Lutz,

Here is a start-up reply.

The Diode Limiting ( as Feedback ) is a response 

that occurs after the initial wave. 

Therefore it is added to any OPA transient phase shifting. 

ie, the Diode distortion rides on the FB signal, 

therefore is always greater than the Transient delayed FB signal.

Also, if the signal waveform is not Symetrical, 

then the diode limiting will generate more InterModulation distortion.

Re your question: no.

I know of solutions with (J)FETs in the negative feedback path as well as of the quite famous solution using an incandescent lamp (see https://en.wikipedia.org/wiki/Wien_bridge_oscillator - top right corner).

To: U. Dreher.

Thank you for replying.

Yes - there are many well-known alternatives (non-linear elements) for gain control: Diodes, FET as resistor, PTC (lamp), NTC (thermistor), OTA as resistor, OTA as amplifier, hard limiting with filtering, control loop with reference value,...

No I have not seen such an amplitude control scheme.

Interesting idea, Lutz... It would be not so surprising if we remember the last viewpoint at this kind of oscillators in the question below where the differential op-amp input voltage is presented as a difference between the two partial voltages applied at the inverting and non-inverting input... 

https://www.researchgate.net/post/What_is_the_basic_idea_of_Wien_bridge_oscillator_How_does_it_operate

Hi Cyril - thank you for your reminder. However, perhaps I couldn`t make it clear:

I was not surprised about the general possibility to control the amplitude this way (increasing positive feedback will, of course, enhance the loop gain aboive unity). The main point was the QUALITY of this alternative method (better than the classical method).

In order to show the difference between both alternative I have chosen a rather bad design with unacceptable large THD values (in both cases). I will present another set of results for lower TDD values soon.

I dealt with stabilizing of oscillation in circuit of positive feedback in 1978. See ref

[7]            PUNČOCHÁŘ Josef. Stabilizace amplitudy kmitů oscilátoru diodami v obvodu kladné zpětné vazby (Stabilization of the oscillator amplitude by means of diodes in the positive feedback), Sdělovací technika, No. 11, pp. 403–404 (1978)

in https://www.researchgate.net/publication/281830241_WIEN_BRIDGE_OSCILLATOR_WITH_REAL_AMPLIFIERS

It is always AGC I think.

I have to look, I do not remember it  too much :-))

Research WIEN BRIDGE OSCILLATOR WITH REAL AMPLIFIERS

For fixed frequency AGC can be omitted. See pg.300 in

https://www.researchgate.net/publication/281830241_WIEN_BRIDGE_OSCILLATOR_WITH_REAL_AMPLIFIERS

or the picture below from

https://www.researchgate.net/publication/36861209_Operacni_zesilovace_v_elektronice

Research WIEN BRIDGE OSCILLATOR WITH REAL AMPLIFIERS

Book Operační zesilovače v elektronice

Here are some updated results with reduced influence of the non-linearity (with lower distortions):

* Case 1: TRAN analysis: f=1kHz, amplitude 2.8V, THD=0.88%.

* Case 2: TRAN analysis: f=1kHz, amplitude 5.5V, THD=0.325%.

Again the second (novel) alternative for stabilization gives better results (lower distortions) at higher amplitudes. This method seems, indeed, to work better than the "classical" method using gain control - perhaps because gain regulation seems to be the most logical way (but not the best one) ?

"I dealt with stabilizing of oscillation in circuit of positive feedback in 1978."

Hi Josef - I am not sure if we speak about the same method of amplitude stabilization. The method I have named "case 1" is the classical one based on gain control. However, the alternative I have named "case 2" is a kind of phase control. 

This works as follows: A mismatch of the WIEN network causes a phase shift for the complex loop gain (if compared with the tuned case) - connected with an increase of the feedback factor above 1/3. Thus, safe start of oscillations is ensured. Now - if for rising amplitudes this mismatch is corrected because of the diodes effect, the phase is shifted back to the "ideal" design value and the feedback factor - at the same time - is also reduced to its ideal value (1/3). The gain of the amplifier remains constant due to the two fixed feedback resistors.

