AD8145 doesn't work exactly.

Dear Sir/Madam,

I need your help.

My schematic:

cf5b8619e13a6629748eb9fcb37b1341.bin

VREF is provided directly by AD5765. I set the common mode voltage of Vin to 0.5 V and the differential mode voltage to 1.0 V.

AD5765's output voltage is set to 0 V.  Than I measure the voltage of AD8145's REF pin. However, it's not 0 V(Of course, AD5765 doesn't work exactly now). The voltage on REF pin has following features:

Its frequency is the same as Vin;

The following pictures show how its Vp-p  varies with frequency:

ddcb67b663c8130f42f55ea6aa07677c.bin

58866f4eb63dcf28941b89e8d64e5890.bin

My question is that why REF pin of AD8145 is not 0 V ?

  • Hi zhifu,

    Please verify that you are using +/-5V supplies on the AD8145.

    Thanks.

    --Jonathan

  • Hi jpearson,

    I'm using +/-5V supplies on the AD8145.

    Thanks.

  • Hi zhifu,

    Do you know the source impedance of the output that you're using from the AD5765?  REF needs to be derived from a source with very low impedance, otherwise some of the output voltage will appear at the REF pin. If the source impedance is high, the best solution is to use a buffer with bypass capacitors to provide a broadband low-Z source.  Can you connect REF to ground?

    Best regards.

    --Jonathan

  • I got a reference circuit of AD8130 in application note AN-1214:

    The note said that "The two 49.9 Ω resistors( in the red square) in series with the AD8130 input pins improve the overall distortion performance of the circuit". How these tow 49.9 Ω resistors improve the overall distortion performance of the circuit? After all, the differential input impedance of AD8130 is quite high(6 MΩ). If these tow resistors are close to the differential input port of AD8130, serious reflection will appear.

  • Hi zhifu,

    A moderate amount of load capacitance (a few pF to a few hundred pF) directly applied to an op-amp output is generally problematic since it produces a second pole in the loop gain response before the 0 dB loop gain crossover, but putting something like a 10 uF load can often be done successfully because it nonlinearly drops the open-loop gain magnitude response and pushes its 0 dB crossover back in frequency where the phase shift is less.  The nice thing about this is that you get the stabilizing effect of inserting attenuation in the loop gain without increasing the closed-loop gain.  This allows the circuit to be operated at a gain of one, which is most often the desired gain.  The only reason it works is because the op-amp is not ideal, and its open-loop magnitude response drops with the heavy cap load.  The AD8031 is a popular op-amp in this application, and has been well proven to be stable with a 10 uF load at unity gain.  I recommend that you use the AD8031 in a gain of one, with a 10 uF capacitor and 0.1 uF capacitor in parallel between its output and ground.  (You can always change the 0.1 uF value or place additional smaller caps in parallel if necessary, once the 10 uF cap is in place.)  This will give you a broadband low-Z source to drive the "GAIN" input.

    Best regards.

    --Jonathan