Bubba Oscillator and THD

[LvW]

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 ?

Sameer, please make our life more easy. I couldn`t find your AC2DC section - perhaps I did not spend enough time for searchig. More than that, when you ask us how to reduce "ripple", you should describe what kind of ripple (sinus? frequency? amplitude?). How should we answer without knowing the characteristics?
My assumption is the following: Within each control loop there is a control signal (error signal) that swings around the wanted reference value. This leads to a signal amplitude that is not ideally constant (it swings around the nominal value) and is called signal "breathing".
Do you refer to this effect? Its frequency is determined by the time constant of the control loop (mainly the time constant of the rectifier unit) and should be at least ten times the signal period.

 

[FvM]

If there is ripple in AC2DC section does it has any relationship with distortion and over all performance?

Yes, why you are asking, you already know.

If yes, then how can I lower the ripple?

By adjusting the time constants.

If you think about the problem, you'll find, that you can simulate the distortions introduced by the variable gain element and the control loop separately, operating it with a sine voltage source.

 

[sameerdhiman]

Sameer, please make our life more easy. I couldn`t find your AC2DC section.

A section (full-wave rectifier) sitting between output of wien and differential amplifier.

You should describe what kind of ripple

the output from full-wave rectifier is not smooth there is around 30mV pk-to-pk voltage.

 

[LvW]

For my opinion, the rectifier is to "advanced". According to my experience, a simple diode followed by a load capacitor and a large resistor in parallel (peak detection) does the job. Your ripple is something between 1kHz and 2kHz - identical to the oscillation frequency, right?

 

[sameerdhiman]

Your ripple is something between 1kHz and 2kHz - identical to the oscillation frequency, right?

Right, Oscillation frequency is 1KHz and after full wave rectification ripple frequency is 2KHz with 30mV pk-to-pk value.

 

[LvW]

Sameer,

I have done a simulation of a WIEN oscillator (3.5 kHz) with rectification: 1 diode, C=5uF, Rp=100 kohms.
The result is enclosed: ripple app. 1..2 mvolts.
Remark: watch the shape of the ripple (loading/deloading of the capacitor).

 

[LvW]

I don`t know why, but I have some problems with the attachement.
Here is another attempt.

 

[FvM]

Obviously the rectifier ripple must cause harmonic distortions, because the FET performs a signal multiply. So a half-wave rectifier will generate 2nd order harmonics and the full-wave 3rd order. The exact amount can be calculated from the FET characteristic or determined in a simulation. With a few 10 mV ripple, it's most likely still acceptable.

I agree, that a simple half-wave rectifier would work, too. But oscillators with "precision amplitude stabilization" can be found in literature. For a fixed frequency oscillator, there's a simple solution by making the control loop slow. But if frequency tuning is involved, the control loop dimensioning gets more complicated.

 

[LvW]

Quote: For a fixed frequency oscillator, there's a simple solution by making the control loop slow. But if frequency tuning is involved, the control loop dimensioning gets more complicated.

Yes, unfortunately this is true. Changing the oscillation frequency requires adaption of the detection circuitry at the same time.

 

[sameerdhiman]

1. So a half-wave rectifier will generate 2nd order harmonics and the full-wave 3rd order.
2. Changing the oscillation frequency requires adaption of the detection circuitry at the same time.

Yes, these kind of notes are very helpful to me. Thank you so much.

It appears that you use Orcad PSPICE. Am I right ?

 

[FvM]

Both PSPICE and LTSPICE.

 

[sameerdhiman]

Today I found, 2SK104 and 2SK105 are the only available JFETs in our local market.

 

[FvM]

Today I found, 2SK104 and 2SK105 are the only available JFETs in our local market.

They should work, although Vgs,thr is specified with a rather wide range. By the way, who is manufacturing these old NEC types now? Or are they just from surplus stock?

 

[sameerdhiman]

The shop keeper had only 6-7 pieces. It seems these are from surplus stock. Yesterday, I bought one piece of 2SK104 (Rs 30/-) and one piece of 2SK105 (Rs 20/-).

In India you may still get pair of SL100 and SK100 (Medium power BJT) which are obsolete but CDIL is still producing them.

 

[LvW]

Sameer, with respect to frequency variation and the impact on the amplitude regulation circuitry (FET, variation of time constants) - did you consider already the thermistor principle? I have experienced very good THD results using a thermistor as amplitude-dependent resistor if the surrounding temperature does not vary to much. However, it is necessary to select a proper (large) thermal time constant for the thermistor.

 

[sameerdhiman]

Actually thermistors are not readily available in our local market.

---------- Post added at 20:54 ---------- Previous post was at 20:36 ----------

I have some lamps with me

Torch: 3V
Old decoration lamp: 6.2V
New decoration lamp (Chinese): wild guess around 12V

Today I setup my work bench after 12-13 years (I am feeling very happy).

 

[LvW]

Sameer, by mentioning the thermistor I was referring to an NTC (negative temp coeff), not a tungsten lamp.

 

[FvM]

Incandescent lamps have been often used as PTC for Wien-Bridge amplitude stabilization. But the lamps should have a very low rated currents of a few 10 mA to allow the circuit be driven by an OP.

 

[sameerdhiman]

My first attempt with Wien and PTC (tungsten lamp)

R=10K
C=0.022uF

Rf=33 Ohm
Rg=10 Ohm + 4.7 Ohm (lamp)

Oscillation frequency = 730Hz

Negative going swing is having clipping.



---------- Post added at 23:43 ---------- Previous post was at 23:40 ----------

Sorry forgot to mention

Output swing = 2.5V (pk-to-pk)

 

[FvM]

The load resistance is too low for most OPs.