Non-Inverting Op-Amp Theory

[BennyBoy]

Hey all,

Just had a quick question that I am not sure how to formulate in searches for info but I was curious about the theory behind non-inverting op-amps that have the negative feedback connected to reference voltage rather than ground. In what cases would this design choice be made, and how does this effect gain calculations? 

[Rob Strand]

Just had a quick question that I am not sure how to formulate in searches for info but I was curious about the theory behind non-inverting op-amps that have the negative feedback connected to reference voltage rather than ground. In what cases would this design choice be made, and how does this effect gain calculations? 

The assumption is Vref is a low impedance source (for AC) so it makes no difference to basic calculations.

If Vref isn't quite a low impedance source then that complicates things a lot because you can't ignore the Vref impedance.

In practice connecting the feedback to GND or VREF can produce different levels of noise an hum.   You can analyse this fairly easily but it's outside of the general hobbyist's world.

Connecting the feedback path to GND produces a DC voltage across the cap C6.   This can affect small package SMD devices.

[BennyBoy]

Thanks, Rob! That makes a lot of sense.

[phasetrans]

Connecting the feedback path to GND produces a DC voltage across the cap C6.   This can affect small package SMD devices.

Specifically, a DC bias across the SMD capacitor changes the spontaneous polarization of the dielectric (typically barium titanite). In practice this means the capacitor has less capacitance with the DC voltage applied, than without.

[Rob Strand]

Specifically, a DC bias across the SMD capacitor changes the spontaneous polarization of the dielectric (typically barium titanite). In practice this means the capacitor has less capacitance with the DC voltage applied, than without.

Even SMD ceramics.  I posted info and plots a few times about it on the forum.  It can be a strong effect for larger capacitance values in smaller packages.


These results are indicative of the problem:

[ElectricDruid]

...so stay safe, kids! Use traditional big, ol'electrolytic caps and avoid the problems! 

[antonis]

!!!..DON'T PUT GARBAGE IN Vref..!!!

[amptramp]

If you have multiple stages connecting to Vref and any of them are driving any current into it, it may be preferable to establish more than one Vref to keep feeedback from subsequent stages from upsetting the Vref voltage.  This may be easier than trying to get the impedance of Vref down to a minimum.

[ElectricDruid]

If you have multiple stages connecting to Vref and any of them are driving any current into it, it may be preferable to establish more than one Vref to keep feeedback from subsequent stages from upsetting the Vref voltage.  This may be easier than trying to get the impedance of Vref down to a minimum.

I've been studying old drum machine schematics recently, and you see this *all over*. Practically every separate sound circuit (snare, bass drum, cymbals, whatever) has a separate RC filter from the main supply, and pretty much *every* op-amp has its own Vref. I guess the type of clicks and thumps these things are *designed* to generate are pretty much the worst case for coupling through to other circuits, especially when they're all jammed up together on one PCB, as they tend to be for cost reasons.

[phasetrans]

Specifically, a DC bias across the SMD capacitor changes the spontaneous polarization of the dielectric (typically barium titanite). In practice this means the capacitor has less capacitance with the DC voltage applied, than without.

Even SMD ceramics.  I posted info and plots a few times about it on the forum.  It can be a strong effect for larger capacitance values in smaller packages.

Small, large value caps have more interleaved metal / dielectric alternating layers inside. So the local electric field strength is higher in the dielectric, and the effect on the inherent polarization of the ferroelectric is higher.

Long ago as an undergraduate, I worked in a lab where we looked at the polarization behavior of ferroelectric materials. As well as nucleation and growth of the ferroelectric domains.