need help on bandgap vdd=2.5~6v

[sunnyboy]

hello,everyone!

I am a freshman to Analog IC Design, now i want to design a bandgap circuit which vdd will change from 2.5v to 6v use CMOS tech
as i know: CMOS bandgap design have
1 use OPA + diode

2 use cascode mos + pnp diode


but cascode bandgap need Hi vol
so it seems that OPA + diode is the only way ,but i have a bandgap use OPA + diode ,but vdd only change from 2.0v to 2.8v

can anyone give me some help?
thanks

 

[melodyseu]

Try to use 2-order cascode+OPA+PNPdiode,and a proper bias can help you

 

[qutang]

can you add some layer?
use a depletion &enhancement NMOS to design the reference.use pmos as pass element.it works fine.

 

[electronrancher]

haha - NO, i don't think you are allowed to add extra layers whenever you want them!! you're funny!


back to the question - any small op-amp should be able to operate from 1.5v-6v with no problem at all.

so if you are in cmos tech and only have pnp, use pnp to ground - no problem. 8:1 pnp ratio for 54mV VPTAT, 5.4k Rptat for 10uA, and you can find the proper resisor to gain Vbe up to Vbg.

Build the op-amp separate from the bandgap. Make sure the common mode and VDD works in your range. Go for about 60dB DC Gain, and make sure to compensate for 50-60d phase margin. Once you have this op amp, build the bandgap, i guarantee it will work.

 

[qutang]

but the pnp transistor has less gain.

 

[ysz]

haha - NO, i don't think you are allowed to add extra layers whenever you want them!! you're funny!


back to the question - any small op-amp should be able to operate from 1.5v-6v with no problem at all.

so if you are in cmos tech and only have pnp, use pnp to ground - no problem. 8:1 pnp ratio for 54mV VPTAT, 5.4k Rptat for 10uA, and you can find the proper resisor to gain Vbe up to Vbg.

Build the op-amp separate from the bandgap. Make sure the common mode and VDD works in your range. Go for about 60dB DC Gain, and make sure to compensate for 50-60d phase margin. Once you have this op amp, build the bandgap, i guarantee it will work.

Hi,electronrancher
Could you tell me how to reduce the Vos of amp? thank you very much!

 

[rtpq]

Current-mode will help you to operate at a very low supply voltage like 2.5V. A low-voltage cascode current mirror topology will help more and improve PSRR.

 

[rtpq]

Usually it would not be easy to design rail-to-rail opamp with uniform gain with very low supply voltage.

haha - NO, i don't think you are allowed to add extra layers whenever you want them!! you're funny!


back to the question - any small op-amp should be able to operate from 1.5v-6v with no problem at all.

so if you are in cmos tech and only have pnp, use pnp to ground - no problem. 8:1 pnp ratio for 54mV VPTAT, 5.4k Rptat for 10uA, and you can find the proper resisor to gain Vbe up to Vbg.

Build the op-amp separate from the bandgap. Make sure the common mode and VDD works in your range. Go for about 60dB DC Gain, and make sure to compensate for 50-60d phase margin. Once you have this op amp, build the bandgap, i guarantee it will work.

 

[nickooeda]

u can find some useful paper in IEEE for LVLP bandgap circuit.good luck

 

[sutapanaki]

You all seem to concentrate on the difficulties of building the opamp at low supply voltages. It is also not easy to built it to work at high voltages because of the SOA of transistors. If the guy has a CMOS that can normally work at 3.3v, then the transistors will not sustain 6v. And what if the technology is 0.18? Well, it is true, it is not said in the initial post what the technology is.
If it is for example 0.35u you can build an opamp working at 2.5v. Then use some sort of a clamp for the supply voltage of the BG - for example follow the supply if it changes from 2.5 to 3.3 and then limit it to 3.3 - 3.5 for the part from 3.3 to 6v.
This will also reduce the sensitivity to supply variation (PSRR) for the second half of the voltage range. It can do the same also for the first half of it if there is proper filtering.

 

[electronrancher]

?? I like a certian half micron process with Vth=0.7v. I can build a 1.5-6v op amp with no problem. You probably don't need 0.35u for this application.

ysz - if you have VOS in simulation, I bet you have mismatched tail current to common source current in simple miller op amp. tail current must be twice the common source stage if nmos are the same size.
pm me

 

[electronrancher]

sorry - i guess i should mention. high swing folded cascode is really not hard to build, that was my choice for this application. plus, you know your output will be about 1.25v so you have plenty of headroom at 2.5v..

 

[hhun]

you can look "RAZAVI"-design of analog cmos integrated circuits
l design a bandgap is "2v~12v to dc 3.3v"

 

[Colbhaidh]

If you run 0.35um at 6 volts, you are asking for reliability problems and field returns. It like driving through a red light, no matter how many times you succeed, there is one time you will not.

 

[segabird]

I agree sutapanaki's opinion because the spice model maybe not correspond with the mos's action if the supply voltage is larger than 4v in 0.35um process.

 

[codec]

cascode mos + pnp diode is a good choice.
I have designed a BGR with 0.005V variation thru 2.3V~5.8V.
It is easier If you can use native tranisitors.

 

[sunnyboy]

my process is 0.6um
can anyone give me a example?

 

[gevy]

You can find the useful information on a bandgap here www.circuitsage.com
Usually I use two various vertical P-N-P transistors as diodes, operational amplifier with feedback and sturt-up circuit.

 

[gevy]

You can find the useful information on a bandgap here www.circuitsage.com
Usually I use two various vertical P-N-P transistors as diodes, operational amplifier with feedback and sturt-up circuit.