Curvature-Compensated BiCMOS Bandgap with 1-V Supply Voltage

[crystalballs]

curvature compensated bicmos bandgap

Hi all,

I'm trying to do the design described in this Malcovati's paper (attached):
"Curvature-Compensated BiCMOS Bandgap with 1-V Supply Voltage"

The circuit schematic is this:

**broken link removed**

Does anyone has already done a similar design?

Thanks in advance.
CBs

 

[amriths04]

band gap bicmos

yes. what is your doubt?

 

[crystalballs]

Re: Curvature-Compensated BiCMOS Bandgap with 1-V Supply Vol

Hi amriths04 and thanks for you answer.

Do you know that paper? Have you read it?
I'm trying to design that circuit, following the equations described but it seems no possible to reach a performance below 10ppm/^C.
I'm using a 0.35 CMOS tech, not a BiCMOS, but other papers describe about the same circuit reaching about the same performances (about 7 ppm/^C).
Moreover, starting from the simplest 1st order compensated circuit descibed in fig 1 it is hard to reach 20ppm/^C.

If you have already done this kind of design what's for example your values for the current in the BJTs and the resistances?

TIA,
CBs.

 

[amriths04]

Re: Curvature-Compensated BiCMOS Bandgap with 1-V Supply Vol

Hi amriths04 and thanks for you answer.
Do you know that paper? Have you read it?

I know about this paper, but have not read it.

I'm using a 0.35 CMOS tech, not a BiCMOS,

.
That matters. Generally for cheaper design, people use parasitic pnp from a CMOS technology. If you use a BICMOS you get good beta for your BJTs.

but other papers describe about the same circuit reaching about the same performances (about 7 ppm/^C).

.
Did you try simulating across corners and tried to see which is the best ppm/^C you get?
Did you try using high impedance(cascoded) current mirrors for proper mirroring?
Did you choose the resistance values properly? It's critical.

If you have already done this kind of design what's for example your values for the current in the BJTs and the resistances?

Anyway I did it long back, i don't remember. I think the current through each branch was some 30uA. The resistances were some where in Kohms and Mohms. I used 180nm technology. Got some 12ppm/^C.



TIA,
CBs.