How to design a voltage divider?

[leonken]

input voltage:1.2V, do not use some big resistor divider,
how to generate a 0.82v,0.75v output?
Thank you!

 

[atmaca]

search for linear scaler realized by cmos'

 

[mikem]

input voltage:1.2V, do not use some big resistor divider,
how to generate a 0.82v,0.75v output?
Thank you!

What's wrong with resistor divider and buffer ?? If you are trully after a circuit that divides one voltage into another voltage, then you will need to look at using a analogue voltage multiplier IC in division configuration. good luck

 

[leonken]

search for linear scaler realized by cmos'

I searched but I didn't find any material, can you help me to provide me some website? thank you!

 

[leonken]

input voltage:1.2V, do not use some big resistor divider,
how to generate a 0.82v,0.75v output?
Thank you!

What's wrong with resistor divider and buffer ?? If you are trully after a circuit that divides one voltage into another voltage, then you will need to look at using a analogue voltage multiplier IC in division configuration. good luck

Consider the low power operation, resistor divider network is not a good choice.
so, I want to find some good approaches.

 

[mikem]

input voltage:1.2V, do not use some big resistor divider,
how to generate a 0.82v,0.75v output?
Thank you!

What's wrong with resistor divider and buffer ?? If you are trully after a circuit that divides one voltage into another voltage, then you will need to look at using a analogue voltage multiplier IC in division configuration. good luck

Consider the low power operation, resistor divider network is not a good choice.
so, I want to find some good approaches.

You need to be more specific ! Are we talking nA of current drain ? A couple of tens of µA drain would not be difficult and impractical to achieve with R-divider. If you're truly requiring nA current drain, then I would still consider R-divider (because of simplicity) and concentrate on PCB layout issues with minimizing leakage currents around high-Omic resistances and/or possible conformal coating of PCB to reduce moisture and hence leakage currents ... however this gets way too messy ... Basically I'm asking how long is a piece of string ? µA, nA or pA ?

 

[leonken]

input voltage:1.2V, do not use some big resistor divider,
how to generate a 0.82v,0.75v output?
Thank you!

What's wrong with resistor divider and buffer ?? If you are trully after a circuit that divides one voltage into another voltage, then you will need to look at using a analogue voltage multiplier IC in division configuration. good luck

Consider the low power operation, resistor divider network is not a good choice.
so, I want to find some good approaches.

You need to be more specific ! Are we talking nA of current drain ? A couple of tens of µA drain would not be difficult and impractical to achieve with R-divider. If you're truly requiring nA current drain, then I would still consider R-divider (because of simplicity) and concentrate on PCB layout issues with minimizing leakage currents around high-Omic resistances and/or possible conformal coating of PCB to reduce moisture and hence leakage currents ... however this gets way too messy ... Basically I'm asking how long is a piece of string ? µA, nA or pA ?

The nA current is acceptable for me,because I am designning a very low power IC.And I want to integrate this subcircuit on one chip.
The power supply voltage is 3V. If R-divider is selected, you can see,1uA current needs 3megohm, 1nA needs 3000megohm. so ,how to realize it in one chip?
So, I need a special approach.

 

[mikem]

input voltage:1.2V, do not use some big resistor divider,
how to generate a 0.82v,0.75v output?
Thank you!

What's wrong with resistor divider and buffer ?? If you are trully after a circuit that divides one voltage into another voltage, then you will need to look at using a analogue voltage multiplier IC in division configuration. good luck

Consider the low power operation, resistor divider network is not a good choice.
so, I want to find some good approaches.

You need to be more specific ! Are we talking nA of current drain ? A couple of tens of µA drain would not be difficult and impractical to achieve with R-divider. If you're truly requiring nA current drain, then I would still consider R-divider (because of simplicity) and concentrate on PCB layout issues with minimizing leakage currents around high-Omic resistances and/or possible conformal coating of PCB to reduce moisture and hence leakage currents ... however this gets way too messy ... Basically I'm asking how long is a piece of string ? µA, nA or pA ?

The nA current is acceptable for me,because I am designning a very low power IC.And I want to integrate this subcircuit on one chip.
The power supply voltage is 3V. If R-divider is selected, you can see,1uA current needs 3megohm, 1nA needs 3000megohm. so ,how to realize it in one chip?
So, I need a special approach.

Have you considered metal-oxide resistors on silicon substrate ? I belive digikey have them on their stock list ... but don't quote me on that. good luck !

 

[jake]

input voltage:1.2V, do not use some big resistor divider,
how to generate a 0.82v,0.75v output?
Thank you!

You can use a LDO to realize them
the paper"Analysis and Design of Monlithic, High PSR, linear Reulators for SOC Applications" maby help you.

 

[itheman]

One way of doing this would be to use MOS transistors as resistors. Use
a MOS device with a Laerge L and minimum W., to get a large R.

 

[JPR]

What is the application? There are a couple of ways that I would consider to create a very low current voltage divider - switched capacitor and switched resistor. Both would have sampling characteristics. A continuous time divider would take more effort, and I will not go into that detail unless needed:

If you expect the input signal to be slow changing, try resistor divider that is powered for a small percentage of the time:

Example:
If you use a 300kOhm total resistor divider, which is connected for 1uSec every 10mSec, with the output held on a capacitor. This will give you a current draw of 10uA for 1uSec that occurs once every 10mSec. This will cause an average current of only 1nA. The capacitor will hold the value between samples from the resistor divider. The ability to integrate this solution would depend upon the size of the resistors needed and the size of the holding capacitor needed.

A switched capacitor circuit may be able to be run at a faster rate, yet still acheive low enough current. By using small capacitors and switching at the lowest rate you can reliably use, you might be able to keep a low power consumption. Power consumption will depend upon the accuracy that you need (size of capacitor to use) and the rate that it needs to be switched. The power goal that you have may or may not be attainable at the current that you desire. The real advantage here is in area. Smaller current would be achieved by a physically SMALLER capacitor. The trade-off will be on noise and accuracy of the result. Noise can be reduced by increasing the size of the filter capacitor at the output. There will be a limit to the accuracy that will only be acheieved by making the switching capacitors bigger, causing more current draw.

 

[wued]

if switching rate is higher, the power consumption is higher.

 

[Btrend]

if the current consumption is nA scale, using switching cap like method is no good,
because there is charge injection, feedthrogh and settling problem.
I think possible solutions are:
1. using high resistance poly or make some special ultra high resistance type resistor.
2. using capacitor in series

 

[JPR]

I agree that nA scale is difficult to attain. As I said before, it may be possible with switched cap if the frequency response is slow. If the signal bandwidth is less than 50 Hz, for example, you can clock your switched capacitor circuit at 200Hz (could be as low as 100Hz, but you might start to have aliasing issues.)

At 200Hz, you could switch 200pF at 1.2V and still maintain 50nA. I would suggest that if extreme accuracy is not needed, the circuit could probably be done with something on the order of 5-20pF, or even less, which should be able to be done in less than 10nA. This all assumes that a SLOW clocking frequency is okay (which has yet to be confirmed.)

If the goal is to divide a DC signal into a smaller voltage, say a bandgap reference voltage being divided down for a comparitor reference, the bandwidth of interest is all EXTREMELY slow, and the clock rate can be adjusted to keep the current consumption low.

I am not claiming that this approach would be the best for all requirements, but it CAN be made to run at very low current if a very low clocking speed is usable. The impact from switch feedthru is essentially the same, no matter what the clocking frequency is used. There will be other factors at such low clock speeds (leakage currents) that will also impact accuracy, especially at elevated temperatures.