order oscillators, where the poles exhibit a walk down the root locus in time. They stick on the real axes for the linear part, and only when the oscillator switches the poles move toward the imaginary axis, and at that point go up/down (becoming complex).

I suggest using phase planes (nonlinear dynamics theory), theory on moving poles, etc, to get deeper understanding of the problems you see.

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by buddypoor on Dec 1^st, 2011, 3:15am

loose-electron wrote on Nov 30^th, 2011, 11:14am:

You may want to look at this:
http://web.mit.edu/klund/www/weblatex/node4.html
To me, this all goes back to control systems and Blacks Law.

Loose-electron, thank you for this link to a "funny" contribution from K.H. Lundberg (2002). I know this text since several years - and I also have given to Lundberg some comments and corrections.
But as it seems he sees no necessity to modify his conclusions like the first and the last sentence:
"The Barkhausen Stability Criterion is simple, intuitive, and wrong."
"Down with Barkhausen"
Even the headline is wrong: Barkhausen never has formulated a "stability criterion".
Perhaps he likes such provocative formulations - even when the are wrong.
To me he has (a) never read the original text from Barkhausen and (b) never heard about the difference between "necessary" and "sufficient".
None of these terms is even mentioned in his article.
In the mean time (within the last 3...4 years) a discussion took place in the magazine "Analog Integrated Circuits and Signal Processing" dealing with the validity of Barkhausen's condition (based on some so called "counter examples" introduced by V. Singh). For my opinion, there is a wide agreement that Barkhausen has formulated a necessary condition only (I can confirm this as I have his book on my desk).
Besides this, of course Black's formula is the basis for all calculations of feedback systems. And - as I have mentioned - I have no problems to explain the differences between the behaviour of both alternatives (ideal vs. real opamp model). My only concern is: Does it make sense to investigate per simulation a "circuit" based on an ideal opamp that never can be build - and that behaves completely different if real amplifiers are used?
Thank you

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by buddypoor on Dec 1^st, 2011, 3:20am

Lex wrote on Dec 1^st, 2011, 1:31am:

................
I suggest using phase planes (nonlinear dynamics theory), theory on moving poles, etc, to get deeper understanding of the problems you see.

Hi Alexander, thanks for replying.
However, as mentioned in my reply above you see that I have no technical problem. I can explain and verify the behaviour of both alternatives (including pole movements).
Nevertheless, thank you again.
Regards

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by raja.cedt on Dec 1^st, 2011, 11:02am

hello,
@ alexder: can you please tell me where to study about moving poles (i donno any thing about moving poles)

@ buddypoor: hello buddypoor, you have written "Surprisingly, if this circuit is transferred to a simulation program and if an ideal opamp model (gain not frequency dependent) is used, the output shows a sinusoidal signal having a frequency fo", is really worked, i guess you have to add some passive ckt which makes gain drops at least and infact with gain dependent amplier also works because after all you are having some gain drop in the opamp so you will reduce the phase shift required by the passive network. Correct me if i am wrong.

do you have the book which decribes counter examples? please tell me the exact title so that i can read.

Thanks,
Raj.

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by buddypoor on Dec 1^st, 2011, 11:52am

raja.cedt wrote on Dec 1^st, 2011, 11:02am:

...........................
i guess you have to add some passive ckt which makes gain drops at least and infact with gain dependent amplier also works because after all you are having some gain drop in the opamp so you will reduce the phase shift required by the passive network. Correct me if i am wrong.

do you have the book which decribes counter examples? please tell me the exact title so that i can read.
Thanks,
Raj.

What do you mean with "you have to add...".
In contrary, I do not want to add some circuitry to let the gain drop because I want to demonstrate the influence of an ideal opamp model.
Of course, as I have mentioned already, the real model behaves differently.

Regarding "counter-examples" I only can give you a reference to some magazine articles. Have you access to "Analog integrated circuits and signal processing" ?

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by raja.cedt on Dec 1^st, 2011, 11:53am

Hello buddypoor,
what i mean to say is when you have ideal opamp without any frequency dependent gain, there should be some netwrok in the loop which reduces the loopgain to 1 at some frequency.

regarding articles, please send me if you can otherwise please tell me the name so that i can find some otherway to download.

Thanks,
Raj.

