How this bandgap works? [from IEEE paper]

[mike_bihan]

Hey, folks:

I can not understand how this Bandgap works. How the output could be accurate?

according to the paper, the vref=~(VBE3+Im3*R2)*R3/(R2+R3).

Pls give your advice. Many thanks.

 

[ambreesh]

Look that in this circuit the temperature range is small.
Has large variation in terms of TC.
Also I doubt if circuit could work at high temperature unless we have twin tub and very low Vt NMOS input device.
i have simulated it but it is more of a hit and trial as i foud out.

Added after 2 minutes:

Also one cannot go below diode voltage drop as reference voltage

 

[eda4you]

The left part is a "normal" band gap reference with a simple opamp-feedback, but what the hell should do the right part? is there also a start up circuit drawn?

 

[mike_bihan]

I agree that there is also a lot of draw back.

The most important thing I wondered is that how the equation is given?

It looks like that author believes that Iptat applied to a structure as "R3 in parallel with R2/B3" can generate accurate reference?

 

[nEuron]

hi mike_bihan,

u can obtain the final equation by solving the following equation:

Vref/R3 + (Vref-VEB)/R2 = IM3

 

[Fom]

If solve above equations you will get:

Vref= ((R2/R1)*Vt*ln8+Vbe)/(R2/R3+1)

You can select R2/R1 to make it temperature independent and select R2/R3 to make Vref lower than 1.2V.