Bubba Oscillator and THD

[LvW]

I mean frequency variation in general - independent on the way to realize it. Or is the frequency fixed in your case?
Background: If it is required to build an oscillator that allows frequency variation (tuning) some topologies have advatages and some other are not suited. Therfore my question.

 

[sameerdhiman]

Here is the circuit. ZIP contains multisim file and PDF is simple circuit diagram.

Please note that this circuit does not have linearization around FET.

 

[LvW]

Why do you confuse us?
You do not answer my question? Instead you suddenly are posting a WIEN oscillator circuit diagram although up to now we were discussing a BUBBA topology.

 

[sameerdhiman]

Please do not get confused. Only to test my AGC network I connected it to WIEN topology. If my AGC network works well I'll insert it into BUBBA topology. I hope you will understand.

Yesterday I got late at my workplace. When I reached home I found my self totally exahusted So I only posted circuit diagram. I am sorry for delayed answer.

I am willing to build two oscillators:

1. Selectable fixed frequency using rotary switch to select between RC networks with lowest possible THD. It will be used to measure Audio pre-amps and power-amps THD figures.

2. Variable frequency oscillator (VFO) using potentiometer for general purpose use e.g. bode plot etc.

Please suggest me right topology for above mentioned purposes.

---------- Post added at 13:49 ---------- Previous post was at 13:13 ----------

Please discard posted circuit diagram. I found a mistake around differential amplifier. I'll post corrected circuit diagram today including linearisation.

---------- Post added at 14:07 ---------- Previous post was at 13:49 ----------

FvM: I tried to plot the 2N5486 voltage/current like you did with BF245C but failed to do so. Could you please tell me the steps as I am not much familiar with FETs.

 

[LvW]

I am willing to build two oscillators:

1. Selectable fixed frequency using rotary switch to select between RC networks with lowest possible THD. It will be used to measure Audio pre-amps and power-amps THD figures.
2. Variable frequency oscillator (VFO) using potentiometer for general purpose use e.g. bode plot etc.

Sameer, before selecting a particular oscillator topology, one of the most important question is if variation resp. tuning of the oscillating frequency is required. Therefore my last question regarding tuning capabilities.
It seems that you have decided to use the BUBBA structure with 4 identical RC stages. Do you really intend to change 4 RC time constants at the same time (rotary switch) in order to select different frequencies?
As far as the second option (VFO) is concerned: Be aware that there are only a few oscillator topologies that allow frequency variation by changing only one resistor or capacitor.

 

[FvM]

Could you please tell me the steps as I am not much familiar with FETs.

R_R1         $N_0001 0  1Meg  
R_R3         $N_0002 $N_0001  1Meg  
X_J1         $N_0003 $N_0001 0 BF245C/PLP 
V_V2         $N_0003 0 0V
V_V1         $N_0002 0 0V

.DC LIN V_V2 -3 3 .1 
.STEP LIN V_V1 -1 -11 -2

 

[sameerdhiman]

It seems that you have decided to use the BUBBA structure with 4 identical RC stages. Do you really intend to change 4 RC time constants at the same time (rotary switch) in order to select different frequencies?

LvW: If there is better topology please tell me I'll adopt that one as you have expertise in this field. As I have mentioned my requirement I need oscillator which could provide lowest THD figures and selectable fixed frequency range e.g. 10Hz - 100KHz.

For second option notch based oscillator would be suitable or not ?

FvM: Thanks for the netlist.

LvW and FvM: I corrected the errors in AGC section please have a look and tell me if there are any pending errors or If there is still a scope for further improvement.

 

[FvM]

The controller uses the FET threshold voltage as it's setpoint, but misses an option to adjust setpoint and controller gain independently. No problem, if the dimesioning works by chance. Apart of this point, everything seems correct.

You'll find a suggestion in literature to fine adjust the resistor ratio of R1/R17 for minimal distortion. I must confess, that I didn't try yet, the point is also ignored in the said variable resistor thread. I would expect, that you can exactly cancel the 2nd harmonic for a particular oscillator amplitude.

 

[LvW]

LvW: If there is better topology please tell me I'll adopt that one as you have expertise in this field. As I have mentioned my requirement I need oscillator which could provide lowest THD figures and selectable fixed frequency range e.g. 10Hz - 100KHz.
For second option notch based oscillator would be suitable or not ?

Hi sameer, at first I like to make one general statements about oscillators:
Each (new) proposed harmonic oscillator is evaluated under two viewpoints:
1.) Is it possible to adjust/tune the frequency without impact on the oscillation condition?
2.) Is it possible to do this adjustment with only one single component (possibly grounded)?

Both criteria are quality measures because each circuit has to compete with other known circuits.
For selection of a suitable circuit - based on these criteria - the question of THD is of minor importance because the THD is more or less dependent on the amplitude regulation circuitry and NOT on the oscillation principle.
There are a few - not many - oscillator circuits that can fulfill both requirements: For example, the double-integrator principle.
Another option is the GIC-based oscillator principle.
That means, you have to decide if such an adjustment feature is of importance for you.

