fast amp gain tweak

What's the difference between pin 1 and 6? Just to make the output physically available closer to the inputs?

E2_Gain_Tweak.jpg

How stable is the "about 60dB"?

More importantly, after tweaking, how well does the thing hold to the desired gain over temperature and aging?

Yes, a few fast opamps do that. It keeps the load current/wirebond inductance voltage drop from getting into the feedback path, and makes layout nicer.

Yes. At first it looked awesomely good, but I soon found the ADA4817's pin 1 connection to be a pain. For amplifiers with post-feedback resistor networks, and where you're relying on a low 0.1 or 0.15pF capacitance of the resistor, the close pin 1 proximity is presents unwelcome extra feedback. For a TIA using a pole-zero correction network (AoE III Fig 8.80.C) on the feedback resistor, it's a killer! I snip the lead off the so-8 package and use a special footprint without pin 1.

I kept a second version of the PCB with pin 1, but this looks to be valuable only for above 50 to 100MHz circuitry.

Nice, thanks. I don't see anything about the series R(R15) in the data sheet,

Well there is talk on page 16 fig's 47,48,49 of a series R with the SOIC pac, and Gain =1. Where did you learn this trick?

George H.

I've found ICs to be pretty consistent over time. The number and configuration of the transistors inside isn't going to change.

It's not as if I can buy a GHz opamp with 120 dB open-loop gain.

What I like about that amp is its basically zero bias current and noise current. Current-mode amps are terrible that way. I need its specs for my input stage, not shown, so I may as well use them in the dumb 100 MHz inverters too.

The 4817 looks like the best choice, because it's probably the only choice. The low open-loop gain is a mild annoyance.

A couple tenths of a pF of feedback C shouldn't hurt here.

I nabbed Phil's ADA4817 model, but it doesn't include distortion or output stage subtelties, so it's not much better than tweaking UniversalOpamp2.

I saw it sort of intermittently in a TI data sheet, and asked. The response was vague. I have used fast current-mode amps before, with the NI input hard grounded, without problems. It does make a nice way to tweak gain without involving the main feedback path.

So, sometimes people think this is a good idea.