Bubba Oscillator and THD

[LvW]

LvW and FvM please ping the below ip address and send me the screenshots. I want to see how much latency is there because I'll provide you the FTP link for videos directly from my PC. Videos are quite large around 250MB each.
IP Address: 202.89.69.35

Sameer, I don`t like to download any videos because I don`t know what I should learn from it.
But - for my opinion - up to now you didn`t tell us not exactly
(a) what you want (exact specification), and
(b) what you have reached presently.

As far as I`ve seen, your oscillator does work! So, what is the problem? Amplitude? THD? Frequency variation capability?
You should try to describe your problems as clear as possible - otherwise it´s really not easy to help you.

 

[sameerdhiman]

As both of you needed to see the wave shape at Gate pin in calibration mode, I captured it into video beacuse wave is ever shifting without adjusting sweep variable.

Ok, as you wish, I stopped FTP service.

a. Amplitude is not stable over frequency range.
b. Almost same THD over entire range (target is to achieve 0.001% 10-ppm)

 

[FvM]

As both of you needed to see the wave shape at Gate pin in calibration mode

I don't necessarily need to see it calibrated. I wanted to know the frequencies. My suggestion is, that you determine the frequencies and report it.

Amplitude is not stable over frequency range.

How unstable? Seems to indicate, that the amplitude control doesn't work correctly.

 

[sameerdhiman]

OK, what ever input you need from me just make a list of points and let me know. I'll try to measure it but if I stuck while measurement I'll need your help to cross the hurdle.

Is it fine ? (yes/no)

 

[LvW]

Sameer,
I repeat some parts from my yesterday posting:

(a) what do you want (exact specification), and
(b) what do you have reached presently.
As far as I`ve seen, your oscillator does work! So, what is the problem? Amplitude? THD? Frequency variation capability?

*Specification: Frequency (range), Amplitude (value, allowed variation), THD max ?

Quote: "Amplitude not stable over frequency range"

What shall we do with such a vague description? What kind of variation ? Which frequency range?

I think, when you really understand the function of the whole circuit (that means: of each single part) it should be not a problem for you to do some modifications in order to come closer to the design goal (that we do not know in detail)!
In this context: have you used any circuit simulation up to now? This tool can give you very important information and can help to improve the circuits operation - especially for harmonic oscillators. Why?
Answer: Such oscillators are "critical" circuits because of an inherent contradiction. This becomes clear with the following sentence:
In order to operate as a "linear" oscillator with a sinusoidal output the circuit must contain a certain degree of non-linearity.


Thus, you have the task to find the "best" compromize between these two conflicting requirements. Here, a simulation program can be very helpful.

 

[LvW]

Hi Sameer,
as you probably may have noticed, I am interested in analog electronics – and particularly in harmonic oscillators (since several years).
Therefore, I have performed some circuit simulations for a WIEN oscillator topology with
(a) your rather complicated signal rectification and
(b) with a simple diode-based rectifier circuitry (see attachement).

Already on 26th of February I have told you that – for my opinion – your detection circuitry is to "advanced" and that a diode-based circuitry even gives better results.
Therefore, my question: Why have you selected the rather complex full-wave rectifier?
For a good-quality signal the rectifier time constant must be carefully chosen in accordance with the signal period – and this causes problems for your circuit when signal tuning capabilities are required (as you already have noticed and reported).

In the attachement my simulation circuit is shown – together with some results.
In short:
Amplitude: 3 V at 720 Hz (period Tp=1.4 msec),
Control voltage across RL||CL: -2.35 V with ripple of 6mV (only!).
Drain voltage: 120 mV (only!).
Time constant (rectifier): 500 msec (adjusted/selected for signal period: Tp=1.4 msec)
R,ds (operating): app 1.2 kohms (please note R11=5 k in parallel, further reducing the non-linear R,ds influence)

(In comparison, the best results with your rectifier circuitry: 6 V output amplitude and 80 mV (eighty!) ripple riding on the control signal).