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by loose-electron on Dec 1^st, 2011, 2:13pm

buddypoor wrote on Dec 1^st, 2011, 3:15am:

My only concern is: Does it make sense to investigate per simulation a "circuit" based on an ideal opamp that never can be build - and that behaves completely different if real amplifiers are used?
Thank you

Completely different? Yes and No.

Use of ideal models are a simplified method to gain a better understanding of something.

We all use those all the time.

Also, an ideal op-amp can be dropped into the middle of something, and then add gain limitations, and explicit phase-gain properties outside the ideal amplifier.

That lumped model allows investigation into the separate parameters of the model.

i.e. adjust gain, phase characteristics without a need to redesign an entire device down at the transistor level.

Depends on what you are trying to achieve? Understanding the characteristics of something or building a circuit to plug in and turn on?

We all do both in this
business/profession/area of study/obsession
(pick one) :D
to varying degrees.

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by buddypoor on Dec 2^nd, 2011, 1:33am

loose-electron wrote on Dec 1^st, 2011, 2:13pm:

Completely different? Yes and No.
Use of ideal models are a simplified method to gain a better understanding of something.
We all use those all the time.
Also, an ideal op-amp can be dropped into the middle of something, and then add gain limitations, and explicit phase-gain properties outside the ideal amplifier.
That lumped model allows investigation into the separate parameters of the model.
i.e. adjust gain, phase characteristics without a need to redesign an entire device down at the transistor level.
Depends on what you are trying to achieve? Understanding the characteristics of something or building a circuit to plug in and turn on?

Hi Loose-electron.
In principle, I agree with all of the above.
But I think my description of a specific case (my first posting opening this thread) can serve as an example that your explantions/justifications of idealized models do not always apply.
As I have reported, there are circuits that show a behaviour that is strongly dependent on the amplifier model used for simulation (ideal or real). It is obvious that this cannot lead to a "better understanding" of the circuit and it's function.
On the other hand, if one is able to understand the reason for the observed phenomena, this certainly will enlarge the knowledge of system theory and related areas.
Perhaps it's useful to give you a very simple example:
Try to simulate an IDEAL opamp with positive resistive feedback (10k/1k).
Of course, such a circuit will not work as an amplifier. However, all simulations (OP, AC, DC, TRAN) will result in a stable inverting amplifier with a gain of 20 dB. Does this lead to a better understanding of amplifiers?
And it is a fact, that the simulator did not make any error at all. The result is correct!

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by boe on Dec 2^nd, 2011, 2:04am

buddypoor wrote on Dec 2^nd, 2011, 1:33am:

...
Perhaps it's useful to give you a very simple example:
Try to simulate an IDEAL opamp with positive resistive feedback (10k/1k).
Of course, such a circuit will not work as an amplifier. However, all simulations (OP, AC, DC, TRAN) will result in a stable inverting amplifier with a gain of 20 dB. Does this lead to a better understanding of amplifiers?
And it is a fact, that the simulator did not make any error at all. The result is correct!

Interesting example. And the result is correct because an IDEAL opamp model contains a control loop....

- B O E

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by Garrett.Neaves on Dec 2^nd, 2011, 6:48am

buddypoor wrote on Dec 2^nd, 2011, 1:33am:

Perhaps it's useful to give you a very simple example:
Try to simulate an IDEAL opamp with positive resistive feedback (10k/1k).
Of course, such a circuit will not work as an amplifier. However, all simulations (OP, AC, DC, TRAN) will result in a stable inverting amplifier with a gain of 20 dB. Does this lead to a better understanding of amplifiers?
And it is a fact, that the simulator did not make any error at all. The result is correct!

This example is useful in understanding the question in your initial posting.

I would not refer to this ideal opamp with positive feedback circuit as an amplifier because it is only functioning as an amplifier because it is excessively idealized.

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by boe on Dec 2^nd, 2011, 7:08am

Garrett.Neaves wrote on Dec 2^nd, 2011, 6:48am:

...
I would not refer to this ideal opamp with positive feedback circuit as an amplifier because it is only functioning as an amplifier because it is excessively idealized.

I agree. I would also consider an opamp model that is stable with positive feedback only as excessively idealized.

- B O E

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by buddypoor on Dec 2^nd, 2011, 7:48am

Garrett.Neaves wrote on Dec 2^nd, 2011, 6:48am:

This example is useful in understanding the question in your initial posting.
I would not refer to this ideal opamp with positive feedback circuit as an amplifier because it is only functioning as an amplifier because it is excessively idealized.