Remark: In principle, this method is not new - such a phase-shifting effect is connected also with the known stabilization methods used for phase-shift oscillators. Here, we make the gain larger than ideal (-29 for example) - and we think (perhaps) that we have just changed the loop gain (for starting oscillations). But that`s not true - we also have influenced the phase of the loop gain.

However, this phase-correcting method was - as far as I know - not described or proposed also for WIEN-type oscillators up to now.

Lutz, almost certainly you're right. There is currently unavailable my archives - locally. As far as possible, this information will be given here. Josef

In view of the new results as given above I tink I must reformulate (or completely drop) the second sentence of following statement which can be found in my contribution „On a rigorous oscillation condition“ (page 4, chapt. 2.1.1, see link at the end):

The ability of these circuits to oscillate has been already demonstrated in various contributions, for example in [2, 6, 12] and for No.2 in [7, 10]. However, circuits No.3 and No.4 are not recommended for practical use because they don’t allow automatic amplitude control without influencing the oscillation frequency.

My recent investigations have shown that also circuits No.3 and No.4 allow an effective amplitude control method - most probably, even more effective than for circuit alternatives No.1 and No.2.

https://www.researchgate.net/publication/255171263_On_a_rigorous_oscillation_condition?ev=prf_pub

Data On a rigorous oscillation condition

Hi Lutz

investigated circuit from 1978 is below

Josef

Josef - thank you. Most interesting. Any results (THD)?

Hi Lutz

see https://www.researchgate.net/publication/312899871_Stabilizace_amplitudy_kmitu_oscilatoru_diodami_v_obvodu_kladne_zpetne_vazby

Article Stabilizace amplitudy kmitů oscilátoru diodami v obvodu klad...

Hi Josef - thank you for the document. As you probably know - I cannot read the paper, therefore my question: Did you explain in the paper WHY you haven chosen this particular stabilizing concept? I ask this question because - as you know - this is in contrast to the well-known concept of regulating the gain (negative feedback loop) of the amplifier (as documented in all textbooks and other published articles).

Lutz

You are right, it was some unusual. I did some measurements. An originator of this idea is

 Hileman D.: Common silicon diodes stabilize oscillator, Electronics, January 9, 1975

"Originator's" description was only qualitative, so I worked a little bit.

In my opinion, the circuit is not too favorable. This is a modified Wien oscillator. Required gain of amplifier is about 10, and distortion (k) much of anything. But it is very interesting matter :-))

Josef

Glen, I believe that the "control range" is suitable only for the "identical elements" in "Wien divider".

Josef,

Agree.

This is a modified Wien oscillator. Required gain of amplifier is about 10, and distortion (k) much of anything.

Josef - a gain of 10 ? This sounds strange to me. Nevertheless, I cannot find this "Hileman-article" in the net. Please, can you send me a copy?

To Glen: I never have heard about a "Hewlett diode". Perhaps you have mixed it with the "Hewlett lamp"? (In his first WIEN type oscillator, W.R. Hewlett  has used a tungsten lamp (PTC) for stabilizing purposes).

Lutz see tables: 1 + R1 / R2. I'm sorry, but I do not have copies. In "communist times" there was no "copy apparatus" (or locked) :-)

Josef

Lutz

Mathematical description you can see in my article - this was not in Hileman´s article - I am sure.

Josef - tomorrow, I will try to compare my theoretical/mathematical results with your paper. For now, I can give you some simulation results.

Based on the WIEN circuit as given on page 1 of this thread (same values) I have done a simulation with the stabilization scheme as given in your last paper. I used a resistor of 50kOhms in series with the two diodes (connected between R and C in the feedback path and ground).

The result was surprisingly good: THD=0.2% with an amplitude of 4volts.

This THD is better than the results obtained with the two other stabilization schemes (shown in the figure on page 1). However, the oscillation frequency was somewhat higher: 1.14 kHz (instead of 1kHz). These results deserve some more investigation. Thank you again for the new stabilization principle I have never seen before.

Lutz

That was in my salad days when I solved the problem above :-))

Good luck to You

Josef

Hi Glen,

for my opinion, it is not necessary to invent ("suggest") a name of this pair of diodes because such a name does already exist: "Antiparallel diodes". As far as your suggestion is concerned ("Hewlett Diode") I have some objections - primarily because the diodes are used to DECREASE a resistance with rising amplitudes. This is in contrast to a tungsten lamp which as a positive tempco and thus will INCREASE the resistance within a feedback path. For this reason, I am afraid the term "Hewlett Diode" is completely misleading.