The simple rectification principle has the advantage that the time constant Rp||Cp can be modified together and in the same amount as the signal period (RP, RS, RL to be tuned in parallel). This offers the possibility to maintain oscillation with constant amplitudes over a rather large frequency range.
For example, I have tuned these resistors between 1k and 100k resulting in good quality signals (THD not calculated) with nearly constant amplitudes (frequency between 72 Hz and 7.2 kHz).
Perhaps you should think about redesigning your circuit?
Regards
LvW

 

[sameerdhiman]

My suggestion is, that you determine the frequencies and report it.

FvM: it is exactly at 50Hz. (same as our mains frequency)

*Specification: Frequency (range), Amplitude (value, allowed variation), THD max ?

I'll mention them shortly.

In this context: have you used any circuit simulation up to now?

LTSpice and Multisim.

as you probably may have noticed, I am interested in analog electronics – and particularly in harmonic oscillators (since several years).

That's why I chose both of you as my teacher.

Therefore, I have performed some circuit simulations for a WIEN oscillator topology with
(a) your rather complicated signal rectification and
(b) with a simple diode-based rectifier circuitry (see attachement).

Sorry, I could not find your attachment here.

Therefore, my question: Why have you selected the rather complex full-wave rectifier?

I was in the impression that full-wave rectifier will output smooth DC control voltage as I got during simulation also. This rectifier circuit is from National's opamp circuit collections.

your detection circuitry is to "advanced"

I am still confused in this concept. I need some basic tutorial.

 

[LvW]

Sorry, I could not find your attachment here.

Ohh sorry - I am getting older and older...
Here it comes...
LvW

 

[FvM]

it is exactly at 50Hz. (same as our mains frequency)

So it may be trivial mains hum and no indication of instability. You can set the oscilloscopy trigger to mains anc check, if the low frequency signal is exactly mains synchrone, also if the signal stays when changing the amplitude control setings or oscillator frequency.

Mains hum would be mainly a breadboard setup problem, but the respective amplitude modulation also questions the oscillator signal quality.

 

[sameerdhiman]

Your design has very simple rectifier circuit

LvW: Are you comfortable with LTSpice simulator ?

Please give me some tutorial and example for Detection circuit advancement.

 

[LvW]

Your design has very simple rectifier circuit
LvW: Are you comfortable with LTSpice simulator ?
Please give me some tutorial and example for Detection circuit advancement.

Yes, the rectifier circuitry is rather simple - nevertheless, I think it is rather effective and my simulations did not reveal any advantages (that means: better rectification) for the rather complex circuitry as proposed by you.
More than that, severe problems can be expected in case of (a) parts tolerances (10k, 20k between JP2 and JP3) and (b) in case of frequency tuning.
By the way: You can remove the buffer after the RC lowpass (between JP3 and JP1). It is not necessary.
Regarding LTSpice: Yes, it is on my PC - however, I do not use it very often. Mostly, I rely on PSpice.

General remark: I do not know about the status of your project now. But I like to repeat: Try a logical and systematic design of the control loop. That means: Start with an analysis of the FET and it's resistance function in order to select a proper value. Only then, all the other resistor values can be determined.

 

[sameerdhiman]

LvW: Please provide some small write-up or tutorial with example about the theory of "Control-loop time constants". It would be great for me as well as for the other beginners.

 

[LvW]

LvW: Please provide some small write-up or tutorial with example about the theory of "Control-loop time constants". It would be great for me as well as for the other beginners.

I think, it is not appropriate to write a "turorial" about this subject.
There are two extremes as far as the time constant tau of the control loop is concerned:
1.) If tau is to small (if compared with the signal period) each single half-wave is "disturbed" - that means limited in its amplitude (like the diode limitations method), resulting in a "bad" (not optimum) THD.
2.) If tau is to large, the control loop does respond to slowly - resulting in an unwanted amplitude modulation (amplitude "breathing"). For even larger tau values the oscillation stops and starts again for a short period (like 100% amplitude modulation).
3.) The optimum choice for tau consists of a compromize between both effects - that is a tau value that results in an oscillation signal (nearly) without any amplitude modulation and without disturbance of a single wave. In practice, the factor between tau and the signal period Tp should be at least app. 50...100.

