questions about band-gap design

band-gap design

1. When designing a band-gap reference circuit, I only consider the temperature,PSRR,and start-up circuit,is there anything else that I should consider?
2. Considering PVT, is there any way to generate a high precision current reference?

Re: band-gap design

yes these are main features of band gap reference:

constant output voltage
PSRR
Noise analysis
output current
Quiscent current

first possibly complete structural design soon. then do more characterization

Re: band-gap design

meleo1988 said:

Considering PVT, is there any way to generate a high precision current reference?

Click to expand...

With a bandGap reference circuit a PVT-independent voltage is generated. Then you only need a T-independent resistor, which can be approximated by an appropriate combination of NTC and PTC resistors. The achievable precision usually is limited by the accuracy of the T-compensation of the on-chip resistors (order of (several) 100ppm/K within the commercial temperature range).
Depending on the current value needed, an additional buffer or even a voltage regulator might be necessary. In this case the achievable precision is closer to 500 .. 1000ppm/K .

Re: band-gap design

I am not sure that a T-independent resistor is a must. The output voltage is Vbe+dVbe*k, where k and dVbe depends only on the relative values (not absolute ones). Vbe depends logarithmically on the current, which is dVbe/R.

But I totally agree that a resistor with no temperature coefficient would improve the bandgap's temperature dependence.

Re: band-gap design

tzg6sa said:

I am not sure that a T-independent resistor is a must. The output voltage is Vbe+dVbe*k, where k and dVbe depends only on the relative values (not absolute ones). Vbe depends logarithmically on the current, which is dVbe/R.

But I totally agree that a resistor with no temperature coefficient would improve the bandgap's temperature dependence.

Click to expand...

May be this is a misunderstanding, tzg,

you are right saying a T-independent resistor is not absolutely necessary in the bandGap circuit; the OP (meleo1988) however -additionally- asked for any way to generate a high precision current reference, and for this, of course, a low TC resistor - in connection with the bandGap reference - would be rather welcome ;-) .

Re: band-gap design

Hi meleo,
To get a high precision current reference with PVT you need to have a high precision bandgap reference. To get that you need to trim your part to remove the Process variations since a properly trimmed bandgap removes the temperature variations then all you need to design for is a bandgap circuit which is insensitive to the Supply variation. With this you have a PVT independent bandgap reference.
To generate a current either you can use a resistor off chip and specify its accuracy so all you need is a low offset buffer (offset should be determined by the precision you want on the reference after adding up the other errors like the bandgap variation and the resistor variation) and place the voltage over that external resistor to get your current.
If it has to be generated on Chip then you need a temperature independent resistor as erickl said and you also need to trim that to remove the process variations in that resistor. Now you should have precision current reference over PVT variations.

Re: band-gap design

you are right saying a T-independent resistor is not absolutely necessary in the bandGap circuit; the OP (meleo1988) however -additionally- asked for any way to generate a high precision current reference, and for this, of course, a low TC resistor - in connection with the bandGap reference - would be rather welcome ;-) .

Click to expand...

Sorry. I missed the "current" part.

Re: band-gap design

aryajur said:

Hi meleo,
and you also need to trim that to remove the process variations in that resistor. Now you should have precision current reference over PVT variations.

Click to expand...

How to trim to remove the process variation of the resistor? How could I get a resistor both with little process variation and good temperature independent?

Added after 4 minutes:

erikl said:

meleo1988 said:

T-independent resistor, which can be approximated by an appropriate combination of NTC and PTC resistors. .

Click to expand...

So u mean make a resistor using several kinds of resistors combination to get both temperature independent and process variation? What else do I have to pay attention to when doing the layout?

Added after 8 minutes:

girih192002 said:

yes these are main features of band gap reference:
Quiscent current

Click to expand...

So I think u mentioned Quiscent current as considering the Quiscent power and according to what I have read, the power depends mostly on the structural of band-gap. And if I want to design a utra low power band-gap, which structural could I consider?
thanks,

Click to expand...

Re: band-gap design

meleo1988 said:

erikl said:

... T-independent resistor, which can be approximated by an appropriate combination of NTC and PTC resistors. .

Click to expand...

So u mean make a resistor using several kinds of resistors combination to get both temperature independent and process variation?

Click to expand...

Regarding temperature compensation pls. s. this thread, especially my last contribution from Thu, 14 Jan 2010 1:14 .

Regarding process variation compensation, I think the only way is trimming, as aryajur stated above, at least if you need an absolute accurate current.

meleo1988 said:

What else do I have to pay attention to when doing the layout?

Click to expand...

Nothing else, if you provide trimming anyway. If not, try to match the resistor(s) with the bandGap's resistors.