How to get a common-mode voltage (Vdd/2) with high PSRR?

[rockycheng]

common mode voltage

Dear all,

Are there any special ways to obtain a Vcm voltage that is half the supply voltage? A simple voltage divider seems unable to restrain the noise from the power supply. Any suggestions are welcome! Thank you!

 

[MSSN]

high common-mode voltage

You can use a bandgap reference,for more information u can check razavi's book

 

[rockycheng]

comparator common mode voltage

But the output of a bandgap is a fixed voltage. The supply voltage could be different in different applications.

 

[renwl]

voltage reference supply, tracks vdd/2

is it possible?
if you want vcm=vdd/2, then the small interference signal with different frequency on power supply is also divided by 2.

 

[jonashat]

vdd/2

Isn't this why people use LDOs?

 

[renwl]

buffer good psrr

LDO output stable voltage.
and now, we need a common-mode voltage (Vdd/2) with high PSRR. it's proportional to vdd.

 

[Jepeto]

If you need the ½ VDD for OP biasing try this

 

[rockycheng]

Thank you for the replies!

Jepeto,
The buffer you used can increase the driving capability which is good. But the noise is still there, isn't it? Maybe adding some capacitance after the buffer will help.

Are there any better ideas?

 

[elbadry]

Since you need a Vdd/2 voltage, the DC rejection should be zero ( so the output can track the supply )

To improve rejection at higher frequencies, you can use high resistance values for the resistive divider and add an additional series RC filter. Something like shown below:

Vdd
-
-
R
-
---R(very high) --
- |
R C (large)
- |
- |
gnd gnd

The series R can be a very large MOS resistor. Depending on the needed attenuation value and the frequency where you need this attenuation, this technique may help.

 

[rockycheng]

Do you mean this figure? thanks!

 

[szekit]

how about an on-chip regulator biased off from bandgap? you can trim the resistor divider to get whatever voltage you want. as for high PSRR, you can increase the gain of the loop and add decoupling caps.

 

[liuyonggen_1]

you could use a LDO to get a voltage reference drived from a bandgap reference.
i think it's a very popular structure in LED applications.
the bandgap reference is applied to a unity gain amplifier, the other terminal named "A" of the amplifier is connected with a resistor, and an another resistor is connected between the output of the amplifie and the terminal "A". so, you could get a arbitrary voltage reference by adjusting the ratio of the two resistance.

 

[Siniša]

There are specific IC's designed exactly for this purpose and they are called rail splitters or virtual ground IC's. Have look here : https://tangentsoft.net/elec/vgrounds.html

 

[rfsystem]

OK give me also a try:

1. Use a bandgap VBDG=1.2V

2. At chip startup sample the ratio of VDD/2 to VBDG by using a opamp and a ladder network.

3. Use a comparator which compares the scaled VBDG with VDD/2

4. Use binary search to fix the scale factor

5. Finish the initial calibration. So the scale factor multiplies the VBDG to VDD/2 at the time of startup.

6. If later VDD changes happen a window comparator should trigger the process again.

 

[melc]

Why didn't you defined first what means "high PSRR" for you in standardised measuring units. Then you may apply one or other advices.
Mostly the RC divider is usually enough if the supply is a clean one.

 

[rockycheng]

Sorry, "PSRR" should be corrected as "KSVR".

 

[rajeshwari]

how self biased bandgap reference can be improved to have high PSRR

 

[northeast1]

Dear all,

Are there any special ways to obtain a Vcm voltage that is half the supply voltage? A simple voltage divider seems unable to restrain the noise from the power supply. Any suggestions are welcome! Thank you!

You can use resistor devider and connect to a large external cap.

 

[newsailor]

Use a resistor divider firstly,then use a buffer to follow this VDD/2,
But it is important that this buffer should have good performance.

Added after 13 minutes:

Use a resistor divider firstly,then use a buffer to follow this VDD/2,
But it is important that this buffer should have good performance.

MoreOver :
1, The buffer's input differential pair tansistors must be MOS, not be BJT, just to get high input impedence.
2, The PSRR of AMP in buffer must be high, to make good PSRR of VDD/2.