ADA4851 differences between single and dual

hello everyone,
i am having a hard time understanding the huge difference in behavior that i see between the single (ADA4851-1) and dual (ADA4851-2) versions of this amplifier.
my supply is +- 5V
my input is a DC to 10mhz, 0 to 1v sinewave
i am testing both versions of the chip (dual and single)

the input signal goes directly to the non-inv input,
the feedback res is 100k and another 100k goes from the inverting input to a variable voltage to control the output offset.
the output has no load except a 10x probe

the dual version, works somewhat as expected, but for the single version the output gain varies hugely for some frequency ranges.
i tried lowering the resistor values to 10k and to 1k, but then i get something that looks like a crossover distortion on the output (when i vary the offset control voltage).

in general, the desired output is a fixed gain of about 2 or 3, relatively stable across the DC to 10mhz input range, with wide output offset control.
the dual version seems to do ok (regarding the large resistor values), but the single version does not.

this is my 1st post here, so excuse me if i missed some important formality or something like that.
thank you for your time and help,
- Chris

  • The ADA4851 has a very large input (2.2 uA) bias current, so you need to keep your feedback resistance values low. 100k is hopeless, 10k may be acceptable for some applications but I would stick to less than 1k. It's possible that the behaviour you are seeing is just the device to device variation.

  • thank you,

    it seems i got confused by my own previous mistake on the offset setting voltage.

    while it was a buffered DC voltage, it was not stable when i used low resistor values on the high speed op amp.

    when i realized that i added a cap to keep the offset setting voltage stable.

    this completely removed the distortion on the output of the high speed op amp that i was seeing and that initially caused me to try the higher resistor values.

    your answer got me off the "carried away mode", i placed 1K resistors and found that the distortion cause was outside the op amp... thanks!

    however, out of curiosity i tested about 10 other high speed op amps (this model and others) and got the same differences when comparing single or dual chips.

    i.e. when using the wrong (too big) resistor values, the dual packages seem to not care while the single just dont work.

    so i guess there is some difference here aside from the device to device variations.

  • Hi Chris,

    That's an interesting observation you have there. Could you provide more details on the differences that you see? Maybe if you could share screen shots, we could help analyze the data.



  • Supply: +- 5V,
    Inputs: pseudo-differential sinewave (all positive, but phase 180), 350mV offset, 700mV peak-to-peak,
    channel 1 (yellow) is the dual package, channel 2 (blue) the single,

    linear symmetric frequency sweep, 100KHz - 10MHz,
    all 4 resistors are 100k

    Same as above except:
    linear symmetric frequency sweep, 100KHz - 30MHz,
    all 4 resistors are 10k

    Note: on the initial test, the input was not differential, just a positive sinewave, and the amplifier on non-inverting setup. There, the dual package had a much smaller gain drop on high frequencies, even with 100k resistors. i.e. the yellow channel above (dual chip) was almost flat, while the blue (single chip) was about the same as seen here.


    Chips types tested (2 of each) were: ADA4851-1(single), ADA4851-2(dual), MAX4389EUT (single), AD8056ANZ (dual).

    On all cases, the single-dual package differences were the same. The blue channel looks like a: “too high resistance with too high capacitance” on the feedback?

    But anyway the difference between the single and dual packages was surprising to me.

    By the way, the initial distortion that was going away with bigger resistors and looked like a crossover distortion was just that… the lm358 that I was using to produce the output offset control voltage.