 

[sameerdhiman]

There are two extremes as far as the time constant tau of the control loop is concerned:
1.) If tau is to small (if compared with the signal period) each single half-wave is "disturbed" - that means limited in its amplitude (like the diode limitations method), resulting in a "bad" (not optimum) THD.
2.) If tau is to large, the control loop does respond to slowly - resulting in an unwanted amplitude modulation (amplitude "breathing"). For even larger tau values the oscillation stops and starts again for a short period (like 100% amplitude modulation).
3.) The optimum choice for tau consists of a compromize between both effects - that is a tau value that results in an oscillation signal (nearly) without any amplitude modulation and without disturbance of a single wave. In practice, the factor between tau and the signal period Tp should be at least app. 50...100.

Very nice write up, brief and to the point. Now my fundamental about control-loop time constant (tau) got cleared. Thank you Lutz.

FvM: So it may be trivial mains hum and no indication of instability.

I'll try to check, tonight whether suspicious wave vary with gain/frequency OR it remains stable over the gain/frequency variation.

LvW:
(a) parts tolerances (10k, 20k between JP2 and JP3).
You can remove the buffer after the RC low-pass (between JP3 and JP1). It is not necessary.
Start with an analysis of the FET and it's resistance function in order to select a proper value.

I'll try this on coming Saturday night and whole Sunday.

I must specify that all resistors used in physical circuit are CFR 5%. In our local market they do not sell MFR 1% but they can provide MFR 1% on bulk order (not less than 100 piece of each value).

 

[sameerdhiman]

FvM: Confirmed, its trivial mains hum. Does not change with gain or frequency variation.

 

[Audioguru]

It seems that every circuit complained about on electronics chat forums built on a breadboard has the same problems of mains hum pickup.

I always build a prototype circuit on stripboard with a compact layout. The strips are shortened to the required short length so there is no mains hum pickup. All the connections are soldered so they are not intermittent and it is easy to change a resistor or capacitor value.
Usually I built custom designed circuits where only one was needed. My stripboard prototype was usually the only one and it was the one that was sold and installed.

 

[sameerdhiman]

Hello AudioGuru,

This is the first time when I used breadboard for the prototype circuit.

Usually, I develop PCB at home even for the prototypes. When I did not had the PC, I used to use the Drawing board, Drawing sheets, Drafter, Set Squares, Circle master etc but now I use toner transfer method (easy and fast). In my developed PCB I never faced these kind of problems. This was the first time when I saw the hum and I am happy that I learnt something.

 

[FvM]

I generally agree with Audioguru about breadboard problems. In this case, the circuit has high impedance nodes, that are succeptible to picked up hum. I assume, that the power supply is well stabilized, otherwise there would be another possible reason. You can try to operate the circuit in a metal box connected to circuit ground, or at least placed upon a grounded metal sheet. Then the injected hum should disappear.

 

[sameerdhiman]

Well, the power supply is like the following

Transformer (15-0-15V 1A) -> 1N4007 Bridge -> 4700uF per rail -> 7812 & 7912 -> 10uF per rail -> Output

On breadboard power supply is decoupled using 100n polyester capacitors.

Last night, I noticed, oscilloscope was showing distorted 50Hz wave when the probes were not connected to anything. The settings were Time: 10mS/Div, Volt: 5mV/Div, Trig: Line. I'll confirm the exact amplitude tonight and also post the snap.

My doubts (other than circuit):
1. Oscilloscope is placed near the steel Almirah.
2. 5KVA voltage stabilizer in the mains.
3. There is no Earthing in the house.

Please suggest other possible reasons, if any.

 

[FvM]

Last night, I noticed, oscilloscope was showing distorted 50Hz wave when the probes were not connected to anything. The settings were Time: 10mS/Div, Volt: 5mV/Div, Trig: Line. I'll confirm the exact amplitude tonight and also post the snap.

In my opinion, this is just a more verbose description of what I called "trivial mains hum". Just consider that it's there, mainly cause my electrical fields and in some cases (e.g. near to transformers) also by magnetical fields. Any high performance audio or precision measurement equipment should have a metal case connected to circuit ground. If it's not appropriate for some reason, an inner metal shield would be necessary.

Simply check what are the suitable counter-measures for your design.