Analog front end for HS10X oscilloscope. - https://www.martinloren.com/hs10x-diy-oscilloscopes/ - http://hscope.martinloren.com/HS10X-custom-modules.html Aiming for 1MOhn/~20pF input for use with 1X/10X oscilloscope probes. Ranges: - 0-20V or - +-5V and +-15V attenuate to 0-3.3V ADC input. 5V available. Martin Loren published a multi-range example with ~1MOhm Zin, but: 1. single channel 2. Maximally attenuates then variably amplifies (tho MCU noise prolly buries any penalty. but still.) 3. magic amp (an interesting item and cheap-ish, but function baked into H-Scope ap appears hard to hack) Challenge: buffering very high impedance signal with jellybean parts. - Because spending more on op amps than on the uc board would violate the spirit of the problem - Seemed like it would easy Non-answers: - LM358 - slow & input bias current - TL072 - fast & FET input, but needs ~4V from lower rail - FETs on hand - 0ish bias current, but all trade high gate C for low Rdson - NPNs on hand - slow - BFR93 - jellybean RF NPN is fast enough; can sim a Darlington emitter-follower with speed, lo Zout & hi Zin but gain is unity-ish for small signals and 0-3.3 (+Vbe) of 5V isn't that. - swinging 0-3.3V on 0-5V supply hits slew limits of devices with nominal b/w. Answer: AD823 fast FET input low V op amps. good from lower rail to above 3.3V at 5V supply. jellybean cheap at dimes/unit in so8 dual packages from AliExpress. (tho >$1@ for DIP8 & >$3@ from reputable sources). Looks good in sim. Ordered.