OpAmp with slightly negative supply for ADC input

Question

I would like to discuss a design idea around the supply of OpAmps that drive a ADC cannels of a microcontroller or DSP. Let the ADC be supplied by 3.3V and 0V ("analog" supplies separated from similar "digital" supplies), with an internal or external reference of 3.0V, that defines the full-scale input.

^{simulate this circuit – Schematic created using CircuitLab}

If the OpAmp is supplied by the same 3.3V - 0V, even if it is a rail-to-rail model, there is always a "footroom" limit of at least some tents of milliVolts. Therefore, my idea is to use a "Single Supply OpAmp" where its negative supply is connected to a slightly negative voltage, in order to use its complete voltage swing compensating for its rail-to-output difference both on the positive and the negative side of the amplifier.

What do you think about?

Answer 1

Yes, though -500mV might be a bit on the high side, many chips specify -300mV max. This is more of a practical issue at elevated temperatures, but the abs. max. values typically don't distinguish.

In fact, as @DKNguyen says, you can buy chips that provide a drop-in solution, such as the LM7705.

I believe Maxim offers a similar part.

Answer 2

Should be fine.

A distant concern is if OPA output goes fully negative (-0.5V), it may activate the lower ESD diodes on the input pin of the ADC chip (typically two diodes, Vss -> pin and pin -> Vdd). At -0.5V it is not critical. If it was -0.7V or less it would be questionable. An alternative, if the -V cannot be well controlled, is to pull down a small fixed current on the pin, so the the OPA is always sourcing current. Then the "footroom" is not a problem for the OPA, but rather for the pulldown current sink.

Answer 3

Take a look at the LTC6360. It is an OpAmp with a built in charge pump inverter to handle this exact case. The negative supply is also important for linearity, not just true 0 output. OpAmps become less linear as they approach their supply rails.