LabJack, Linux, data aquisition/control

[Doug Coulter]

Just received a LabJack U6-pro and am testing it on my Linux system here. I am thinking of adding this to our lab data acquisition stuff, and it looks like it's going to be a fantastically great thing -- worth the bucks even for someone who usually rolls their own.

I managed to crash it once on some example (a re-plug fixed that), but in general this is one fine thing indeed. I opted for this particular model as the ethernet one was out of stock and too high priced (they have an app note on how to put the version I got onto a wireless dongle much cheaper anyway -- which gives the additional benefit of galvanic isolation from everything else in the system -- no ground loops and lightning can't wipe your whole system), and I wanted the temp sensor for thermocouple cold junction compensation this will provide. This is not meant to be fast-time-deterministic, but is pretty good regardless if you can handle a millisecond or so of time jitter on acquisitions from the high resolution a/d channels. It is not simultaneous sampling -- it's muxed, but for what I want, things like power supply parameters, gas gages, and perhaps even setting things like the power supply, it looks just about perfect. It does do a few khz on the a/d scanning the channels. Fast enough for things like that.

It seems someone who knows a goodly bit about real world data acquisition designed this thing. Stuff like the inputs can be gain programed, the digital I/O has pullups, there is pwm output, counters, all the good stuff -- the digital stuff looks like CMOS on input but will drive TTL on output, nice low output impedances on the dacs and so on, and some protection from EMI.
In fact, it looks for all the world like someone did a PIC design and added some pretty nifty hardware to it for the few things a PIC alone won't do (like high res A/D).

Installation was easy for someone used to the make, make install thing one does for a lot of Linux software, and who knows what is meant by dependencies and how to run apt-get or synaptic package manager -- it all just worked on the first try. All the code is GPL open source, simple C, with source code provided. Next I will try to get this built as a perl module so we can integrate it into our other stuff that shoves timestamped data into a MySQL database during runs.

Takeaway -- this is one nice thing, I may get more. If we produce useful code for this re things we do with physics, I'll put it up here for all. I have to say, this is the very first bit of scientific type gear that I've ever had this little trouble getting to run on Linux (in this case Ubuntu 10.4). No emulators (Wine), no Virtual Box and windows, just runs right on the metal, fast and efficient.

On a side note, we also found some tools to integrate the GDS-2204 4ch 1ghz digital scope up and running, similar results, so that will go with our data aq project too so we can take snap shots of fast events (which has become important here). It's its own system that uses Gnuplot and a custom shell, so we will be modding their source code to integrate into our master code. Funny thing is, it works much better than the supplied code from GW-Instek for that product!