Science/Engineering/Technical forums

Hello all,

I'm Pascal Dennerly, from Nottingham in the UK. I've been a semi-lurker on fusor.net for about 4-5 years now. My day job is writing software (mostly C++ with a heavy hint of TMP) but somehow I manage to fit a little in after work too. My spare time is mostly spent learning how decorating or driving to see various family in opposing corners of the country.

(And apologies for not using my real name in my profile - I haven't quite figured that out yet!)

Pascal

Welcome! Our software section hasn't been getting enough hits lately, so maybe you can help with that. Right now, we are doing a massive data aq effort for the fusor that will result in a lot of good content there when we get it a little more done. I made my living as a systems guy/programmer/hardware engineer, so we should find plenty to talk about.

We are taking inputs from:
A pic box that counts counters and has a few a/d channels and an accurate time stamp generator -- rs 232 to the PC.
The soundcard to record counter pulses for later PHA work, and to pick up audio comments. Built in.
A 1 GHZ 4 ch digital o scope (screen captures and data and settings) USB
And adding:
10 ch DAW workstation (96 khz, 24 bit) PCI card
A labjack U6-pro -- USB

And driving a brace of arbitrary waveform generators (USB).

The data is all put into a normalized MySQL database during runs -- all timestamped to help tease cause and effect apart. I'm not a DB maven, but I read the books and came up with a reasonable schema setup. There are separate tables the for example, hold info on what's hooked up to all those inputs on each device, what type of run we are doing, and so on, and of course, the data tables themselves.

Once we get finished there, the real fun begins with data mining software. Sure, open office can suck up mysql data and make pretty plots, but....there's a lot more we want to be able to do.

This is all in linux, insofar as possible, but we have a couple of things (mass spectrometer) that won't run without real windows (dcom stinks) so we also run a copy in Virtual Box on the Data AQ machine. The db is replicated to a more powerful machine elesehwere for later analysis and safety -- sometimes the odd lightning bolt around a fusor will kill the proximate machine.

We are using a mix of perl and C for this. Mostly perl and Tk for the gui at this point (which btw means it can run on other platforms with zero changes). C gets in there for things that have to move a lot of data quick, usually then becoming a perl module, as perl is great for "ducktape" jobs, but not something you crunch numbers in quickly.

So, when this gets done, that will be my big input to the software section, and of course, we will GPL it and give out the source code.