a question about differential ring oscillator

[incol]

in order to decrease the influence of temperature to the frequcecy, I use positive temperature coefficient current as bias current. But I meet a problem, when VCC=2.4v, the frequency is not changed as temperature, but when VCC= 5.5v, use the same value bias current, the frequency changed, I dont know why using the same bias current, the frequency changed. Is frequency changed a lot as VCC?

 

[incol]

no one know this question?

 

[gajen]

There can be multiple reasons for this. First the bias current itself will change due to finite output impedance of the current source. Then gm of input devices will change due to body effect. Can you post your archiecture.

 

[incol]

There can be multiple reasons for this. First the bias current itself will change due to finite output impedance of the current source. Then gm of input devices will change due to body effect. Can you post your archiecture.

the current of I1

 

[incol]

There can be multiple reasons for this. First the bias current itself will change due to finite output impedance of the current source. Then gm of input devices will change due to body effect. Can you post your archiecture.

the current of I2

 

[incol]

bias current

 

[incol]

bias current

delay cell

 

[gajen]

From all circuits, Current I1 will increase due to power supply change. As trasistor M43 is single transistor and node 18 will change with power supply. If you can try using a cascode for this transistor, then impact of power supply on I1 will be minimum. I2 looks perfect. Then "I2-I1" i.e difference of these 2 currents also has some single transistor before it goes to delay cell. See if you can improve these, then power supply variations impact can be minimized.

 

[incol]

From all circuits, Current I1 will increase due to power supply change. As trasistor M43 is single transistor and node 18 will change with power supply. If you can try using a cascode for this transistor, then impact of power supply on I1 will be minimum. I2 looks perfect. Then "I2-I1" i.e difference of these 2 currents also has some single transistor before it goes to delay cell. See if you can improve these, then power supply variations impact can be minimized.

ok, thank you. And is current the main influence of the frequency? so if we meet the frequency changing, the first thing is to check the current? For the R and C is not always changed.