I'm in the processes of buying a new data acquisition system for my company to use for various projects. At first, it's primary purpose will be to monitor up to 20 thermocouples and control the temperature of a composites oven. However, I also plan on using it to monitor accelerometers, strain gauges, and to act as a signal generator.

I probably won't be the only one to use it, but I have a good bit of programming experience with Atmel microcontrollers (C). I've used Labview before, but ~5 years ago. Labview would be good because it is easy to pick up on for both me and my coworkers. On the flip side, it's expensive. Right now I have a NI CompactDAQ system with 2 voltage and one thermocouple cards + Labview speced out and it's going to cost $5779!

I'm going to try to get the same I/O capabilities with different NI hardware for less $ + labview to see if I can get it for less $. I'd like to see if anyone has any suggestions other than Labview for me.

Thanks in advance!

+1  A: 

LabWindows CVI is NI's C IDE, with good integration with their instrument libraries and drivers. If you're willing to write C code, maybe you could get by with the base version of LabWindows CVI, versus having to buy a higher-end LabView version that has the functionality you need. LabWindows CVI and LabView are priced identically for the base versions, so that may not be much of an advantage.

Jim Lewis
+4  A: 

Welcome to test and measurement. It's pretty expensive for pre-built stuff, but you trade money for time.

You might check out the somewhat less expensive Agilent 34970A (and associated cards). It's a great workhorse for different kinds of sensing, and, if I recall correctly, it comes with some basic software.

For simple temperature control, you might consider a PID controller (Watlow or Omega used to be the brands, but it's been a few years since I've looked at them).

You also might look into the low-cost usb solutions from NI. The channel count is lower, but they're fairly inexpensive. They do still require software of some type, though.

There are also a fair number of good smaller companies (like Hytek Automation) that produce some types of measurement and control devices or sub-assemblies, but YMMV.

+1  A: 

Given the range of measurement types you plan to make and the fact that you want colleagues to be able to use this, I would suggest LabVIEW is a good choice - it will support everything you want to do and make it straightforward to put a decent GUI on it. Assuming you're on Windows then the base package should be adequate and if you want to build stand-alone applications, either to deploy on other PCs or to make a particular setup as simple as possible for your colleagues, you can buy the application builder separately later.

As for the DAQ hardware, you can certainly save money - e.g. Measurement Computing have a low cost 8-channel USB thermocouple input device - but that may cost you in setup time or be less robust to repeated changes in your hardware configuration for different tests.

+1  A: 

I've got a bit of experience with LabView stuff, and if you can afford it, it's awesome (and useful for a lot of different applications).

However, if your applications are simple you might actually be able to hack together something with one or two arduino's here, it's OSS, and has some good cheap hardware boards.

LabView really comes into its own with real time applications or RAD (because GUI dev is super easy), so if all you're doing is running a couple of thermopiles I'd find something cheaper.

+1  A: 

This isn't necessarily a software answer, but if you want low cost data aquisition, you might want to look at the labjack. It's basically a microcontroller & usb interface wrapped in a nice box (like an arduino (Atmel AVR + USB-Serial converter) but closed source) with a lot of drivers and functions for various languages, including labview.

Reading a thermocouple can be tough because microvolts are significant, so you either need a high resolution A/D or an amplifier on the input. I think NI may sell a specialized digitizer for thermocouple readings, but again you'll pay.

As far as the software answer, labview will work nicely with almost any hardware you choose -- e.g. I built my own temperature controller based on an arduino (with an AD7780) wrote a little interface using serial commands and then talked with it using labview. But if you're willing to pay a premium for a guaranteed to work out of the box solution, you can't go wrong with labview and an NI part.

+1  A: 

There's a lot of misconception about what will and will not work with LabView and what you do and do not need to build a decent system with it.

First off, as others have said, test and measurement is expensive. Regardless of what you end up doing, the system you describe IS going to cost thousands to build.

Second, you don't NEED to use NI hardware with LabView. For thermocouples your best bet is to look into multichannel or multiple single-channel thermocouple units - something that reads from a thermocouple and outputs to something like RS-232, etc. The OMEGABUS Digital Transmitters are an example, but many others exist.

In this way, you need only a breakout card with lots of RS-232 ports and you can grow your system as it needs it. You can still use labview to acquire the data via RS-232 and then display, log, process, etc, it however you like.

Third party signal generators would also work, for example. You can pick up good ones (with GPIB connection) reasonably cheaply and with a GPIB board can integrate it into LabView as well. This if you want something like a function generator, of course (duty cycled pulses, standard sine/triangle/ramp functions, etc). If you're talking about arbitrary signal generation then this remains a reasonably expensive thing to do (if $5000 is our goalpost for "expensive").

This also hinges on what you're needing the signal generation for - if you're thinking for control signals then, again, there may be cheaper and more robust opitons available. For temperature control, for example, separate hardware PID controllers are probably the best bet. This also takes care of your thermocouple problem since PID controllers will typically accept thermocouple inputs as well. In this way you only need one interface (RS-232, for example) to the external PID controller and you have total access in LabView to temperature readings as well as the ability to control setpoints and PID parameters in one unit.

Perhaps if you could elaborate on not just the system components as you've planned them at present, but the ultimaty system functionality, it may be easier to suggest alternatives - not simply alternative hardware, but alternative system design altogether.

edit :

Have a look at Omega CNi8C22-C24 and CNiS8C24-C24 units -> these are temperature and strain DIN PID units which will take inputs from your thermocouples and strain gauges, process the inputs into proper measurements, and communicate with LabView (or anything else) via RS-232.

The Omega CNi8... are controllers that are meant to stand alone, with some communication with a computer. They're probably inappropriate here for a couple of reasons -- (1) He wants to read 20 thermocouples, so would need 20 units -- also hard to talk to 20 separate devices over serial, although I think omega sells an ethernet version. (2) If the feedback output depends on more than one input (e.g. heater 3 needs to know about sensors 1-4) these can't handle that.
Agreed. It was meant as a starting point more than anything, and I did suggest a multichannel interface as well - like the e-net connected ones you're talking about. OP didn't give the impression that he knew PID existed so I figured it would be somewhere to start his own search.Ideally, I'd probably do a system like this with a small, low-cost PLC system, to be honest - deal with all the communication over OPC. You could build one for less than $5k and it would allow for lots of expansion. Above a certain point it does, as you say, become cumbersome to do it all in a modular way.
Again, without knowing exactly what the system is it is difficult to suggest options, but for an integrated multichannel controller things like this exist : temperature sensing. With this you could monitor temps but not control them. It all depends heavily on what the OP is intending to use his 20 thermocouples for. Without more information about the system goals it's difficult to suggest effective options.
+1  A: 

A few thousand dollars is not a lot of money for process monitoring and control systems. If you do a cost/benefit analysis, you will very quickly recover your development costs if the scope of the system is right and if it does the job it is intended to do.

Another tool to consider is National Instruments measurement studio with VB .NET. This way you can still use the NI hardware if you want and can still build nice gui's quickly.

Alternatively, as others have said, it is perfectly viable to get industrial serial based instruments and talk to them with LabVIEW, VB .NET, c# or whatever you like.

If you go down the route of serial instruments, another piece of hardware that might be useful is a serial terminal (example). These allow you to connect arbitrary numbers of devices to your network. You computers can then use them as though they were physical COM ports.