Looking for ethernet 16AI DAQ solution

[hfettig]

I am working on a project that will require a central controller with some analog inputs and digital IO but also two to four remote 8 to 20 channel AI devices. For the central controller we are considering cRIO or sbRIO. The remote DAQ devices could be up to 200m from the central controller. All they need to do is read voltages on demand. No high speed required. However, I have to be able to talk to in from LV RT.

I would welcome any suggestions.

Cheers,

Heiko

[ShaunR]

QUOTE (hfettig @ Apr 11 2009, 04:56 AM)

I am working on a project that will require a central controller with some analog inputs and digital IO but also two to four remote 8 to 20 channel AI devices. For the central controller we are considering cRIO or sbRIO. The remote DAQ devices could be up to 200m from the central controller. All they need to do is read voltages on demand. No high speed required. However, I have to be able to talk to in from LV RT.

I would welcome any suggestions.

Cheers,

Heiko

Unless you really need real time acquisition I wouldn't go with RIO (very expensive). PC as the controller (pc cards are cheap and you have the full flexibility and scalability of LV without the LV RT and memory restrictions) and any 3rd party ethernet acquisition modules (better channel count than NI e.g http://www.acromag.com' rel='nofollow' target="_blank">Acromag).

[Neville D]

QUOTE (hfettig @ Apr 10 2009, 08:56 PM)

For the central controller we are considering cRIO or sbRIO. The remote DAQ devices could be up to 200m from the central controller. All they need to do is read voltages on demand. No high speed required. However, I have to be able to talk to in from LV RT.

For the remote DAQ devices if you want ruggedness & a nice DIN-Rail mountable form factor, you could go with compactFieldPoint, which is not fast and has a lot of useful features like onboard datalogging to compactFlash, range of processor options etc, and of course programable and running LV-RT.

If you don't care about ruggedness or form factor, you could use a regular PC in ETS mode with PCI NI DAQ cards, and running LV-RT. You will need to check that the ethernet chipsets are supported for ETS mode, and that your DAQ cards will run on LV-RT (some older ones don't) but apart from the initial headaches of selecting and setting it up, it will run just as well as any NI LV-RT target.

This has the advantage of being very fast, fairly cheap (you need an LV-RT runtime licence) and equipped with Gig-E support as well.

If you want to go really fancy, you might use the new NI wireless DAQ modules..?

Neville.

[ShaunR]

cFP are not cheap. You start at about £1,000 just for the controller and once you've added the backplane, and IO you are generally looking at about £2K+. RIO's (as far as I'm concerned) are just repackaged fieldpoints. Considering you will have multiple end points, you will need a controller for each point (NI will love you a lot).

DO NOT GO WIRELESS!!!! (especially if you are in an industrial environment) unless you really have to. There aren't many pro's but shedloads of cons.Wires don't drop connections and interference is rife in the 2.4GHZ band. Also solid structures (like walls, cabinets and even people) attenuate the signal immensely. Just search the net for problems people have getting a signal from one side of their house to another let alone 200m!.

[Neville D]

QUOTE (ShaunR @ Apr 13 2009, 12:32 PM)

cFP are not cheap.

True, but you are paying for the ruggedness and ease of use of the platform (hot pluggable modules etc.) if your application demands it, then its worth it.

I have used Fieldpoint to develop a Hydrogen fueling station for fuel cell industrial vehicles. It worked flawlessly even at 1 degC (parking lot outside in fall).

QUOTE (ShaunR @ Apr 13 2009, 12:32 PM)

DO NOT GO WIRELESS!!!!

(especially if you are in an industrial environment) unless you really have to. There aren't many pro's but shedloads of cons.Wires don't drop connections and interference is rife in the 2.4GHZ band. Also solid structures (like walls, cabinets and even people) attenuate the signal immensely. Just search the net for problems people have getting a signal from one side of their house to another let alone 200m!.

Again, your mileage may vary but I regularly use wireless connectivity to view camera output for alignment and calibration in a vision application in a saw-mill. The distance to wireless node is about 50m through lots of steel walls with no problems. I get about 9 frames/s but thats good enough for my application.

This is using a regular laptop with 802.11g with commercial OTS networking hardware (nothing fancy, custom or expensive).

