I need to shift ±2.5 V signal into 0 V to 5 V for my A/D converter.
I found several designs, most of them using either voltage divider or an opamp.
I wonder, why not to use just a single reference to shift the voltage? It works on the breadboard so far... What are things I should be concerned with? Noise? Dynamic characteristics?
simulate this circuit – Schematic created using CircuitLab
PS: The Signal comes from opamp and is connected directly to ADC
Nothing wrong with using a shunt voltage reference - I've used one myself for shifting exactly the same p-p AC voltage exactly the same amount (2.5V) to exactly the same recipient (an ADC). I'll also add that it worked great from DC to 30 kHz - just make sure that you put a small capacitor across it so its dynamic impedance isn't "tested" too much at higher operating frequencies.
You need to maintain a minimum current for a shunt reference to work, so you need to have a Vcc in excess of 5V - preferably a fair bit more so the reference current does not change too much with input voltage. As Andy says, a parallel capacitor helps, but it cannot supply currency indefinitely if the input voltage is at or very near Vcc.
For example, a TL431 needs 1mA to work properly. Of course that current must be sunk by your input driver, which could cause various problems depending on where the input is coming from (it would be a bit rude to make a product with a general-purpose input like that). In your case, you control the input amplifier so it may not be an issue. If you don't have much voltage headroom (say a couple volts) you could consider replacing the resistor with a current mirror made from discrete transistors.
By the way, a voltage divider won't work in your case because you want the output voltage swing to be the same as the input voltage swing. A passive-only solution will reduce the swing, for example two equal resistors to go from +/-5V to 0-5v. If you could reduce your ADC reference to 2.5V, however (maybe using that shunt reference...) it could work quite well.