Thanks for replying and - yes - of course I agree (more or less).
On the other hand, the degree of simplification is the same for interchanged input nodes leading to the classical circuit configuration, which is widely used.
Thus, the "excessive idealization" alone cannot be used as an argument.
Rather, I think it is the knowledge we have (contrary to the computer resp. the simulation engine) about the conditions for a stable dc operating point.
Are there other (additional) aspects?

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by boe on Dec 2^nd, 2011, 8:11am

buddypoor wrote on Dec 2^nd, 2011, 7:48am:

...
On the other hand, the degree of simplification is the same for interchanged input nodes leading to the classical circuit configuration, which is widely used.
Thus, the "excessive idealization" alone cannot be used as an argument.

However, this simplification is only allowed if overall feedback is negative.

[Added:]
Quote:

Rather, I think it is the knowledge we have (contrary to the computer resp. the simulation engine) about the conditions for a stable dc operating point.

IMO it is because we use the model also it is no longer valid.
- B O E

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by Garrett.Neaves on Dec 2^nd, 2011, 8:19am

buddypoor wrote on Dec 2^nd, 2011, 7:48am:

... the degree of simplification is the same for interchanged input nodes leading to the classical circuit configuration, which is widely used.
Thus, the "excessive idealization" alone cannot be used as an argument.
Rather, I think it is the knowledge we have (contrary to the computer resp. the simulation engine) about the conditions for a stable dc operating point.

I think we are seeing the same thing from slightly different points of view. From my point of view, when a model becomes idealized to the point that it gives fundamentally contrary behavior then I would say that this is a sufficient condition for describing it as "excessively idealized".

The configuration of an ideal opamp with resistive divider negative feedback does not give fundamentally contrary behavior and therefore is not necessarily " excessively simplified. In fact, it is often a very useful model.

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by buddypoor on Dec 2^nd, 2011, 9:19am

Garrett.Neaves wrote on Dec 2^nd, 2011, 8:19am:

This sounds reasonable - but the problem is: How to know in advance (and for unknown circuits) if the simplification is "excessive" (means: not allowed and not applicable) or not?
But thanks to this discussion we are approaching now the core of the question in my first posting.

B O E: I agree, because your condition "dc negative feedback" is identical to my condition regarding a stable dc operating point.

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by Garrett.Neaves on Dec 2^nd, 2011, 10:16am

buddypoor wrote on Dec 2^nd, 2011, 9:19am:

but the problem is: How to know in advance (and for unknown circuits) if the simplification is "excessive" (means: not allowed and not applicable) or not?

I like the kind of thinking represented by this question.

I think that there is no general way to know in advance.

In specific cases, we can see a warning as soon as we see that the ideal modelling result is contrary to what we think we know about physical circuits. Then, we can do some investigation by, for example, replacing the ideal models of the most complicated components by ones known to reasonably model physical components. Such as replace, the ideal op amp with a known physically representative model.

In general, we need to always ask if our modelling results are reasonable by testing them against something.

Of course, you already do that. I am just saying that I think that what you are already doing is the right thing to do as far as I know. I do not know of any general way to know in advance.

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by loose-electron on Dec 2^nd, 2011, 11:59am

If you are familiar with the quirks of simulators, you can set up a large number of cases that show simulation tools failing to produce a proper result.

Take 3 logic inverters and put them in a ring - should oscillate right? Well, no it won't due to a lack of noise in the model.

Simulation tools are not a substitute for enough knowledge to understand what the expected behavior is.

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by buddypoor on Dec 2^nd, 2011, 12:24pm

loose-electron wrote on Dec 2^nd, 2011, 11:59am:

But you shouldn't blame the simulator in this case. The same applies to each linear oscillator that is simulated without any "kick-off", that in reality is provided by the switch-on action. The user must know that such an "artificial" help to start is required.

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by buddypoor on Dec 2^nd, 2011, 12:34pm

Garrett.Neaves wrote on Dec 2^nd, 2011, 10:16am:

........
........
I do not know of any general way to know in advance.

Hi garrett,
in response to your last reply I accidently have found a paper from an author (Czech Republic) describing a simple simulation procedure to check if the dc operatimng point of a circuit with feedback will be stable or not. It is based on the following criterion:
DC operating point is stable if the circuit is consumer for arbitrary fluctuation to be inserted. .

