Post Go back to editing

LTC6404 input circuit

Dear Sir or Madam 

I am implementing my design with LTC6404-2 , and not sure if the  circuits  will  work well , especially about the input part , where switches are inserted  between  the input resistors (R1,R2)  and feedback resistors(R5 , R6) ,  Besides , are there any issues if I place the R1 ,R2 far away from LTC6404-2 input pins respectively  in my layout design ?

Is the circuit OK?

Thank you in advance  for your advice.


Best regards


  • Hi Anglic,

    Please clarify exactly what problems you expect to get and what are you afraid of?

    Yes, this amplifier is optimized for high-speed operation and does not tolerate unity gain, this is stated in the specification. But I see resistors R3 and R4 that can provide increased noise gain when S1 and S2 are open.

    I'm confused about capacitors in the feedback circuit, why do you use them?



  • Oops ,  My bad.

    It was LTC6404-1 , however , being used as a half unity gain (0.5V/V) . Is it OK? 

    The problems are ,  Doesn't it matter if  R1 R2 are placing far away from LTC6404-1 on PCB board ?

    According to the datasheet P25 , feedback capacitance can help to mitigate peaking in frequency response . 

    And , here is the modified circuit , where the input resistor is 130ohm , the feedback resistor is 68ohm .

    Another problem is , can LTC6404-1 be used to drive AD9650 ?

    Best regards!


  • Hi Anglic,

    I posted a response to your other post about using LTC6404-1 at lower than 1V/V gain here:

    In general you'd want to minimize all trace lengths with a device as fast as LTC6404-1. You probably have to experiment with it on a prototype board to make sure though.

    I see the plot you've noted about cap across RF to mitigate peaking. That's certainly a valid technique:

    You had asked if LTC6404-1 can be used to drive AD9650 ADC.

    From AD9650 datasheet:

    • Analog input: 2.7Vpp (Vref = 1.35V)
    • Input CM voltage: 0.9V

    From LTspice, these conditions look doable as shown below:

    With some drivers, approaching the ~0.2V from ground may be troublesome. I've seen designs where the FDA is used with its V- supply pin powered from a slightly negative supply such as -0.7V to alleviate the swing so close to the V- rail. But, if LTspice can be trusted (which I tend to do), first order seems like a doable proposition (unless distortion terms become too large with swing so close to the V- rail).

    Here is the LTspice file for your review:

    LTC6404-1 AD9650 driver EZ 2_26_20.asc



  • Oh, I see .

    Thank you , Hooman.

    Really appreciate for your help.

    Best regards .


Reply Children
  • Hi Anglic,

    The main feature of a fully differential amplifier is that it always works in inverting mode. Therefore, the noise gain is always greater than one. Even when it works with a gain of 1 it corresponds to a noise gain of 2. Most very high-speed amplifiers are decompensated and do not allow operation with a noise gain close to or equal to 1.

    Since you want a fully differential amplifier to work with a gain less than one, this corresponds to working with a noise gain close to 1 (specifically in your case, 1.5). It is necessary to check the behavior of the amplifier at the extreme value of the noise gain - 1 to find out about potential stability problems. I suggest this model:

    The noise gain here is strictly equal to 1

    You can see a peak that indicates a small phase margin. In principle this should not lead to self excitation but will cause overshot an impulse response. You can try using a gain 0.5, I think.



  • Thank you very much , Kirill .

    Best regards.