However, in this context the amplifier used within the circuit has some influence because it is important if the amplifier performs hard or soft limiting. And in this respect, there is one opamp type that seems to be very well suited for oscillation purposes - and that is the opamp AD844.
This opamp is a current feedback amplifier (CFOA) that has - in addition to the voltage output - a high impedance node with external access. Therefore, this amplifier can be used as a "current conveyor (CC)" and offers a lot of interesting oscillator applications with grounded parts (resistor, capacitor). Try to find some examples by google for "CFOA oscillator".
One interesting feature of the AD844 is the fact that the gain characteristic allows soft limiting and, thus, leads to very good THD values - very often in connection wit a proper stabilization circuitry.

Regards
LvW

 

[LvW]

Sameer, something else just came into my mind:
As far as I`ve understood your approach up to now, you plan to use post filtering (bandpass) with the aim to improve the THD.
Are you aware that you have to change the mid frequency of the filter according to the selected oscillation frequency?
For my opinion, this complication is a severe argument against usage of post filtering.
Instead, you should concentrate all your effort on an optimal amplitude stabilization concept that completely avoids an additional filter stage.
(By the way, I`ll send you a paper regarding AD844 usage as a PM).

 

[sameerdhiman]

but misses an option to adjust setpoint and controller gain independently

how can I do that ? please give me some tips.

You'll find a suggestion in literature to fine adjust the resistor ratio of R1/R17 for minimal distortion.

I am reading Siliconix AN-105 but I could not find their old application note AN73-1 where things has been explained more elaborately. Do you have Siliconix AN73-1 notes ?

1.) Is it possible to adjust/tune the frequency without impact on the oscillation condition?
2.) Is it possible to do this adjustment with only one single component (possibly grounded)?

Now, I got your point.

THD is more or less dependent on the amplitude regulation circuitry and NOT on the oscillation principle.

Very important point.

There are a few - not many - oscillator circuits that can fulfill both requirements: For example, the double-integrator principle.

Found your thread https://www.edaboard.com/threads/175550/

Are you aware that you have to change the mid frequency of the filter according to the selected oscillation frequency?

Yes I am very much aware of that. I have already dropped the idea to use BPF (Adjusting mid band frequency with respect to oscillation frequency outcome will be added complexity)

LvW: I must admit, I was in the wrong impression of WIEN-Bridge stability. Yesterday, when I was making corrections in AGC loop WIEN Bridge shown me rock solid stability and it also shown 0.002% THD (simulation figure not trust worthy) in comparison with BUBBA+BPF 0.001% (simulation figure not trust worthy). This has also proved your above statement about THD.

 

[LvW]

..................
LvW: I must admit, I was in the wrong impression of WIEN-Bridge stability. Yesterday, when I was making corrections in AGC loop WIEN Bridge shown me rock solid stability and it also shown 0.002% THD (simulation figure not trust worthy) in comparison with BUBBA+BPF 0.001% (simulation figure not trust worthy). This has also proved your above statement about THD.

But remember: To adjust the frequency the WIEN oscillator parallel modification of R or C is required.

 

[FvM]

how can I do that

Adding a DC offset to the differential amplifier would allow to adjust the setpoint independently. The gain can be changed at many places, e.g. R7 or the differential amplifier.

Do you have Siliconix AN73-1 notes?

I have neither of them. But you can simply derive, that the resistor network is compensating the quadratic term in the FET Id versus Vds characteristic. If so, you can find a setting for each operation point (combination of control voltage and AC level), where the quadratic term is exactly cancelled, in other words, no second harmonic is generated by the FET.

 

[LvW]

Do you have Siliconix AN73-1 notes ?

Sameer, I have this AN as a hard copy. If it is really very important for you - tell me.
One more remark: Note that FET`s are not the only component that can realize voltage controlled and grounded (!) resistors.
You also can utilize OTA`s.

 

[sameerdhiman]

I am pretty amazed with AD844 oscillator circuit. it uses very few components.

 

[LvW]

I am pretty amazed with AD844 oscillator circuit. it uses very few components.

I think, there are also classical opamp based filter circuits with the same number of
components (WIEN-bridge based circuits) - however, AD844 allows ground referenced control.

 

[sameerdhiman]

Adding a DC offset to the differential amplifier would allow to adjust the setpoint independently. The gain can be changed at many places, e.g. R7 or the differential amplifier.

I'll try today.

Sameer, I have this AN as a hard copy.

If you have scanner then send it. Otherwise AN-105 would solve the purpose.

You also can utilize OTA`s.

Umm...then have to replace FET with BJT or will it work with FET also ?

 

[LvW]

If you have scanner then send it. Otherwise AN-105 would solve the purpose.

OK, try to work with the AN-105

Umm...then have to replace FET with BJT or will it work with FET also ?

My only intention was to point to other alternatives. Knowing the options is the basis of a good and succesful design.
Explanation: The term OTA refers to an integrated opamp with (a high resistance) current output.

 

[sameerdhiman]

Explanation: The term OTA refers to an integrated opamp with (a high resistance) current output.

OTA = Operational Trans-conductance Amplifier (Voltage In Current Out).

 

[sameerdhiman]

Logically thinking, If there is ripple in AC2DC section does it has any relationship with distortion and over all performance ? If yes, then how can I lower the ripple ?