bandgap important problem

[alasliu]

1 how to improve PSRR for bandgap using OP
2 if resistor ratio is not integer, how to do layout. for example ratio is 11.2

 

[eng_Semi]

For the second question, you have to choose a unit resistance so that both resistors can be composed of it

For example, if the resistor values is R1=1k ans R2=11.2k
then if u choose 1k to be ur unit resisiance,
R1 will be composed of one resistor
R2 is 11 units in series + 5 units in parallel series with the previous ones

 

[kobeghost]

as for the PSRR,I once met a problem,I use a ac source in the VDD source,however,from the magnitude and phase of the output I got a result graph which is reverse with the expected graph,and I do not know how to settle this problem,please anyone who have done the simulation before give some advise or list some paper about the psrr simulation,many thanks.

 

[abcyin]

Using the cascode structure for the current mirror in the bandgap design is a way to increase the PSRR.

for kobeghost, what do you mean the simulation result is reverse with the expected graph? how is your simulation results, can you post it here?

 

[kobeghost]

thank you abcyin for taking care my question,what i mean is:the amplitude of |PSRR| should decrease with the frequency increase(from ALLEN's book),but my result is just the opposite and the graph seems to be unnormal,which confused me much.

 

[abcyin]

thank you abcyin for taking care my question,what i mean is:the amplitude of |PSRR| should decrease with the frequency increase(from ALLEN's book),but my result is just the opposite and the graph seems to be unnormal,which confused me much.

it's strange that the PSRR will present such trend, as to the theory, the PSRR should vary just like Allen's book suggested, I suggest that you can first check all the devices in your schematic, maybe some devices are not in the right working condition.

hope it helps!

 

[alasliu]

thanks for you advices

 

[smilelangjun]

use the feedback with amp can make psrr much better

 

[mario1980]

kobeghost : Could it be that at the output of the opamp in your schematic, you have a PMOS driving the R/PNP loop, and the compensation capacitor(at the output of the opamp) is to VSS? If it is, than do yourself a favor and put it to VCC, so that PMOS to have constant Vgs.
Another way to increase PSRR, is to increase the gain of the opamp.
And regarding the PSSR curve, it decreases with f, until a point (where the loop gain tends to 1), and after that it starts increasing again.

Regards!

 

[gevy]

use the feedback with amp can make psrr much better

And to increase gain of amp.

 

[caosl]

1.increase the gain of OP
2.decouple the power supply
3.add decouple cap between Gate and Souce,surpose PMOS regulate transistor you use
4.pre-regulate you power supply, then supply you bandgap circuit.

I guess these methods can make you PSRR met you spec

 

[leo_o2]

I don't think increase OP gain is a good way to improve PSRR. In many cases, you will find PSRR becomes worse or unchanged though OP gain is increased. I think PSRR is much more dependent on the structure you employed.