Let's call the top resistor of a voltage divider R1 and the bottom R2. The output impedance is R1 // R2, and the gain is R2/(R1+R2). Set these two equations equal to what you want. Now you have two equations and two unknowns. Solve accordingly.
Note that R1 encompasses the 50 Ω of the source, so the actual top resistor (what you add directly in series with the signal generator) will be 50 Ω less than what you solve for R1. At your very low gain, this will make little difference though.
The design looks ok on paper.
I fried the chips by: having wrong circuit (e.g. wrong connecting two
positive inputs of op-amps together) overheating it when soldering
(very low probable!).
That's possible, if you suspect the chip then replace it and be really careful the temperature of your soldering iron.
I have some wiring mistake, but I checked everything several times.
Its probably better to do a simple two layer PCB than spend hours of debugging, then you only have to check the PCB file. I think producing PCB's usually costs less than the time it takes to cobble something together and debug it. Its definitely easier to check than wires everywhere.
Chips I bought are not AD8512, or AD8512A is having totally different devices inside its package. But the fact that one chip works
through pints 5-6-7 and another does not is a fact against this
hypothesis.
If this is suspect, then order new chips from a reputable distributor or sample opamps directly from ADI. It is never worth the time cost of questioning whether your circuits working to save a few bucks on parts, especially during the prototyping phase.
There is no "AD8512A" that I can find on the internet, what I mean by that is you have an AD8512 with an A 'grade' suffix. The suffix on all ADI parts I know of delineates the package and grade. The datasheet for an AD8512A is the same for an AD8512. The AD8512 is what you have and the A means 'A grade' (A and B grades have different voltage offsets). There is an AD8512ARMZ or AD8512ARZ or AD8512BRZ the branding is B8A#. So you either have a really old part or perhaps a fake. A#325 does not jive with any of the revisions of the datasheet (the part was released in 2002) and has always had a B8A.
I'm not sure what kind of gauge wire your using but it has parasitic inductance in the 10's of nH range and you don't have decoupling capacitors on the board. The wiring also could be creating a ground loop, but this would create noise, you still should see some output.
The best way to solve this problem is:
1) Replace the part from a reputable source (mouser,digikey, arrow, newark, ect. NOT: ebay or alibaba).
2) Fix the cap problem
3) diagnose the signals with an o-scope probe on the input and output of the amplifier.
4) Double check your power rails with a scope
Best Answer
If you make a resistor potential divider using a 1 kohm and a 0.1 ohm resistor you can convert a pulse of peak amplitude 1 volts to an output pulse of peak amplitude of 100 uV. Instead, if you used a 10 kohm resistor your output amplitude peak would be 10 uV. If you want a 1 uV output pulse use a 100 kohm resistor and 0.1 ohms.
If you need variations of these make several - they can all operate parallel driven together because the input impedance is in the region of tens or hundreds of kohm. The output impedance is very low (circa 0.1 ohm).
Don't use an op-amp for this - it will only disappoint unless you are prepared to pay tens of dollars or GB pounds.