I'm trying to design a bandgap reference circuit with reference voltage 1V and TC=0 from -20°C to 100°C. I've generated the reference voltage close to 1V but struggling to make the TC zero. I have less idea about how to make the TC very much close to zero. I'm using the circuit shown in the attached picture. Can anyone help me?

 

I'm trying to design a bandgap reference circuit with reference voltage 1V and TC=0 from -20°C to 100°C. I've generated the reference voltage close to 1V but struggling to make the TC zero. I have less idea about how to make the TC very much close to zero. I'm using the circuit shown in the attached picture. Can anyone help me?

Hi,

What theories are you working with so far that got you to that circuit?

 

Hi,

What theories are you working with so far that got you to that circuit?

I saw some youtube videos, read some research papers. I found the circuit in a research paper.

 

I saw some youtube videos, read some research papers. I found the circuit in a research paper.

It would help to see that paper.

 

quite close https://repository.iiitd.edu.in/jspui/bitstream/handle/123456789/640/MT16100_NEHA DALAL.pdf?sequence=1&isAllowed=y
and another https://www.researchgate.net/public...h_an_Inherent_Curvature-Compensation_Property , https://www.researchgate.net/public..._bandgap_reference_design_techniques_a_survey

 

quite close https://repository.iiitd.edu.in/jspui/bitstream/handle/123456789/640/MT16100_NEHA DALAL.pdf?sequence=1&isAllowed=y
and another https://www.researchgate.net/public...h_an_Inherent_Curvature-Compensation_Property , https://www.researchgate.net/public..._bandgap_reference_design_techniques_a_survey

Hi,

Ok i'll see if i can read all that and spot something that could help.

 

I'm trying to design a bandgap reference circuit with reference voltage 1V and TC=0 from -20°C to 100°C.

Are you trying to do this with discrete circuits?

There is no way to get to TC=0 with any real circuit, it's just how close do you really need.
And you are highly unlikely to do better than the IC references already available.

 

Are you trying to do this with discrete circuits?

There is no way to get to TC=0 with any real circuit, it's just how close do you really need.
And you are highly unlikely to do better than the IC references already available.

Hello there,

Sometimes we refer to a zero TC reference when it is just very good even if not perfect.
Since this is the homework section though, it may be possible in pure theory.

I grew up using thermistors and resistors to compensate meters. That can really be a pain having to do all the tests in an oven that can be set to different temperatures. Sometimes took hours or days. At the time, the military applications would not allow anything fancy they wanted tried and proven techniques only.

Oh and also in one of those papers the voltage stability is touted to rival some crystals frequency stability. if it works it would represent a new age of voltage references. I take it that is why there is interest in this area.,

 

quite close https://repository.iiitd.edu.in/jspui/bitstream/handle/123456789/640/MT16100_NEHA DALAL.pdf?sequence=1&isAllowed=y
and another https://www.researchgate.net/public...h_an_Inherent_Curvature-Compensation_Property , https://www.researchgate.net/public..._bandgap_reference_design_techniques_a_survey

Hello again,

If you read one of those papers you will see that the goal is to get a voltage temperature stability equal to some crystals frequency stability. That's amazing if it can work out.

But what is it that you found did not work and how are you testing this?
Also, you should know that the component matching has to be super perfect, and that's not the end of it, one or more of the resistors need trimming to get everything working right.

It would help to know what you are having a problem with though.

 

i happened to read an article about designing a state of the art voltage reference that were some 2x 3x more complex than the 1 OP posted and there was a resistor at a specific place in circuit that was used to T-compensate the entire thing . . . i might have got an unauthorized access to it . . . or i just cannot find it - my old 'puter got crashed . . . some archives went lost . . . and i didn't payed too much attention to it at the time . . . or (it's stored at the wrong folder/bookmark . . . must scan the old IE ones . . . )

written by some old school guy . . . over his 50-s at the time - i guess . . . some 90 pages or more i guess

 

i happened to read an article about designing a state of the art voltage reference that were some 2x 3x more complex than the 1 OP posted and there was a resistor at a specific place in circuit that was used to T-compensate the entire thing . . . i might have got an unauthorized access to it . . . or i just cannot find it - my old 'puter got crashed . . . some archives went lost . . . and i didn't payed too much attention to it at the time . . . or (it's stored at the wrong folder/bookmark . . . must scan the old IE ones . . . )

written by some old school guy . . . over his 50-s at the time - i guess . . . some 90 pages or more i guess

Hi,

Might be interesting to see. I came up with a method myself some years ago but it's hard to implement.
There are several bipolar designs out there that are pretty good though using maybe 5 bipolar transistors. They are usually in the range of 1 percent over the normal temperature range which is good enough for a lot of things.
This new type is really interesting though with an unheard of voltage stability.

I took the easy way out though as Cruts suggests, i purchased a voltage reference IC chip for meter calibration that is really super accurate and does the trick for most regular meters. If i remember right the output was 4.096 volts and i tested it with two different very very expensive 5 1/2 digit meters and they agreed to within 0.001 volts. That voltage makes it great for an A/D converter reference, which was one of the main intended uses.

NOTE:
Actually they were 5 3/4 digit meters and they agreed to within 0.0001 volts. It was some years ago.
Supposedly the aging factor for those references was pretty good too.