Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

analyze a Twin_T oscillator

akbarza

Full Member level 2
Full Member level 2
Joined
Dec 18, 2013
Messages
131
Helped
0
Reputation
0
Reaction score
0
Trophy points
1,296
Activity points
2,556
I uploaded a schematic that was drawn in ltspice. I wanted to upload its pic and tried two times but the pic uploaded to 100% but paused on it. I also sent two messages to this site about this problem but I did not receive any response.
if it is possible for everyone, please upload the image of the circuit here, and I thank him/her for it.
my question:
1) what is the problem that we do not have oscillation at output? what must be changed?
2) as I know, for oscillation loopgain must be -1, how can analyze the loopgain of this circuit?
3) suppose I want oscillatio at a frequency of 50hz, and the opamp type is not given then what condition must be satisfied by the opamp? (as gain and bandwidth)
as you see in the pic(I hope), in the pic is written that the diodes are used for amplitude limiting. how do diodes limit the amplitude of oscillation?
4) finally, please explain about: "To minimize distortion, R5 must be high value.R1 is a variable resistor and used for amplitude adjust."

thanks.
 

Attachments

  • twin_t_osc.zip
    2.3 KB · Views: 110
If your oscillator has fc of a few 100 Hz, you should run your simulation a bit longer than 2 ms, isn't it?

1702298490505.png


Run time increased to 200 ms gives this waveform.

1702298235749.png
 
I wanted to upload its pic and tried two times but the pic uploaded to 100% but paused on it.

The system accepts certain image formats (and not others). See the Rules section. It's best if your image is sized under 500k.

For a couple weeks Edaboard's interface has given everyone long delays in all actions (logging in, navigating, submitting posts, etc.). Although system programmers have mostly gotten us back to normal a few bugs still remain to stamp out.
 
Start with Design Specs:
  1. Linear Sine Osc
  2. f = 50 Hz
  3. f Error = ? tolerance stackup on 6 parts vs cost
  4. Single supply, e.g. 15V
  5. BJT Op Amp e.g. LM358
  6. Steady-state time? quick or "X" = don't care , let's say < 5 cycles or 100 ms max
  7. Amplitude: ? fixed or maximum =? tolerance error = ? Let's say maximum of Vcc
  8. Distortion: ? fixed V or % harmonic from clipping on high side (Vcc-1V), let's choose no clipping.

If using Twin-T approach , how does it work?

- basically an all pass filter with phase-cancelling notch with null phase shift at resonance with null output for ideal components.
- This promotes oscillation with ANY amount of positive feedback and negative feedback for DC self-bias.
- Clipping deletes small-signal feedback thus forward and reverse gain goes to zero when limited or open loop gain and zero gain when clipped, so steady-state maximum amplitude is balanced.

  • 1st design the twin-T filter 1st then scale RC for low current.
  • I chose www.Falstad.com/afilter and chose > circuits> simple> filter> twin-t > adjusted slider to 50 Hz < 1%. after reducing spectrum slider to minimum
  • One can change all Ohms to k and uF to nF> copy all and paste into a blank page of http://www.falstad.com/circuit/circuitjs.html (^a to select all then {delete} or shortcut https://tinyurl.com/ydh8bqav
  • I added an LM358 using >Draw> Active building blocks> Add real Op Amp> then V to add battery> w for wires
  • Since LM358 has an asymmetric saturation 1V on the high side only, I choose a lower value for Vcc/2
  • Then connect wires to Vin- and Vout, It oscillates full scale to saturation immediately then decays very slowly because the -ve gain is barely <|1| which means even 1 pF of positive feedback might sustain oscillation and the balance between saturation and full open loop gain is achieved automatically at x1 so the GBW product is irrelevant ( < 500 Hz ) unlike high Q oscillators and filters where Q is squared on GBW so
  • it becomes GBW= Av²fo ( Stewart's Rule of Thumb}
  • Now I have an almost perfect osc.
  • What-if tweaks on component tolerances can use mouseover slider then ^z to undo. "spacebar to change mode to move/stretch/drag parts. (many shortcuts)

With practice, this takes about 5 minutes.
1702330505389.png

--- Updated ---

After reconsidering amplitude stability , I chose 1N4148 diode Vf @ If = 1uA positive feedback and R1 for gain rather than for the unity negative feedback.
1702337291546.png
 

Attachments

  • 50 Hz twin_t_osc.zip
    1.1 KB · Views: 100
Last edited by a moderator:
The system accepts certain image formats (and not others). See the Rules section. It's best if your image is sized under 500k.

For a couple weeks Edaboard's interface has given everyone long delays in all actions (logging in, navigating, submitting posts, etc.). Although system programmers have mostly gotten us back to normal a few bugs still remain to stamp out.
hi
the file format is .png and its size is 60kB. with this info, does uploading the pic have a problem?
Start with Design Specs:
  1. Linear Sine Osc
  2. f = 50 Hz
  3. f Error = ? tolerance stackup on 6 parts vs cost
  4. Single supply, e.g. 15V
  5. BJT Op Amp e.g. LM358
  6. Steady-state time? quick or "X" = don't care , let's say < 5 cycles or 100 ms max
  7. Amplitude: ? fixed or maximum =? tolerance error = ? Let's say maximum of Vcc
  8. Distortion: ? fixed V or % harmonic from clipping on high side (Vcc-1V), let's choose no clipping.

If using Twin-T approach , how does it work?

- basically an all pass filter with phase-cancelling notch with null phase shift at resonance with null output for ideal components.
- This promotes oscillation with ANY amount of positive feedback and negative feedback for DC self-bias.
- Clipping deletes small-signal feedback thus forward and reverse gain goes to zero when limited or open loop gain and zero gain when clipped, so steady-state maximum amplitude is balanced.

