OTA - C oscillator - complete flat response

[push_du]

OTA - C oscillator

I am trying to simulate a CMOS OTA-C oscillator using Pspice. The OTA is simulating the correct transfer curve. But when I need to generate an oscillation, it gives a complete flat response. I have tried it out using ways like initial charge on a capacitor/ inputting a damped sinusoidal input and also trigerring it with a current pulse.

I do not understand how to make it start oscillations. Do I need to change the values of the components used or otherwise, what changes should I make ??

I am attaching my Pspice code and the circuit diagram I used for the differential input single-ended output OTA. Please answer my query.

 

[LvW]

Re: OTA - C oscillator

There are a lot of different OTA oscillator topologies.
However, the one selected by you is unknown to me - and I doubt if it can work.
Question: What is the source of this circuit ? Is there any description or explanation regarding the operation principle?

 

[push_du]

Re: OTA - C oscillator

It is a paper by Dr. Senani from where I used the circuit. Kindly tell me what problem is persistent in my way of simulation. Thanks.

 

[rfsystem]

Re: OTA - C oscillator

First you can make an oscillator from two integrators built up from a transconductance cell and a capacitor. If the transconductance and the capacitance is ideal that lead to oscillation where the amplitude is undefined. The oscillation could stay for infinite. The amplitude depend on initial condition. A real transconductance cell have a small output conductance and a small phase delay at the target oscillation frequency. Because of the ouput conductance the oscillation will not start. You have to built an additional negative selftransconductance with limited amplitude. So in effect parallel to the output of the transconductance cell you need a negative impedance for small signals but positive impedance for higher amplitudes. That lead to stable, defined amplitudes.

The two caps should linear independend. So you could not connect both in series with a voltage source.

I did not analyse the proposed one but I know the simple OTA-C oscillator where both caps are connected to ground and to the output of the transconductance cell. You have to add the small negative selftransconductance cells. That's all.

I think the series connection of the caps in the proposal gives no specific advantage. The disadvantage is that you can not derive a fully symmetric circuit from the this single ended proposal because at least one cap is not grounded. Also if both caps are grounded you can use different caps or different transconductances for the same frequency. So the proposed advantage of using different caps values is not unique.

 

[LvW]

Re: OTA - C oscillator

It is a paper by Dr. Senani from where I used the circuit. Kindly tell me what problem is persistent in my way of simulation. Thanks.

Hi, push_du

1.) At first, a more general comment:
I would be very surprised if the circuit as given by you would originate from Prof. Senani, because he is a very experienced analog "guru".
And I am right: The paper you have mentioned is not from Senani, but from M. T. ABUELMA'ATTI who claims to have simplified and improved Senani`s circuit.
This is a good example proving that even contributions in Electronics Letters are not always correct.
2.) So, what is wrong with the circuit? It's loop gain is unstable by itself! This can be easily verified by inspection (positive feedback of one OTA stage) and by simulation in the time domain. From the system theory viewpoint: The loop gain has a pole in the right half of the s-plane.
3.) And what about the oscillation criterion? As you have mentioned, the Barkhausen criterion can be shown to be fulfilled (result of an AC analysis for the loop gain).

This example shows a very important and very interesting fact, which is NOT mentioned in most textbooks:
A circuit with feedback which fulfills the Barkhausen criterion will NOT necessarily oscillate! There is a common misunderstanding about this criterion, since it is believed that it is a necessary as well as a sufficient oscillation criterion. This is not true!
Barkhausen never has claimed that a circuit with a loop gain of unity at one frequency will oscillate at this frequency. The other way round: He has stated that a circuit to operate as an oscillator must have a loop gain of unity. That´s the big difference between "sufficient" and "necessary".
I am afraid, that M. T. ABUELMA'ATTI was not aware of this.

Comments are welcome! Especially, I am interested to learn if anybody has seen a book which contains a sufficient oscillation criterion .
Regards to all
LvW