Op-amp selection

Well, I've been short-listing op-amps. I chose another in-amp for the ohms converter, an MCP6N11 from Microchip. It has auto-zero, enable, analogue feedback and can be selected for minimum gain. It has 10pA input bias current - much better. I don't see a spice model of it on their web-site though. I chose the min gain = 1 version, but maybe I should have chosen the min gain = 2 version. Still not fixed the details of the converter yet.

So the rather painful process of selecting an input amp for the ac and dc volts and amps ranges is under way. I started looking at ones with bias current of 10pA or less for this, but I think it actually needs to be 1pA. Still there's plenty to choose from. So I've been finding out how to determine other important characteristics. I've started learning about noise. Very complicated.

So I'm now looking for bias current, supply voltage, noise current, noise voltage, slew rate and bandwidth. I actually didn't think slew rate and bandwidth were important for this application, until I realised the rms converter can give useful results up to nearly 1MHz, and not so useful results up to 15MHz. So I started looking at bandwidths of at least 1MHz, and since it's a true rms converter, and since different waveforms have different rms values, it had better be able to handle a square wave. A handy internet calculator gave a slew rate of at least 80v/μS for this.

This still leaves a huge choice of op-amps, comparing across different manufacturers, so the next criteria are cost and number of channels. Only 1 channel is actually needed for this input amp, but then extra channels could fulfil the other amp's requirements with room to spare. Ultimately it comes down to cost.

I came across a fascinating device from Linear Technology - the LTC1043, dual switched capacitor building block. It transfers small amounts of charge so as to present a high impedance input to things. Don't know how useful it might be at the input (or anywhere else) of my meter though.

So that's the main design figured out, component selection under way. Now it's details, details, details. Oh yes, and pcb design. According to a TI application note, better than creating a guard ring, is lift the inputs of the amp off the pcb and point wire them.

Bedtime.