How to ensure negative feedback factor more than positive?

[nuiscet]

In the circuit of bandgap , the feedback signal produced by the op amp returns to both of its inputs. so how to ensure that the negative feedback factor is more than the positive feedback factor ? Isn't necessary to ensure an overall negative feedback? or the bandgap circuit must be negative feedback and stabe?

 

[jcpu]

Re: How to ensure negative feedback factor more than positiv

Hello

feedback cueents into two different load to
generate + and - voltage into OPamp,
(1) 1 x BJT: whose dynamic resistance is Vt/Ie
(2) N x BJT + Re: whose dynamic resistance is N x Vt/Ie + Re

Solution: pass the negative feedback current
(which ever closer to 180 degree phase shift at low freq end)
through higher resistance arm among (1) or (2)

have fun!

 

[nuiscet]

Re: How to ensure negative feedback factor more than positiv

how about the factors in this figure?

 

[nuiscet]

The negative feedback factor is Betan=gm*(R1+1/gm2),here gm is the transconductance of the PMOS and gm2 is the transconductance of the transistor with area of nA.
The positive feedback factor is Betap=gm*1/gm1 , here gm1 is the transconductance of the transistor with area of A.
Are the factors right?which is more?how about the 1/gm1 and 1/gm2?

 

[jcpu]

Re: How to ensure negative feedback factor more than positiv

Hello:

without exact SPICE model, we can only say
(1) gm2~ n * gm1
which of R1+1/gm2 and 1/gm1 is larger?
Need to run SPICE to resolve it.

 

[pixel]

Re: How to ensure negative feedback factor more than positiv

Hello:
which of R1+1/gm2 and 1/gm1 is larger?
Need to run SPICE to resolve it.

you nornally design that R1+1/gm2 is much larger than 1/gm1.
value of 1/gm is small and you can not achieve 1/gm1>>R1+1/gm2 for large enough values of R. As Feedback at the right side is stronger it is connected to achieve phase of 0(opamp) + π(mosfet) =π (negative feedback).

 

[jordan76]

Re: How to ensure negative feedback factor more than positiv

Hello:

without exact SPICE model, we can only say
(1) gm2~ n * gm1
which of R1+1/gm2 and 1/gm1 is larger?
Need to run SPICE to resolve it.

Normally we choose the two branches current ratio to be 1:1 for better matching.
Since for bjt, gm=Ic/Vt where Vt is the thermal voltage, gm2=gm1 if the two branches current are equal. Hope it helps!

regards,
jordan76

 

[nuiscet]

What's the exact parameter for the bipolar transistor? gm=Ic/Vt ? Isn't undepending on the Is? here Is is the saturation current, and Is2=nIs1 since the area ratio of them is n:1.

 

[bageduke]

Re: How to ensure negative feedback factor more than positiv

Ic=A*Is*exp(Vbe/Vt)

gm=dIc/dVbe=(A*Is/Vt)*exp(Vbe/Vt)=Ic/Vt

right? Ic itself depends on Is and area of BJT

 

[huojinsi]

Re: How to ensure negative feedback factor more than positiv

bageduke's derivation is absolute right!

 

[nuiscet]

So we can get a conclusion of 'gm2=gm1' if the two branches current are equal just as jordan76 said , and gm*(R1+1/gm2) > gm*1/gm1 must be right with the matched size of mirror PMOS. An overall negative feedback can be ensured.

 

[jcpu]

Re: How to ensure negative feedback factor more than positiv

Yes Sir:

What pixel said was right.
checking previous designs
all the R1+1/gm2 goes to (+) input of the OPamp
just as in your attached schematic.