I'll try to evaluate and justify this sentence.
Regards

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by raja.cedt on Dec 2^nd, 2011, 2:05pm

hello buddypoor,
could you please send those moving poles paper titles or papers and this recent simulation procedure paper to my gmail raja.cedt@gmail.com ?

Thanks,
Raj.

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by loose-electron on Dec 3^rd, 2011, 7:24pm

I don't blame the simulator, I blame the model that thjehe simulator is running.

You do not "kick" your circuit in silicon. What you are doing is a "work around" of a deficiency in the model.

You can get an oscillator to self start, but there must be the suitable model of noise in the simulation.

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by loose-electron on Dec 3^rd, 2011, 7:29pm

buddypoor wrote on Dec 2^nd, 2011, 12:24pm:

that in reality is provided by the switch-on action

How about enough gain & phase in a loop to sustain-build an oscillation based upon noise or other non-ideal effects?

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by buddypoor on Dec 4^th, 2011, 2:06am

loose-electron wrote on Dec 3^rd, 2011, 7:29pm:

How about enough gain & phase in a loop to sustain-build an oscillation based upon noise or other non-ideal effects?

Loose-electron, excuse but (due to my limited knowledge of your language) I don't get the contents of this sentence. Do you mean that there can be other "starting aids" than the the power switch-on? (That's clear to me, of course)

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by loose-electron on Dec 4^th, 2011, 12:27pm

"Loose-electron, excuse but (due to my limited knowledge of your language) I don't get the contents of this sentence. Do you mean that there can be other "starting aids" than the the power switch-on? (That's clear to me, of course)"

German? Sorry, I speak some French and Spanish, but not german.

Let me explain with simple language.

1. A tranient simulation does not show the effects of noise or mismatch unless you put a transient noise model into the simulation, or put a set of mismatches into the simulation.

2. Both noise and mismatch are often not included in a transient simulation.

3. Noise and mismatch models can be inserted into a transient simulation if needed.

4. An oscillator, in silicon, not in simulation, starts to oscillate due to sufficient gain and phase parameters in the electronic circuit.

5. The electronic circuit also has noise and/or offsets being amplified in the circuit.

6. A "switch on" is not the only thing that will start a device to oscillate.

7. Sufficient loop gain, will take anything (noise, offsets, injected glitches due to a switch flip) and cause a repetitive amplification of that which builds in amplitude as an oscillation.

Hopefully a bit clearer.

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by buddypoor on Dec 5^th, 2011, 12:17am

loose-electron wrote on Dec 4^th, 2011, 12:27pm:

.............
6. A "switch on" is not the only thing that will start a device to oscillate.
7. Sufficient loop gain, will take anything (noise, offsets, injected glitches due to a switch flip) and cause a repetitive amplification of that which builds in amplitude as an oscillation.

Loose-electron,
thanks for the very detailed information.
However, as already indicated in my question: I am aware that the switch-on effect is not the only cause to start oscillations. But I think the switch-on is the most important and the most reliable "kick", is it not?
And - coming back to the original core of discussion - I still think that in case of simulation problems (false or unexpected results) the user will be the only cause and not the simulation program.
I think the circuit as mentioned in one of my former postings (opamp with positive resistive feedback) can serve as a good example:
The simulation results are correct by 100% - based on the assumptions and conditions that exist during the simulation. Problems can arise only if the user transfer the results to real-world conditions (noise, stable supply, no switch-on,...).

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by buddypoor on Dec 5^th, 2011, 1:03am

Hello Loose-electron,

With posting #4 you gave a recommendation to read Lundbergs contribution to oscillation conditions (http://web.mit.edu/klund/www/weblatex/node4.html).

With posting#6 I have written some critical comments to Lundbergs statements.

Since it was your recommendation I would be very interesting for me to hear about your opinion. In particular, do you know something about an oscillation criterion that is sufficient?
Thank you.

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by Alexandar on Dec 5^th, 2011, 1:30am

buddypoor wrote on Dec 5^th, 2011, 12:17am:

....
And - coming back to the original core of discussion - I still think that in case of simulation problems (false or unexpected results) the user will be the only cause and not the simulation program.
I think the circuit as mentioned in one of my former postings (opamp with positive resistive feedback) can serve as a good example:
The simulation results are correct by 100% - based on the assumptions and conditions that exist during the simulation. Problems can arise only if the user transfer the results to real-world conditions (noise, stable supply, no switch-on,...).

