Bandgap Reference Topic

[dkace]

Hi all!

I want to design a a relatively simple Bandgap reference in CMOS technology. I went through all the literature I could find - Razavi included- but I still have some unsolved issues:

1. It is mathematically proved that Vref=Vout of the OPAMP. I have seen topologies that Vref is the left input of the OPAMP. This contradicts the basic equation that wants TC=0 : Vref=Vbe + Vt Ln n* (1+R2/R3). Any idea?

2. Somebody has to explain me the issue with the OPAMP; We use the OPAMP to maintane equal inputs ( V+=V-) . Then what is the output of the OPAMP and how it corrects any small difference?

3. Second diode has a n - times larger area than first. What is the effect on the current? Which resistor is used to put the current same as reference current? Resostors are teperature dependent. What happens with them in this cell?

I would appreciate if you can back up your answer with maths or references. I am still working on the circuits equations - but I am still a bit rusty with design.

Thanks,

D.

 

[PaloAlto]

I'll try to answer you latter with some more time. But if you want my advice. Keep working a little bit in the circuit equations.

I think that you can still learn the principle better. Put the equations on the BG core and go through them again.

Remember that the bandgap core produces a voltage, so you have to compensate for the TC of resistors when creating currents.

The basic idea is that an area difference in the bjts creates a PTAT voltage across a resistor. Somewhere else you create a voltage that is 1/T and add them. To do the addition, you use currents.





[...................]

After reading it I've realized that I haven't at all explained my self well. I'll try later unless someone else does it better :-(

 

[dkace]

Thanks for the tips, but I have understand so far. My problem is that I am trying to understand how it really works the proposed configuration:

I.E. we have two equal resistance as load. From there we derive equal voltages to both branches ( Say Io).

R3, the resistor of the second BJT is responsible for keeping Io instead of multipling it to n*Io? What do I miss here?

Thanks,

D.

 

[PaloAlto]

I don’t know the exact BG circuit that you are using, but try to identify where the voltages are generated, the PTAT and the 1/T voltages.
My guess is that the currents are equal, the voltage at the input of the opamp is almost 0, so the resistor next to the bipolar is suffering the PTAT voltage and converting it into a current (that includes the res TC). The nominal value of that resistor sets the nominal value of the current. The ratio of that resistor with some other one in the circuit, (the one that sees the 1/T voltage) makes the voltage flat, so, playing with that ratio you can compensate the TC of resistor too. Actually, you could also make the current (and not the voltage) flat with T.

 

[abcyin]

hi, dkace
i think that you can post the bandgap topology you used here, and then we can analyze it together.