Millivolt logger with high sampling rate [closed]

Question

I'm trying to find modern components for a design that would be able to record millivolt range samples at a high sampling rate (~30 kHz) for ~1 hr interval. Key components I can think of thus far are digitizer, amplifier, some type of write circuit (SD card?). Importantly, are there any solutions that might be able to provide a high channel count ~1 - 200 channels? What considerations should I keep in mind recording from such low voltages with minimal noise?

Answer 1

For one or two channels, what you described would be called a sound card. Your sample rate is anything but high! So, that seems like a solved problem. Laptop, microphone input, standard WAV recording program, done. Noise figures of audio equipment is quite nice, and I don't think you mean "minimal" where you write "minimal", or else we need to talk about impossible to avoid quantum noise in cryogenically cooled solid state electronics...

Your high channel count however makes this system pretty complicated, even theoretically, because now you have to keep a lot of sample clocks in lockstep. This is probably the point where you start building smaller modules that you can feed from the same clock source and plug them together, allowing them to multiplex their data onto a single data bus so that a sufficiently capable PC can record them onto an array of SSDs. Your SD card idea is cute, but betrayed you haven't even done back-of-envelope calculations: at 30 kS/s, 200 channels, you get 6 MS/s. At a sample depth of 16 bit (audio equipment usually nominally does 24, so this is already a concession), that's 96 Mb/s sustained write rate. SD cards don't do that for multiple seconds, let alone an hour. (By the way, that hour is about 40 GB worth of data.)

So, your question on "a modern component" is a bit off to a bad start: nothing you're doing here needs anything modern - this rate is just low, and millivolts aren't challenging. But it needs something well-designed, large-scale, so it's not a problem of finding the right component, but the right design.

If recommend opening a new question post that describes what you want to do in actual detail (no shortcuts, describe it, don't try to generate before you know what can and can't be abstracted!), and asks for a design approach. Selection of components comes way, way later in the design process then where you currently are!