I guess the ultimate choice of hardware depends on needs as well as budget.

Neville.

[crelf]

QUOTE (ShaunR @ Apr 13 2009, 03:32 PM)

DO NOT GO WIRELESS!!!! (especially if you are in an industrial environment) unless you really have to.

Make sure there is a requirement to go wireless - some sites specifically rule-out wirelss (military, standards comissions, etc).

[ShaunR]

QUOTE (Neville D @ Apr 13 2009, 09:03 PM)

True, but you are paying for the ruggedness and ease of use of the platform (hot pluggable modules etc.) if your application demands it, then its worth it.

I have used Fieldpoint to develop a Hydrogen fueling station for fuel cell industrial vehicles. It worked flawlessly even at 1 degC (parking lot outside in fall).

I have also used cFP for testing Train Valves, machine control and factory monitoring (many moons ago now). I disagree you are paying for robustness and ease of use. I think you are paying a premium for the NI name. For most of my applications it's the channel count vs cost that is prohibitive. I find dumb 32channel Gigabit Ethernet digital/Analogue IO is far more cost effective and easier to manage (IP65 compliant @ £600 all in) than any of the NI solutions. Even a PLC is 1/3 the price if you really must have real-time.

QUOTE

Again, your mileage may vary but I regularly use wireless connectivity to view camera output for alignment and calibration in a vision application in a saw-mill. The distance to wireless node is about 50m through lots of steel walls with no problems. I get about 9 frames/s but thats good enough for my application.

This is using a regular laptop with 802.11g with commercial OTS networking hardware (nothing fancy, custom or expensive).

Does it matter if you lose a frame now and then?

I looked into using wireless camera's for checking rogue parts from from several bowl feeders that were feeding an assembly line. The process line was spread out over a 180sq ft factory floor with the operator in a cubicle at the end (wireless was an ideal solution). From testing, we randomly lost 1 in 30 frames on average and this went up to 1 in 12 if someone used bluetooth near the furthest feeder. Another example was a wireless remote alarm system that would feed back monitoring info to a central monitor. We had so many false alarms because the IT department had whitewashed the place with their wireless APs (it was cheaper than wiring and.....typically, they wouldn't let us near them!), just so the sales guys could wander around with their laptops and still have internet (although we did cause them grief as well ). I regularly VNC into machines wirelessly, but wouldn't trust it for serious remote control or acquisition.

QUOTE

I guess the ultimate choice of hardware depends on needs as well as budget.

Neville.

NI PCI cards are cheap, reliable, have high channel counts, easy to use and I go for them every time. But for remote acquisition there are far more cost effective and better solutions out there if you have the courage to step away from the warm fuzzy NI bubble.

QUOTE (crelf @ Apr 13 2009, 09:21 PM)

Make sure there is a

requirement

to go wireless - some sites specifically rule-out wirelss (military, standards comissions, etc).

Amen!

[hfettig]

Thanks for the suggestions.

cFP is definitely too expensive for our remote monitoring.

We are considering RT-ETS on a PC, however, once you have paid for the RT license and the PC you are in the same range as a sbRIO.

We will not go wireless for any of the acquisition modules. We might go wireless for an iPod web app user interface. We have done that for a different application and it works well.

Thanks for the Acromag link. I have found a few similar products in the $400-$500 range.

There really seems to be a lack of DAQ equipment for low sample rate. I don't need MS/s, I could do with one sample every second. I am measuring the level of feed in a system feeding livestock. They don't feed that fast :-)

If the remote stations were not as far away I would consider the low end NI USB DAQ like the 6008. But once you put a range extender for USB onto it your back in the $500 range.

Reason were are looking towards LV RT is that the customer should have as little chance as possible to mess things up, i.e. close the program, reboot the PC, etc. We don't really need the real-time part of it.

If we would abandon the idea of remote acquisition over ethernet would there be a way to condition a voltage signal (more or less constant) to be reliably read from a distance of 600ft (200m)?

Thanks,

Heiko

[ShaunR]

QUOTE

If we would abandon the idea of remote acquisition over ethernet would there be a way to condition a voltage signal (more or less constant) to be reliably read from a distance of 600ft (200m)?

Sure there is. Current Loop