Resistive Voltage Divider - What's the alternative?

[malizevzek]

buffered voltage divider

Is there an alternative to resistive voltage divider?
In the case when power dissipation through the resistors (or their size) is not acceptable, what would be a way out and at which cost?

 

[ieropsaltic]

alternatives to using voltage divider

Some people use source followers as attenuators.

 

[ricklin]

voltage divider buffer

Use subthreshold PMOS string to do voltage divide? I guess

 

[Old Nick]

voltage divider alternative

Is there an alternative to resistive voltage divider?
In the case when power dissipation through the resistors (or their size) is not acceptable, what would be a way out and at which cost?

Plenty of alternatives I'm sure, but it depends on what you want the circuit to do. A resistive voltage divider, is a confuguration of components, it's purpose in a circuit is what's important, (attenuation, creating a reference etc.).

 

[malizevzek]

switched cap voltage divider

I was thinking more about creating a reference.
Can I do it with some other configuration, having a negligible power consumption, at the cost of precision, for example?

 

[ieropsaltic]

voltage divider

For high performance references, people usually use bandgap cells. However, this may not be an optimum solution at your case because it may consume considerable power and area.

 

[malizevzek]

pmos voltage divider

One example: LDO regulator for low voltage & ultra low power (uW). When I buffer voltage reference, I need a resistive voltage divider at the output. If the current through them should be low, they need to be large. Can I avoid that?

 

[LvW]

whats the divider

I am not really a specialist in LDO design, but I wonder if one could drop the voltage divider completely. In this case, the reference voltage should be identical to the regulated voltage (and not identical to a part of it). So what ?
Perhaps somebody else can comment to this approach.

After a while some doubts came into my mind concerning common mode properties of the error amplifier, if both inputs are at the regulated voltage level; and at the same time the amplifier supply is between ground and the unregulated input voltage. I think this will not work.

 

[rfsystem]

voltage divider, whats that used for,

To catch up your requirements:

You need a bandgap, a buffer which multiplies the bandgap to the the LDO input and the LDO.

To minimize the area for resistors you can use PMOS with Well connected to source to divide voltage. But it works only for integer division ratios. To reduce the power at all you can power cycle the bandgap and the multiplying buffer and S/H the LDO input voltage.

 

[ytliang]

voltage divider as an ldo

bare in mind that resistor voltage divider is temperature independent because their coefficient cancels out. You might lose this nice property if u were to choose a different approach.

 

[malizevzek]

resistive voltage divider

To catch up your requirements:

You need a bandgap, a buffer which multiplies the bandgap to the the LDO input and the LDO.

To minimize the area for resistors you can use PMOS with Well connected to source to divide voltage. But it works only for integer division ratios. To reduce the power at all you can power cycle the bandgap and the multiplying buffer and S/H the LDO input voltage.

Which configuration exactly do you have in mind when you talk about PMOS relisation? Why does it function only for integer division rations?

 

[drDOC]

cap divider site:edaboard.com

Maybe you can use a switched capacitor idea.

Sample the reference on a cap A. Place an empty cap B in parallel. Charge on A now has to divide over A+B. Voltage has been divided. But this requires some switches and linear caps. Don't know if it safes much power.

 

[rfsystem]

pmos voltage divisor

OK step by step:

PMOS BS-DG connection, so a PMOS diode connection with bulk(NWELL) tied to source

If you try to make a n+1/2 ratio you have to match

VTH+VDSAT1=0.5*(VTH+VDSAT2) some where

which result in

0.5*VTH=0.5*VDSAT2-VDSAT1

These expression changes over process, temp .... no way