SAW oscillator design question

[GDF]

I designed a SAW oscillator with 400MHz center frequency. I got a loop gain
Aβ=6dB by AC simulation , and the phase is 0° at 400MHz. But I can't get
oscillation result in tran simulation. I have given a stimulus of current pulse and
my simulatior is Cadence. Anybody know why I can't get oscillation result?

Thanks

My schematic is as attachment.

 

[biff44]

A "simulation" could fail for any one of a number of weird reasons. But one real reason could be that the open loop gain looks like it is 6 dB into 50 ohms, but when you close the loop the change in impedance (saw to amplifier) could make the round-trip gain less than zero, or shift the phase enough so it hasn't enough gain to oscillate. Put in a dummy gain block at both sides of the resonator and see if it starts up.

 

[GDF]

A "simulation" could fail for any one of a number of weird reasons. But one real reason could be that the open loop gain looks like it is 6 dB into 50 ohms, but when you close the loop the change in impedance (saw to amplifier) could make the round-trip gain less than zero, or shift the phase enough so it hasn't enough gain to oscillate. Put in a dummy gain block at both sides of the resonator and see if it starts up.

Hi biff44,

Thanks to your reply, I do the loop gain simulation considering the loading effect
of the amplifier. I put a laod a cap to the SAW as the input impedance of amplifier.
And, the loop gain is not base on 50ohms system, I didn't add 50ohms load to my
simulation.

I'll try your suggestion. Thanks,

 

[crystalguy]

Two things I would suggest. Put a timed switch on the transistor bias network this should "Kick" the circuit into life. Make sure to set under ".options" "method" Trapezoidal as the method of integration. If this does not work let me know and I will give you the details of the more regular (and harder to explain) method to get this oscillator to simulate properly.
Jim

 

[GDF]

Two things I would suggest. Put a timed switch on the transistor bias network this should "Kick" the circuit into life. Make sure to set under ".options" "method" Trapezoidal as the method of integration. If this does not work let me know and I will give you the details of the more regular (and harder to explain) method to get this oscillator to simulate properly.
Jim

Hi crystalguy,
Thanks to your reply.
First, I can't see the "trapezoidal" in the integration method. I only see
"trap" and "traponly" in cadence simulator. I have tried it. It works.
By the way, I lower the Q of SAW in my simulation, the oscillation is easier
to happen. But I don't know why? Any advice?
Thanks,

 

[biff44]

I would still look at the phase shift. Are you doing a large signal analysis? If so, maybe the round trip phase shift is close enough when there is a large signal, but not close enough when there is a small signal (when it is just starting up). Round trip phase shift will change as the device saturates. Opening up the Q will give you less phase shift vs. frequency and loss in the resonator, so it would be easier to start up close enough to the needed oscillation criteria.

 

[crystalguy]

I would not de-Q the SAW model if moderatly accurate results are needed. The Gain in the circuit is rather low i have noticed. For this type of oscillator it is the usual practice to use an inductor (choke) in the collecter in order to pump up the gain.

Added after 20 seconds:

I would not de-Q the SAW model if moderatly accurate results are needed. The Gain in the circuit is rather low i have noticed. For this type of oscillator it is the usual practice to use an inductor (choke) in the collecter in order to pump up the gain.

 

[GDF]

I would not de-Q the SAW model if moderatly accurate results are needed. The Gain in the circuit is rather low i have noticed. For this type of oscillator it is the usual practice to use an inductor (choke) in the collecter in order to pump up the gain.

Added after 20 seconds:

I would not de-Q the SAW model if moderatly accurate results are needed. The Gain in the circuit is rather low i have noticed. For this type of oscillator it is the usual practice to use an inductor (choke) in the collecter in order to pump up the gain.

Do you think the inductor is suitable for 400MHz operation? Or, If I only use
higher resistor to solve this issue?
Thanks,