I don't agree completely. Surely, the designer needs to have some knowledge on the tool he (or she) is using, but you cannot expect him (or her) to know as much as the software/math guru's who actually made the simulation tools. In that sense, some trust/distrust is always needed. If you don't trust the (or any) simulator at all, your productivity will be fairly low, I expect. But as well, too much trust is not okay: I've seen some junior designers look for hours at a circuit, while the problem was purely simulation (e.g. numerical noise, minimum time step, voltage resolve resolution).

Switch-on/mismatch/current pulse in general is a good way to make sure you don't get stuck in a meta stable point. Switch on is probably the least invasive. For mismatch you could to do a MC analysis w/o altering the schematic.

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by buddypoor on Dec 5^th, 2011, 2:47am

Lex wrote on Dec 5^th, 2011, 1:30am:

Wow...don't agree completely!
Hi Alexandar, thanks for replying.
Perhaps you have misunderstood something.
I did never claim that the designer should "know as much as the software/math guru's who actually made the simulation tools"
My only concern was (and still is) that - in case of some ideal conditions (simulation environment or models) - the designer should know that a simple direct transfer to real conditions can lead to false results.
And I gave an illustrative exaple for that (ideal opamp with pos. resistive feedback). This has nothing to do with general trust/distrust of simulation results. It means simply that the designer should know what he is doing. So - I really don't know what you are arguing against.
And regarding your last sentence: I think it really supports my standpoint. If somebody specifies a minimum time step that is still to large for the expected signal variation - who has made an error? The simulator or the user?
Regards

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by Alexandar on Dec 5^th, 2011, 4:05am

buddypoor wrote on Dec 5^th, 2011, 12:17am:

...
And - coming back to the original core of discussion - I still think that in case of simulation problems (false or unexpected results) the user will be the only cause and not the simulation program.
...

I cite "I still think that in case of simulation problems (false or unexpected results) the user will be the only cause and not the simulation program."

This is what i don't agree to. The story is never black and white. Loose-electron already mentioned modelling. I am mentioning the fact that the user of a piece of software cannot (and is not expected) to know all the details of the simulator. It is the task and responsibility of the software engineer to make its use trivial.

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by buddypoor on Dec 5^th, 2011, 7:20am

Lex wrote on Dec 5^th, 2011, 4:05am:

[quote author=buddypoor link=1322576364/15#29 date=1323073065]

I am mentioning the fact that the user of a piece of software cannot (and is not expected) to know all the details of the simulator. It is the task and responsibility of the software engineer to make its use trivial.

Perhaps this discussion is not very fruitful - however, now I strongly disagree.
Do you really think that the use of a high-quality simulation program should and can be "trivial"? Don't you think it is up to the user only to set a correct and realistic simulation environment?
Think of items like:
Simulation time, max. time step, initial conditions, power switch-on transients, frequency resolution, operational point (offset), stability aspects,.....
Even "modelling" is a task that belongs to the users side (and is not part of the program)
Most important: The user should have in advance an idea how the simulation result should look like. Otherwise, he believes everything - even false results because of errors during schematic entry.
I am working with several different simulators since more than 30 years - and I cannot remember one single case where the simulation program itself was responsible for an unexpected and false result.
(This applies to analog circuits only; I have not much experience with digital circuits).

Please, can you give me such an example?
Regards.

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by loose-electron on Dec 5^th, 2011, 9:19am

buddypoor wrote on Dec 5^th, 2011, 1:03am:

I don't overthink this.

Sufficinet gain in the loop
suitable phase relationship in the loop

you can get fancy with Blacks Law, feedback system analysis poles, root locus etc etc etc etc but for designing things in the real world, the above two items get it done for me.

The open loop Bode plot - both gain and phase tell me I got an oscillator or not. Academics may find fault with that but for me it works for all the designs I have done.

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by loose-electron on Dec 5^th, 2011, 9:25am

Simulators are tools.

You need to know how to use the tools properly and what are the approximate results you would expect from those tools.

You can either use the tools to build houses, or you can cut your fingers off.

In silicon, I have seen the multi-million dollar slicing off of fingers a few too many times.

