Colpitt Oscillator using opamp-uncontrolled oscillations

[abbeyromy]

Hi guys....

Need some advice in designing Colpitt's oscillator. I made a circuit after reading theory, though the oscillations are there but they are uncontrolled.

Please take a look at attached document for schematic, settings, simulation results and a little theory.

My questions is how do we limit the oscillations so that they become of almost same magnitude rather than increasing after every cycle?

Have I missed any design parameter that I need to tune?

Please help
-Hemanshu

 

[LvW]

Hi guys....
Need some advice in designing Colpitt's oscillator. I made a circuit after reading theory, though the oscillations are there but they are uncontrolled.
.........
My questions is how do we limit the oscillations so that they become of almost same magnitude rather than increasing after every cycle?

No, the oscillations are not "uncontrolled". The amplitude increases as each oscillator exhibits increasing amplitudes - if they are not limited using a suitable limiting part or network. Constant amplitudes require a loop gain of exactly unity which is not possible to achieve. Therefore, loop gain must be slightly larger than one with the consequence of rising amplitudes. That`s quite normal.

 

[abbeyromy]

@LvW

I was comparing this with the case of RC phase shift oscillator in which I was able to get constant amplitude sin waves and why because I had made the feedback gain (1/3 of RC network) * Av( 1+ Rf/R1) =1
Also for WIEN BRIDGE OSCILLATOR I was able to achieve the constant amplitude oscillations.

Similarly here as per theory the feedback is b= c2/c1 (1/10) and I am making my Av = C1/C2 = 10.
So I am making the condition of AB=1 then dont you think I should get a good constant amplitude sin wave like in other cases?
Also I am meeting the criteria for 0 or 360 degree phase shift.

Any idea on how to limit the magnitude? I understand that some acceptable rising amplitude can be encountered but in this case it is going above 15V and thus getting clipped.

Please help!!

 

[keith1200rs]

It is a long time since I looked at Wien Bridge oscillators but I seem to remember they have an amplitude limiting mechanism. This always needs to be the case with an oscillator otherwise the limit is the supply rails, or the output swing of the amplifier.

You cannot make the gain 10 and the feedback 0.1 and hope that you have a gain of exactly unity. Either it will be 0.99999 and won't oscillate or it will be 1.000001 and clip.

Keith.

Added after 1 minutes:

Edit:

By the way, if you use a discrete transistor rather than an opamp then the limiting mechanism can be the reduction in gain of the transistor as you drive it out of the small signal region into large signal region.

 

[abbeyromy]

@keith
yeah you are right the A*B is not exactly unity but very close. Also as the feedback mechanism is itself a load to the amplifier causing a reduction in gain.So in practice A*B >1 such that the load compensation happens.

One way to limit gain or amplitude stabalization that I know is adding an extra resistor in feedback circuit and placing 2 diodes across it in back to back fashion. Another can be to use a JFET as voltage controlled resistor in negative feed back path. I am working on this problem in hope of getting a amplitude stablized sine wave...

Can you suggest any modifications in the present circuit?

Thanks a lot for replying!
-Hemanshu

 

[keith1200rs]

I must admit if I am designing a sine oscillator it is usually high frequency and has a tuned circuit somewhere which is where I take the output from (and usually discrete).

The dual diode, JFET or thermistor approaches are the ones in my books for amplitude stabilisation. I will try to give more details when I have time if you need them. JFET is more elegant, I guess.

Keith.

 

[FvM]

In my opinion, the shown circuits isn't a correct Colpitts oscillator implementation, because C2 is effectively shorted by the low OP
output impedance. Generally most LC oscillators are utilizing amplifier nonlinearities rather than an explicite amplitude
control to stabilize the output voltage. To achieve acceptable waveform distortions, you should design a gain that's exceeding
the oscillation condition only by a small amount. It's too high in the shown circuit.

 

[LvW]

Hi abbeyromy

Only now I have seen that your circuit has an error.
The ground connection is wrong.
Connect R1 as well as C1 to ground (instead of the + terminal) - and you have a good oscillator. However for frequencies in the kHz range don`t use the old 741 opamp because of its lousy slew rate.
For R1=110k and R2=10k you get a rising amplitude which can be controlled in the classical way, for example with two antiparallel diodes in the feedback path.

 

[abbeyromy]

First of all thanks everybody for the guidance.
I have attached a modified circuit, simulation settings and simulation output. kindly see if the operation seems ok now. I still have one doubt that when I go for viewing the waveform from 1 to 100ms then I see constantly increasing magnitude. Is this normal? Though I have used back to back diode based arrangement, but please help me understand if something is wrong.

