Why is it around 1.2V for bandgap design ?

[PatrickLee]

I am new to bandgap design.
I noticed that many of the bandgap design regulate at around 1.2V.

What determined this value ?

Is the temperature compensation work best at this voltage point for best cancellation of positive and negative temperature coefficient ?

How to simulate the best compensated point in SPICE? Do temperature sweep ?

 

[alok_ky]

The 1.2V for bandgap design is dictated by the bandgap voltage of silicon and the zero temperature coefficient criterion.

 

[zenisle]

we often design 1.24V for bandgap.

 

[nuiscet]

Very good for your noticing this.
Actually The 1.2v is determined by the bandgap working principle. The intent of designing a bandgap voltage/current is to acquire a voltage/current without dependance on temperature.We knows that the voltage between a diode Vbe is negative temperature dependance. so usually we use another voltage/current with positive temperature dependence to acquire the voltage/current without dependence on temperature. But Vbe is converging to 1.205v at the kelvin zero degree. so the voltage without temperature dependence is just around 1.2v.

 

[PatrickLee]

Hello, Nuiscet,

So that means the fundamental voltage : 1.205V (as you mentioned at 0K) cannot be changed ?
The 1.205V is independent of any wafer foundry process? Right?

So bandgap design is to achieve voltage of 1.205V.

So what is the design criteria during "design phrase".
What should be look for during SPICE simulation for a given set value of R1,R2,R3, AE1, AE2 (parameters as described in Allen and Holberg)?
Should we modify the values and do temperature sweep to get the optimum result ?

 

[nuiscet]

I think the 1.205V is independent of any wafer foundry process. But actually we can't get the accurate 1.205V because that Vbe isn't linear with temperature changing, Vptat is almost linear. So the sum of Vbe and Vptat isn't accurate value. Vbe=kT/q*lnIc/Is, Ic and Is are both changing when temperature is changed. So actuall output is just around 1.2v.

 

[pi331133]

I think 1.2 V is an old concept of the original bandgap reference design, in fact, in current CMOS process, use the parasitic PN junction and current mode circuitry, the temperature independt reference is not limited to 1.2V

Regards,

Added after 2 minutes:

There is a book focus on the bandgap reference design named "Voltage References From Diodes to Precision High Order Bandgap Circuits" IEEE Press. Also many papers are published on bandgap design, it is a big topic. Thanks.

 

[Robert Qi]

I have designed a bandgap ckt with 0.18um TSMC tech and achieved a 1.26v voltage reference. The bandgap gets its' zero temp coefficient at about 40 degree. The circuit has been used in our product CIS (CMOS Image Sensor) and works well.
Just for your reference.

 

[sharas]

Very good for your noticing this.
Actually The 1.2v is determined by the bandgap working principle. The intent of designing a bandgap voltage/current is to acquire a voltage/current without dependance on temperature.We knows that the voltage between a diode Vbe is negative temperature dependance. so usually we use another voltage/current with positive temperature dependence to acquire the voltage/current without dependence on temperature. But Vbe is converging to 1.205v at the kelvin zero degree. so the voltage without temperature dependence is just around 1.2v.

How foes Vbe converge to 1.205V at the Kelvin Zero degree?
Doesn't Vbe depend linearly to KT/q??

 

[|IAngel|]

No, Vbe is dominated by the diode formed at the b-e junction. This junctions has a CTAT behavior and a nearly linear dependance of around -2mV/°C. Although KT/q is still a factor, it is tiny compared to the other device temperature dependancies.