ada4927-1 adding dc offset onto inverting input to maximize adc range

Dear Sir/Madam,

With fully differential current feedback amplifiers like the ADA4927-1 are there any potential problems with adding an positive offset to the inverting input? The use case would be a unipolar input 0 to 1V on the non-inverting input and a 0.5V offset added to the inverting input. With a gain of 2 the full ADC range could be used (ADC with 2Vpp differential). If multiple FDAs are used in this configuration would it be better to use:

a) One LDO for all the FDAs offset

b) One LDO for each FDA offset

c) One LDO for all FDAs buffered using a unity gain buffer for each FDA?

If the unity gain buffer is the better solution would this device require any special attention ie. bandwidth

Thanks Grant

Parents
  • Grant,

    Are you talking about the inverting input of fig 48 of the rev A data sheet?

    If so, the circuit simply sees two signals and takes the difference.  It doesn't

    know that one is a DC offset.

    However, I would NOT use an LDO for the offset.

    -- too much noise

    -- too much tempco

    You didn't say how many bits the ADC is, but if you move 20 deg C, you

    could blow your entire error budget.

    I don't like a) because of routing, and I don't like c) becuase it introduces lots of errors from the amps.

    For b), depending on your budget, I would use multiple ADR4525 refs and divide down.

    (2.5V and 5V arethe most popular, highest volume reference voltages, but 5V has more noise.)

    Depending on the number of ADA4927s, you could use one ADR4525 and carefully route to

    all the amps.

    Harry

Reply
  • Grant,

    Are you talking about the inverting input of fig 48 of the rev A data sheet?

    If so, the circuit simply sees two signals and takes the difference.  It doesn't

    know that one is a DC offset.

    However, I would NOT use an LDO for the offset.

    -- too much noise

    -- too much tempco

    You didn't say how many bits the ADC is, but if you move 20 deg C, you

    could blow your entire error budget.

    I don't like a) because of routing, and I don't like c) becuase it introduces lots of errors from the amps.

    For b), depending on your budget, I would use multiple ADR4525 refs and divide down.

    (2.5V and 5V arethe most popular, highest volume reference voltages, but 5V has more noise.)

    Depending on the number of ADA4927s, you could use one ADR4525 and carefully route to

    all the amps.

    Harry

Children
No Data