Affordable USB DAQ for Geomagnetic Signals

Thanks for the advice Robert. I was reading about python last night. Learning a new bit of software can be fun... if the frustation level can be kept down.

George h.

NO, NO, NO!!! don't use both inputs to look at some RC oscillator's signal AND its chopped signal!

Use one of the OUTPUTS of the soundcard to drive the chop, then you KNOW where it's at.

Cheaper? the bandwidth on 24 bit USB is what? 1kHz? and costs $999compare to soundcard - built-in and can do 24 bits up to 89kHz ! seriously get close to Nyquist rate.

Ahh, OK that's much smarter!

OK lots cheaper! The (~$370) U6-pro labjack I have claims 22 'effective bi ts' on the 24bit A/D but I couldn't find a speed...(I know it's somewhere o n their web site.) perhaps below 1 kHz. But depending on his input noise l evel it may not be needed. Once you're digitizing at/below the noise level you are pretty much done. (I think.. not at all an A/D expert.)

George H.

Den onsdag den 29. januar 2014 15.46.23 UTC+1 skrev Robert Macy:

sorta know, there will be some unknown amount of buffering both going in and going out

-Lasse

Given cheap sound cards, I was figuring someone must have hacked one to do DC. I found this from "our own" Bob Masta,

George H.

Here is another of his designs that uses a chopper.

Barry Latham

Here is another possible solution, the EEG SMT produced by Olimex (99 EUR)

It's ready-made and there is a range of compatible software available for monitoring EEG input that happens to occupy the same bandwidth as the geomagneitc frequencies I am interested in.

Cheaper than a DAQ and free software. Sounds good to me.

Barry Latham

(big snip)

Thanks Robert. That is a truly inspired approach. I will examine further this weekend.

Barry Latham

I've had all sorts of grief with python on windows. Most of the code is written by linux users. Often Python2, and yes the edition matters. The deal is they use a lot of plugins and libraries that are simple to load up on linux and a hard to set up on windows.

Windows is good to run binaries. For code, you really want linux.

Uh, going out becomes your reference so could be a few days and not matter, Coming in? Not sure EXACTLY what you mean here. That delay is fixed, or at least should be fixed. For example, there's an approximate shift of 123 samples at 192000S/s and 91 samples at 44100S/s in my EMU1212 soundcard, so ?? It can be measured. Delays and such don't matter a lot. The point is the system iis synchronized. Also, doesn't matter, set up your software to accept all that 'unknown' stuff. By accept, I mean ignore. Ignore ANY sample shifting just to simplify the code.

Uh, *if* the EXACT phase of the sample is important to you, be prepared to interpolate your sample values to make that more exact like 123.15 sample shift using a truncated [or like hanning window] sync function to determine the interpolated value WITHOUT injecting too much error. You have to do to all samples, just not worth it. Easier to adjust in software by applying the classic phase shift to the FFT, then iFFT back. After all, the pattern MUST repeat here, because YOU are supplying the pattern.

THANKS!

I couldn't tell but that meets the 'patient leakage' requirement right? Not that the patient leaks, but the electrical circuit leaks.

Unless you want it to keep running for more than a couple of years without having to rewrite it for The Cool UI Library Of The Month. I have statically-linked executables on both Windows and Linux that I've been using for years and years, including both of my favourite editors.

I'm a big Linux fan in general (this is being typed on my main box, which runs CentOS 6). One serious problem is that since nobody's really in charge, nobody cares much about binary compatibility--you always get some incompatible change in some shared library, and then if you don't have the source, you're toast.

Cheers

Phil Hobbs

Thanks for the sage advice miso. If I run into problems.. well *when* I run into problems I hope the labjack community can help. (At the moment there is some other 'fire' I need to take care of.)

George h.