Nowadays, the model being simulated is much more often in fault than the simulator itself.

Garbage In = Garbage Out

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by boe on Dec 5^th, 2011, 11:32am

Back again.

buddypoor wrote on Dec 2^nd, 2011, 9:19am:

...
This sounds reasonable - but the problem is: How to know in advance (and for unknown circuits) if the simplification is "excessive" (means: not allowed and not applicable) or not?

This is a very good question. I don't think there is a general answer (applicable to all circuits of any complexity) to this question. But having loose-electron's "capable designers" certainly helps.

buddypoor wrote on Dec 2^nd, 2011, 12:34pm:

...
in response to your last reply I accidently have found a paper from an author (Czech Republic) describing a simple simulation procedure to check if the dc operatimng point of a circuit with feedback will be stable or not.

Buddypoor, could you post the reference, please?

buddypoor wrote on Dec 5^th, 2011, 7:20am:

...
I am working with several different simulators since more than 30 years - and I cannot remember one single case where the simulation program itself was responsible for an unexpected and false result.
(This applies to analog circuits only; I have not much experience with digital circuits).

Please, can you give me such an example?

Although they are rare, simulator bugs do occur; e.g. http://www.designers-guide.org/Forum/YaBB.pl?num=1181226897 (missing @cross events in a mixed-mode simulation).
However, I agree, usually unexpected results are caused by errors on the user's side (models, setup, design, ...).

I also agree with loose-electron's last post (nice wording!).

- B O E

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by buddypoor on Dec 5^th, 2011, 1:58pm

Hi Loose-electron,
thank you for the two replies.
I think, we are on the same track (or line?) - more or less, in particular your last posting.

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by loose-electron on Dec 5^th, 2011, 3:03pm

buddypoor wrote on Dec 5^th, 2011, 1:58pm:

Hi Loose-electron,
thank you for the two replies.
I think, we are on the same track (or line?) - more or less, in particular your last posting.

Not a problem -

Also - making the use of a simulator trivial is a bit risky IMHO, electronics is not Lego's or point and click internet shopping.

People do simple things in electronic block assembly all the time and later get bitten due to not understanding the implications of what they have done. I see this all the time.

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by Lex on Dec 6^th, 2011, 12:26am

Well let me expand a bit on making the simulator trivial to use. If you make it is easier to use, people will make fewer mistakes. This is a matter of organization, clarity and transparency. In my opinion software guys are not working according to these principles.

If important simulator parameters are into 3 menus and 5 sub-menus deep down some archaic software, some people will have a hard time finding them. Or if you use very cryptic names, it takes a lot of time to dig up some PDF file of 700 pages and search for the phrase to finally find out what it means, and which one they need to alter (everybody must have been there, I guess ;) ). This is very unproductive and hence leads to errors.

The fact that the matter is very complex, doesn't mean you can apply some organization, some clarity and transparency. It breaks down the complexity, allows some overview, inducing fewer errors. I am talking about the "A" in CAD.

Down to earth example: if you make your circuit look like spaghetti, obviously it will be difficult and take long for someone else to decipher how it works. Might take half an hour. Some organization and notes might reduce that time to 10 seconds.

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by loose-electron on Dec 6^th, 2011, 2:20am

usability of a software interface is a useful thing, however most EDA tools are not "Apple PC iMac" user friendly.

its a problem, I agree, but EDA is not shrink wrapped Microsoft type products, and getting improved ease of use is often not a priority for the vendor.

I do agree that it is an issue.

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by thechopper on Dec 6^th, 2011, 6:19pm

buddypoor wrote on Dec 5^th, 2011, 2:47am:

[quote author=Alexander link=1322576364/30#31 date=1323077419]

My only concern was (and still is) that - in case of some ideal Hi conditions (simulation environment or models) - the designer should know that a simple direct transfer to real conditions can lead to false results.
And I gave an illustrative exaple for that (ideal opamp with pos. resistive feedback).

Hi Buddypoor,

that should not be a concern since it is design experience what builds designer's knowledge and intuition. This knowledge is the one that will make the designer weigth and take into account the limitations of the model he's running. A good and experienced designer will be aware of such limitation. I think your concern is simply part of the natural process of learning...

