Help me decrease the resistor in opa design for bandgap

[jeremy_zhu]

1、i‘m designing an opa for bandgap，using miller comp tech for getting a PM of around 60. but my nulling resistor has to be around 100k.is it ok for being implemented by CMOS process? as a resistor,it seems to be too large.any other methods to decrease the resistor?

2、opa gain >70db & GBW is about 400KHz,the GBW is ok for bandgap?

ckt is attached here!

 

[edajason]

Re: opa for bandgap

If Your gain is 70dB, and -3dB BW is 400kHz, that may be okay. If your GBW is 400k, then not good for PSRR consideration.
For BG, the higher GBW is needed to supress power supply noise.

 

[jeremy_zhu]

opa for bandgap

my input differential pair works in subthreshold region and the gm is about 20uS, GBW=gm1/Cm,if Cm=5pf，GBW=4/2*pi Mhz=0.64Mhz. -3dB bw is only 213Hz. how to solve this problem?
and how about the resistor?

 

[gevy]

Re: opa for bandgap

1. You may place capacitor between Drain M5 and Drain M11 without resistor.
It is cascode Miller compensation.

 

[edajason]

Re: opa for bandgap

20uS Gm looks a little bit small. increase gm further.
1) on second look at your topology, you do not have to add second stage. If your supply permits, you can add a cascode for your bottom transistors, M3, M8, M9. Basically you are using a folded cascade ota, but without matched output Rp and Rn.
2) In this case, you even do not need R-C compensation.

my input differential pair works in subthreshold region and the gm is about 20uS, GBW=gm1/Cm,if Cm=5pf，GBW=4/2*pi Mhz=0.64Mhz. -3dB bw is only 213Hz. how to solve this problem?
and how about the resistor?

 

[jeremy_zhu]

Re: opa for bandgap

20uS Gm looks a little bit small. increase gm further.
1) on second look at your topology, you do not have to add second stage. If your supply permits, you can add a cascode for your bottom transistors, M3, M8, M9. Basically you are using a folded cascade ota, but without matched output Rp and Rn.
2) In this case, you even do not need R-C compensation.


if i have not the second stage ,i am afraid that my gain is not enough because of my subthreshold input stage(small gm).increasing aspect ratio of the input stage or enlarge the tail current of M3 may increase the gm,but the foemer seems not work (now the W/L is already 240/2),i will try to increase the tail current for the cost of more dissipation .
i can't use the cascode topology because of the low supply voltage (1V less)

place capacitor between Drain M5 and Drain M11 without resistor.does it work? i
will try !

 

[mists]

Re: opa for bandgap

place capacitor between Drain M5 and Drain M11 without resistor.does it work? i
will try !

it work, it is known as ahuja compensation, try to refer a JSSC paper on 1983

 

[jeremy_zhu]

Re: opa for bandgap

it work, it is known as ahuja compensation, try to refer a JSSC paper on 1983

thank you for your ideas ! it seems work for decreasing my big resistor,but a lillte effect on my GBW even makes the GBW less!
BTW, for a bgr ,what's the minimum -3 db BW or GBW should we have ?

 

[fanrong]

Re: opa for bandgap

You can use mos transistor as resistor or voltage buffer or current buffer

 

[hazemafify]

opa for bandgap

I think you have two ways to decrease your resistance
1-increase L of MOS.
2- decrease the current in the mirror.

 

[jeremy_zhu]

Re: opa for bandgap

1-increase L of MOS. 2- decrease the current in the mirror.

thanks, the 2nd method works & decrease about 50% resistance in my ckt

 

[beckchm]

Re: opa for bandgap

GBW is too small, but in DC operation is OK

But the freq characterics is not enough

 

[jeremy_zhu]

opa for bandgap

for this topology,are there any other useful ways to increase the GBW except the above methods ?

 

[pmr9]

Re: opa for bandgap

Hi , actually u dont need to use 2nd stage and put cascode connection in the first stage in ut FC opamp. Then no need for compensation ,if o/p drives less cap load use load cap at o/p that plays compensation role so that u dont need any resistor now.