operational-amplifiervoltage-reference
If I understand correctly, we are using MC1043 because we don't trust Vcc to be stable. I get that the output of MC1043 is 2.5V which is extremely stable.
But I don't get how the output of opamp will be stable. Because this opamp is getting supply from VCC and VEE which are not necessarily stable. What am I missing here?
In normal operation, V+ and V- are the same. If they are different, the op-amp is being used as a comparator.
An important spec in choosing an op-amp is the "input common-mode voltage" range. The voltages at the inputs must be in this range for proper operation.
Some op-amps are "RRIO" or rail-to-rail input/output. As you might guess, this means the input common mode range goes from the Vee rail to the Vcc rail.
Some op-amps which include only the Vee rail in the common-mode range are referred to as "single-supply" op-amps.
Here is the spec in the LM324 datasheet:
Note that the OPA355 input common mode range extends to 0.1 Volt below ground BUT only to within 1.5V of Vdd. We'd need to see your circuit to know if this was significant. Also Vout is "not quite" rail to rail. OPA355 datasheet
I'll not claim to be a master in this context, but the following may be useful:
Similar specs to OPA355in many areas.
Apparently much superior CMRR.
2.75 vs 5.8 nV/Hz^2
Similar GBW
Superior output voltage swing (especially +ve)
280 vs 360 V/uS slew (ie slower)
LTC6252 Datasheet
720 MHz GBW
400MHz -3dB point as buffer
Input common mode rail to rail.
2.5V to 5.25V operation and fully specified at 2.7V and 5V.
CMRR 105 dB ! (!!!!!)
Noise 2.7 nV.Hz^-2
280 V/uS slew :-) :-) :-) :-) :-)
$4.59/1 in stock Digikey (Cheap considering spec).
Single, dual and quad versions available.
Best Answer
The output voltage of an OpAmp is independent from the supply voltage to a very high degree. The supply voltage rejection of the LM348 is typ. 96dB, meaning any ripple on Vcc will only be translated to the output of the OpAmp by a factor of \$10^{-\frac{96}{20}} = 15.8 ppm \$ (parts per million).
The output voltage of the LM348 will be determined almost solely by the stable voltage of the reference source.
But, as Peter Green has pointed out in the comments, the PSRR is typically given for 50 / 60 Hz. At higher frequencies the rejection will be not as good, the circuit designer has to take care to filter out any high frequency noise on the supply rails. Effective decoupling is important with every good circuit design.
Additional:
What might have a higher impact than the supply voltage are other parameters of the OpAmp.
For example the input offset voltage of the LM348 has a maximum value of 6mV. With the amplification of \$A_v = 4\$ this could result in the output beeing off up to 24mV.
But even that is typically not a problem: These high precision sources are all about low drift over time and temperature. As long as the error added by the OpAmp is static it can easily be calibrated for. Important is, that it does not change.