Best
Tosei

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by buddypoor on Dec 7^th, 2011, 12:19am

HdrChopper wrote on Dec 6^th, 2011, 6:19pm:

buddypoor wrote on Dec 5^th, 2011, 2:47am:

Hi Tosei,

I am afraid that I can only partly agree. As you know - in particular in the area of opamps - it is common practice to idealize the gain and to use approximate formulas, which are exact enough for nearly all cases.
We know the limits/conditions for these simplifications and are, thus, aware that the formulas must not be applied for frequencies beyond a certain limit.
But the example I have presented is quite different, I think. Using the idealized opamp model the results don't only deviate up to a certain degree from the correct behaviour, but there is a fundamental difference beween real and ideal: works vs. works not at all.
Thus, there is no chance for the designer to "weigth and take into account the limitations of the model he's running".
I will tell you something about the background of my posting:
There is one author -very well known in the analog world - who has published a paper in an international electronic magazine (also with very good reputation) dealing with an "oscillator circuit" (how he calls it) that never will work in practice. It is a modification of the classical WIEN oscillator - however, all analyses are based on an ideal opamp model only. In detail, he is discussing oscillation condition, frequency, amplitude stabilization - for a "circuit" that never will oscillate in case of real opamps (immediate saturation due to net positive dc feedback).
And my original question was if such a combination of active and passive parts may be called "oscillator".
What do you think?
Regards

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by loose-electron on Dec 7^th, 2011, 9:28am

papers and journal papers get published all
the time that have errors, make incorrect
assumptions and similar.

It is not a perfect error free world.

One of the basic premises of science -

1. Publication of work
2. review of work by peers
3. Reproduction of results by peers
in an independent manner from the
original work.

A lot of the material that you read in
engineering is your
participation in step 2.

Just because it has been published
in a trade magazine, or a peer reviewed journal
(IEEE and others) does not mean that it is
perfectly correct, or even technically feasible.

Welcome to reality!
:D

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by buddypoor on Dec 7^th, 2011, 11:55am

Hi Loose-electron (Jerry),

What shall I say? It seems you have misinterpreted my last posting.
Neither was I surprised nor disappointed to find in a magazine with quite good reputation an article that seems to be not correct in all parts.
Such things I have observed many times in the past - insofar, I feel not to live in an "ivory tower". However, I only have mentioned this particular case to give a background for starting this thread - that's all!
Regards

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by loose-electron on Dec 7^th, 2011, 4:11pm

buddypoor wrote on Dec 7^th, 2011, 11:55am:

Not a problem!

In this forum you have a mix of experience at all levels.
For those with low levels of experience, they still think journals and trade magazines are infallible, and we both know that they are not.

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by thechopper on Dec 12^th, 2011, 6:47am

buddypoor wrote on Dec 7^th, 2011, 12:19am:

But the example I have presented is quite different, I think. Using the idealized opamp model the results don't only deviate up to a certain degree from the correct behaviour, but there is a fundamental difference beween real and ideal: works vs. works not at all.
Thus, there is no chance for the designer to "weigth and take into account the limitations of the model he's running".
I will tell you something about the background of my posting:
There is one author -very well known in the analog world - who has published a paper in an international electronic magazine (also with very good reputation) dealing with an "oscillator circuit" (how he calls it) that never will work in practice. It is a modification of the classical WIEN oscillator - however, all analyses are based on an ideal opamp model only. In detail, he is discussing oscillation condition, frequency, amplitude stabilization - for a "circuit" that never will oscillate in case of real opamps (immediate saturation due to net positive dc feedback).
And my original question was if such a combination of active and passive parts may be called "oscillator".
What do you think?
Regards

Hi Buddypoor,

I still think it is a matter of being aware that you are dealing with a model - and not a real circuit - during the design phase. The ultimate question is: how well the model represents the circuit you are trying to build? And I think that question applies to the example of the published "oscillator circuit" you pointed out, that only works with ideal components.
It is up to the designer to evaluate the limitations (considering second order effects introduced by actual circuits) of the model which was used for developing the circuit. In the publication you cited, if that was not done, then the analysis was incomplete and the validity of it is under question, as reality proved it.

In my opinion, your example is an example of an "oscillator circuit model" and not and "oscillator circuit": if you are talking about circuits you are talking about real stuff...not models.

Best
Tosei

Title: Re: Harmonic Oscillator with ideal Opamp models
Post by buddypoor on Dec 12^th, 2011, 9:00am

Thank you Tosei, I think I can agree with everything.