+ Post New Thread
Results 1 to 12 of 12

15th May 2006, 17:39 #1
 Join Date
 Nov 2005
 Posts
 136
 Helped
 11 / 11
 Points
 2,872
 Level
 12
How to do stability analysis of a common mode feedback loop?
Can anyone guide me how I can do a stability analysis of a common mode feedback loop? I want to plot the phase margin of that over PVT variations. Please help.

Advertisment

15th May 2006, 19:07 #2
 Join Date
 Nov 2005
 Location
 EGYPT
 Posts
 1,347
 Helped
 220 / 220
 Points
 11,124
 Level
 25
Stability analysis
cut the loop at some point, add a probe instance "iprobe" from analoglib between the two points where u cut the loop, from analog enviroment choose stb for stability analysis , fill the required as if u r doing ac analysis "not as if, actually u r doing ac analysis", at select probe push select and select the probe u inserted.
simulate
from analog enviroment >results>direct plot, main form ,stb , "magnitude" and phase ,plot
1 members found this post helpful.

16th May 2006, 12:41 #3
 Join Date
 Nov 2005
 Posts
 136
 Helped
 11 / 11
 Points
 2,872
 Level
 12
Re: Stability analysis
Hi ,
I have done the analysis but there is a problem. The circuit is unstable(I think) at 1GHz(design freq). What should I do to get it stable at 1GHz? I have tried different values for Rc, R and C but to no effect. Should I have to redesign the whole circuit?
Please help. I have attached the plot and circuit diagram.

16th May 2006, 12:55 #4
 Join Date
 Nov 2005
 Posts
 136
 Helped
 11 / 11
 Points
 2,872
 Level
 12
Re: Stability analysis
Oops forgot the plot here it is.

16th May 2006, 13:13 #5
 Join Date
 Mar 2005
 Location
 USA
 Posts
 127
 Helped
 2332 / 2332
 Points
 53,644
 Level
 56
Re: Stability analysis
Hi,
1) First, what is the intent of your main circuit? You have the input at gate of NMOS differential pair tied to gate of an NMOS load (positive feedback?)
2) You have shorted the output with a resistor (Why?).
3) For stability analysis, you have to keep Rc in place  the second picture shows that one of the Rc is missing.
4) In your common mode correction/detection circuit, do not use ideal current source. Use a transistor current mirror. This is important because the impedance of the tail current source affects the common mode response in your circuit configuration.
5) Instead of your current way of probing, I would insert a voltage source probe between the output of the CMFB circuit and the point where it is feeding to the gate of the PMOS current source load.
6) Is the compensation you add (R,C) at the input of the CMFB detection/correction differential pair helping you at all? Is that the location of the dominant pole for the CMFB loop? I guess not, most probably the dominant pole is at the output of the main circuit.
Bharath

Advertisment

16th May 2006, 14:02 #6
 Join Date
 Nov 2005
 Location
 EGYPT
 Posts
 1,347
 Helped
 220 / 220
 Points
 11,124
 Level
 25
Stability analysis
i am little confused about ur circuit , u r using the cmfb for mirroring some current, u should first pass the current "ideal source" throught a the mirror branch , then copy this current to ur circuit (either CMfb and main circuit), also in ur circuit does the diode connected load good for cmfb, it should have low gain , so the cmfb will not precisely adjust ur cm to the Vref , i think u should use a diff. to single ended diff. pair .
a last question what is the name of ur main circuit.(is that some kind of buffer)

16th May 2006, 18:23 #7
 Join Date
 Dec 2005
 Location
 EGYPT
 Posts
 63
 Helped
 7 / 7
 Points
 1,842
 Level
 9
Re: Stability analysis
what about make an ideal CMFB and put it insted of ur circuit version then run stb and vary the gain of this ideal one tell u find the optimum gain to have min error and stable circuit at the same time
u can make the ideal one using veriloga or try to find it in the libs of cadence "i think it is available "

17th May 2006, 06:01 #8
 Join Date
 Nov 2005
 Posts
 136
 Helped
 11 / 11
 Points
 2,872
 Level
 12
Stability analysis
Well I am using the circuit as an LVDS transmitter. Its a commonly used circuit. I have to use a resistor to terminate as specified by LVDS standards. Its a differential circuit.
There is no NMOS load. It is also an input stage. Basically its acting like a switched current source.
Sorry for the pic. The Rc is still there but masked by the font.
I will try using a current mirror and see if it works fine.
Thanks to all for the reply.

18th May 2006, 03:55 #9
 Join Date
 Mar 2006
 Posts
 18
 Helped
 0 / 0
 Points
 1,306
 Level
 8
Re: Stability analysis
5) Instead of your current way of probing, I would insert a voltage source probe between the output of the CMFB circuit and the point where it is feeding to the gate of the PMOS current source load.
Bharath[/quote]
Why insert a voltage source probe there is better?

6th September 2007, 10:03 #10
 Join Date
 Apr 2007
 Posts
 40
 Helped
 0 / 0
 Points
 1,224
 Level
 7
Stability analysis
i am a little confused with your picture ,why LVDS output has no loading cap ,does it not contribute to the phase shift ?

Advertisment

9th March 2008, 16:28 #11
 Join Date
 Jul 2007
 Posts
 4
 Helped
 0 / 0
 Points
 1,628
 Level
 9
Re: Stability analysis
problems about the simulation of CMFB is always hot, maybe the teachers in univ should be responsible for this, they seldom taught the handon simulation tricks (maybe they just don't know,)

1st April 2008, 08:31 #12
 Join Date
 Feb 2008
 Location
 China
 Posts
 403
 Helped
 54 / 54
 Points
 3,875
 Level
 14
Re: Stability analysis
Originally Posted by suhas_shiv
+ Post New Thread
Please login