Lutz,

As you wrote, it is important to see the difference in Function

between  Lamp   vs. Diodes .    It is certainly "inverse".

My point was that it had a proper name and place in history.

However,  it was my mistake, because 

I missed your use of this term in your Original Question 

"2 anti-parallel diodes".

Thank you for the 'proper' name : "AntiParallel Diodes". 

The circuit below is as beautiful as the first OP AMP with tubes.

The great advantage "PTC lamp" (R19) is its operating temperature that is high in relation to ambient temperature. Normal temperature changes, then do not play a big role. It does not apply to semiconductors PTC elements.

Two very interesting references.

Hi Josef - in our student lab we have used the (NTC) termistor-stabilized version of the WIEN oscillator already since 1978 - however, with one big disadvantage: The amplitude  was rather sensitive to variations of the ambient temperature (e.g. air movement). According to my experience and as far as signal quality is concerned, stabilization using a separate control loop with a FET-resistance works best.

Concerning the single-resistor controlled version of the WIEN type oscillator I think, there are other two-opamp based single-element controlled circuits which work better. However, it is interesting to verify HOW the single-element conrol is realized in the circuit given with your second reference.

Lutz, I know it well,

NTC termistor-stabilized oscillator acts as thermometer :-)

But with lamp - it is another story. But it is difficult to get right lamp.

Lutz,

Indeed, a JFET multiplier for amplitude control can be much better than highly-nonlinear clipping diodes or nonlinear resistors such as lamp filaments. For a detailed analysis of the JFET alternative, see

http://www.planetanalog.com/author.asp?section_id=3049&doc_id=563932

The loop dynamics are at

http://www.planetanalog.com/author.asp?section_id=3049&doc_id=563936

An even more attractive option is to use an op-amp synthesized resistance instead of the JFET.

Dennis,

Thanks for the new ideas. 

Yes, the Diode Limiters do introduce InterMod Distortion

even in the 'softest' design.  

In my designs I have avoided "highly-nonlinear clipping diode" circuits 

( schematics are in the linked page , near bottom, from my website ).

Can you provide

a   hint    or    rough schematic

for  the "OPA synthesized resistance circuit" ?

Glen,

There is an example, with design derivations, in a Wien-bridge oscillator that is part of a Z meter here:

http://www.planetanalog.com/author.asp?section_id=3049&doc_id=561342

Dennis,

Thank you for your contribution (and the links to planet analog). May I ask two questions:

* Speaking about a "JFET multiplier" - do you refer to an application where the FET is used as a controllable resistor?

* It is not clear to me how your "op-amp synthesized resistance" looks like. Can it be found in the link you have provided in your answer to Glen? I know how a grounded controlled resistance can be synthesized using an OTA - but with an opamp? Or do you have an NIC-based resistance in mind?

Lutz,

* Speaking about a "JFET multiplier" - do you refer to an application where the FET is used as a controllable resistor?

Yes, though the amplifier which gain is controlled by the JFET is variable-gain, which is essentially a multiplier.

* It is not clear to me how your "op-amp synthesized resistance" looks like. Can it be found in the link you have provided in your answer to Glen?

Yes, the derivations are included. Look at the second page of the article for derivations.

Look at the second page of the article for derivations.

Dennis - thank you. I did not notice that there is a 2nd page. My fault.

I had a look on page 2 and - as I had suspected -  your "opamp synthesized" resistance is an "OTA synthesized" resistance, right?

Dennis,

may I ask you: Why do you think that you must change 3 resistors in parallel for tuning the frequency of the quadrature oscillator - as you wrote on page 3 of your linked contribution?

https://www.researchgate.net/deref/http%3A%2F%2Fwww.planetanalog.com%2Fauthor.asp%3Fsection_id%3D3049%26doc_id%3D561342

The third one is in series with the JFET amplitude control circuit. The other two are op-amp input resistors. At least, this is the way I implemented it in the Innovatia ZM204 (www.innovatia.com). The circuit diagram is attached as a JPG file.