CMOS op amp datasheet: Lies! All lies!

The fact that the quoted value is exactly the shot noise of the typical bias current is extremely suspicious.

Cheers

Phil Hobbs

It must have been a perfect process :)

NT

Well, I tried a few other parts, and they all worked as expected--1-Hz noise flat at 9 uV out to the rolloff at 1 kHz. It's just the Diodes TLV272IS that misbehaves.

The front end has a 200M // 1 pF FB network followed by a x5 amp in the second stage, made from one section of a LM358, so the expected noise is

I like the Microchip MCP6242 so far, $0.27 @ 1ku from Digikey.

Cheers

Phil Hobbs

Well even with a gain of 5 that's hardly going to touch Phil's 9uV/rtHz. (His 200 M resistor has a noise north of 1.3 uV/rtHz.)

George H. (I always think noise should be expressed as a power density, V^2/Hz.. that way it's easier to compare the relative contribution.)

Makes not much difference in this case. The 200M resistor's noise is 40 times higher than that, and the 3 dB point of the e_N*C_diode contribution is about 5 kHz, well above the gain rolloff.

Cheers

Phil Hobbs

You have to tote around a lot of powers of ten that way. I just use a couple of rules of thumb, e.g. that e_N2 = 0.5 e_N1, it increases the total noise by 1 dB, and it's quadratic in dB below that--if e_N2 = 0.25 e_N2, it increases noise by 0.25 dB, and so on.

Cheers

Phil Hobbs

Thanks for the info. Diodes Inc. seems to be in a bunch of only- peripherally-related fields- probably related to acquisitions and JVs in the PRC such as the BCD acquisition.

--sp