@keith-> I am working on a voltage divider bias BJT to acheive the same operation and would share my circuit soon and take your guidance on that.

@FvM-> Your comment gave me an idea and I think the output is kind of acceptable now. The derivation of B =C2/C1, takes into account that current is same in both the feedback path and the output path. in this case the resistor 5K is already there in the inverting path that means to balance this effect i need to put a 5K at the output before I enter the colpitt arrangement just to make sure that the ratio of Vf/vout depends only of C1 and C2.

Theory is attached in my document. Please see the waveform and let me know if something else is required? Is everything ok according to you?

@LvW

What exactly is wrong I didnt get you. One thing I understand is that I need to take output and feed it in inverting input side. this leads to 180 degree shift .The additional 180 degree phase shift require for oscillation is achieved by the fact that the two capacitors are connected together in series but in parallel with the inductive coil resulting in overall phase shift of the circuit being zero or 360 degrees.

I request you to kindly see the new circuit ( i did some modifications but overall its same) and tell me exactly what is wrong. Please see current simulation results and let me know what do you think.

Thanks again for helping me out. i am trying to understand the theory in book by actually doing it on pspice so having a lot of problem. I totally depend upon you guys for advice so please keep helping me. I am now designing the same circuit with BJT and will post it soon.

And yes everybody gets points

 

[abbeyromy]

hey, I have tried to implement colpitt with Transistor arrangement. Please see the doc and let me know if it looks ok.
The oscillations sits on DC level cuz i have directly taken output across the collector. I tried adding a C and load but things went out of control so for now just see if the oscillations are kind of ok..

Waiting for your comments :idea:

 

[E-design]

Your p-p amplitude seems very low if I make out your graph scale correctly. You need a collector resistor of a few kΩ at least.

The bjt circuit looks like if it is squegging because the time constant in the emitter circuit is way too large. Make the C about 10 times smaller.

 

[abbeyromy]

@E-design
Yes you are right, the peak amplitude is very low. But if I increase the Rc(collector resistance) then my gain gets disturbed as the matching of feedback gain * gain of negative path of transistor does not remain near to one.

I am really confused as to how to calculate correct values for C, L and Rc so that osciallations are amplitude stabilized as well as of acceptable amplitude and quality.

How do you guys do it? Any maths you can figure out?

I will try the modifications you suggested and will get back to you.

-Hemanshu

Added after 1 hours 30 minutes:

@E-design
Ok i think I have got it right this time and my mathematics also says that the values of R,C matches theory.

Please give a look at the modified circuit

 

[LvW]

Hi abbeyromy,

since you are arrived now at an oscillator with a BJT as active device it makes not much sense to further discuss the opamp version.
However, just one general remark: Both concepts differ substantially because the BJT acts more or less as a current source and the opamp nearly as an ideal voltage source.

 

[abbeyromy]

@LvW

yes you are right!! so the BJT design looks good to you?

@all
you guys are awesome!! Everybody who has replied to me has motivated me to work further, I hope one day I am able to answer queries like you guys do..

 

[E-design]

I am sure you still have a squegging problem with the emitter time constant of RC being to large. That should show up over a longer simulation time period.

Look at simulation plots below.

Top is with your values. Bot is with small RC in the emitter and reduced loop gain.

Read about squegging here http://www.analoghome.com/articles/an002.pdf

 

[abbeyromy]

Thanks E0design,

squegging was a new concept to me...

 

[keith1200rs]

Cor, squegging - I haven't heard that term for years. It was well covered in my University course (a very long time ago).

Keith

 

[yasir_ali86]

@ abbeyromy

Hey abbeyromy! will you help me now to understand Copitts Oscillator and to design One. please. i am waiting for your Reply

Thanks

 

[abbeyromy]

@yasir_ali86

Go through the attached design in this post and ask your doubts.

 

[yasir_ali86]

well..... first of all thanks a lot for reply.
yr i've studied colpitts Oscillator. and design according to as per theory but did not had the oscillation, as you just designed the and showed in your 1st attachment. may be i did not got oscillation because those days i was use Work bench and now i know that this software do not response to every circuit correctly. well........... yr my question is how did you modified your 1st circuit.? i mean changed the resistor values have been changed and added another resistor with back to back diode in feed back? mean i've never seen such configuration. if you read this from any book tell me name of the book so that i can have a stable sine wave output too. all i have to say is i'm still stuck till your 1st attachment circuit. i wanna read the theory and want to modify my colpitts oscillator too.

Thanks