using seperate V+ and V/2 for opamps

Started by changes, September 13, 2009, 05:42:01 AM

Previous topic - Next topic

changes

I have a design featuring three dual opamps. All of them are simple gain stages and/or buffers. I'm thinking for noise issues to filter each V+ supply seperately and make a new V/2 for each dual opamp.

Will i benefit from that or is it too much?
CHANGES

head_spaz

Will i benefit from that or is it too much?

Well... that depends on whether the powersupply is the source of your noise artifacts.

Few circuits bother to do it that way, so generally speaking, it shouldn't be necessary.

I would begin troubleshooting by installing bypass caps on all of your opamps.
Almost all data sheets mention this... but it is rarely incorporated in the projects or
mentioned in this forum.
Bypass caps local to each opamp is essential for reducing powersupply noise, ripple and
crosstalk that can be modulated between stages via the powerplane.

You can test it easy enough by placing a .1uf to 1uf cap on each opamps' power lead
to the Vbias / Ground.
And be sure to READ the manufacturer's data sheet for more specific instructions.

Good luck.
Deception does not exist in real life, it is only a figment of perception.

alanlan

Could potentially cause a problem in certain circumstances but in practice you should be OK with just one.  The problem can be that you get a tiny amount of coupling from one point in your circuit to another via the V/2 reference and if a circuit using this reference has very high AC gain, then it can lead to problems.  But as I said, it's usually OK.

Just make sure you put a filter cap across V/2 and bear in mind that it's low pass roll off frequency is dependent on the R values chosen so the cap has to be big enough to filter out most of what might appear otherwise.  The roll off frequency is given by 1/(2*PI*(R1//R2)*C) where R1 and R2 are your potential divider resistors - in your case they will be identical values so R1//R2 = R/2.

So say you have 2 100K resistors to make your V/2, and you want to filter down to say 1Hz (3dB down at this point) then:

C must be greater or equal to 1/(6.283*50K*1Hz) = 3.2uF, so you could use 3.3uF or greater.

changes

thanks guys
i'll use the same power supply, V/2 and using 100n V+ and V- pins of the opamps and see how it goes
CHANGES