  • 1st design the twin-T filter 1st then scale RC for low current.
  • I chose www.Falstad.com/afilter and chose > circuits> simple> filter> twin-t > adjusted slider to 50 Hz < 1%. after reducing spectrum slider to minimum
  • One can change all Ohms to k and uF to nF> copy all and paste into a blank page of http://www.falstad.com/circuit/circuitjs.html (^a to select all then {delete} or shortcut https://tinyurl.com/ydh8bqav
  • I added an LM358 using >Draw> Active building blocks> Add real Op Amp> then V to add battery> w for wires
  • Since LM358 has an asymmetric saturation 1V on the high side only, I choose a lower value for Vcc/2
  • Then connect wires to Vin- and Vout, It oscillates full scale to saturation immediately then decays very slowly because the -ve gain is barely <|1| which means even 1 pF of positive feedback might sustain oscillation and the balance between saturation and full open loop gain is achieved automatically at x1 so the GBW product is irrelevant ( < 500 Hz ) unlike high Q oscillators and filters where Q is squared on GBW so
  • it becomes GBW= Av²fo ( Stewart's Rule of Thumb}
  • Now I have an almost perfect osc.
  • What-if tweaks on component tolerances can use mouseover slider then ^z to undo. "spacebar to change mode to move/stretch/drag parts. (many shortcuts)

With practice, this takes about 5 minutes.
View attachment 186837
--- Updated ---

After reconsidering amplitude stability , I chose 1N4148 diode Vf @ If = 1uA positive feedback and R1 for gain rather than for the unity negative feedback.
View attachment 186840
hi, thanks for your reply.
Although I read your reply twice, your details are more complex for me, I did not understand most of what you said. can you introduce to me a reference for theses details?
 
hi, thanks for your reply.
Although I read your reply twice, your details are more complex for me, I did not understand most of what you said. can you introduce to me a reference for theses details?
No I do not have any references.
Did you try to interact with my simulator to get a feel for how to make it decay oscillations by reducing bias current, You want to be on the edge of gain 1% above unity positive feedback. Any more and it distorts.

Try changing values with mouse wheel, move wires and parts to stretch like a rubber band by dragging the end points or move by anywhere in between.

Adjust simulator speed like a strip chart recorder in slow motion, there is no start trigger like a scope.

When you mouse-over the trace, the corresponding wire or part on schematic changes color to turquoise and visa versa. ( wish other simulators would do this.). You can change scales of traces with mouse wheel. You can undo with ^z redo with ^y using ctrl keys.

The theory you can learn after you learn how to search the web with key words or ask Chat GPT
Give a list of words which you did not understand, ok?

User



GPT 3.5

A Colpitts oscillator is a type of LC oscillator circuit that produces sinusoidal waveforms. It is commonly used in radio frequency (RF) applications. The basic Colpitts oscillator circuit consists of a combination of capacitors and inductors that form a resonant LC tank circuit. Here's a basic explanation of how it works:
  1. Tank Circuit: The heart of the Colpitts oscillator is the LC tank circuit, typically composed of two capacitors (C1 and C2) and an inductor (L). These elements create a resonant circuit with a specific frequency determined by the values of capacitance and inductance.
  2. Feedback: The tank circuit provides feedback to sustain oscillations. The key to sustaining oscillations is to have positive feedback, and in the Colpitts oscillator, this is achieved by tapping a fraction of the voltage across the tank circuit and feeding it back in phase to the input.
  3. Transistor Amplification: The tank circuit is connected to the base of a transistor (commonly a bipolar junction transistor or BJT), which serves as the amplifying element. The transistor provides the necessary gain to overcome the losses in the tank circuit and sustain oscillations.
  4. Phase Shift: The capacitors in the tank circuit provide a phase shift, and when combined with the phase shift introduced by the transistor, the overall phase shift becomes 180 degrees, ensuring positive feedback for oscillation.
Tuning a Colpitts oscillator involves adjusting the values of the capacitors and inductor in the tank circuit to achieve the desired resonant frequency. This frequency is determined by the formula:
f=12πL(C1+C2)f=2πL(C1+C2)
1

To improve distortion in a Colpitts oscillator, you can consider the following measures:

  1. Harmonic Control: Use appropriate filter networks to suppress harmonics and unwanted frequencies generated by the oscillator.
  2. Linear Components: Select high-quality, linear components to reduce non-linear distortion. This includes using high-quality capacitors and inductors with low equivalent series resistance (ESR) and equivalent series inductance (ESL).
  3. Biasing: Proper biasing of the transistor is essential to ensure it operates in its linear region. This helps minimize distortion.
  4. Amplitude Stabilization: Incorporate methods to stabilize the amplitude of the output signal, which can help control distortion.
  5. Impedance Matching: Ensure proper impedance matching between the oscillator and the load to prevent reflections and distortion.
It's important to note that while Colpitts oscillators are widely used, the choice of oscillator type depends on the specific application requirements, and other oscillator circuits such as the Hartley oscillator or the crystal oscillator may be preferred in certain situations.
 
Last edited:
hi
the file format is .png and its size is 60kB. with this info, does uploading the pic have a problem?
That is also the frequent type of image I use. The files have .png at the end. Currently uploading gives me no problem.

If the filename has no suffix then it might have a problem since it doesn't notify what is the image format.
 
That is also the frequent type of image I use. The files have .png at the end. Currently uploading gives me no problem.

If the filename has no suffix then it might have a problem since it doesn't notify what is the image format.
I usually just copy & paste
 

LaTeX Commands Quick-Menu:

Similar threads

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top