compactRIO vs controller+PLC for vision

Hi there,
I have a vision project on that has been through some pretty major paradigm shifts since its inception.

Without getting into too many details I started with an Omron FH and global shutter CMOS to Omron NJ via EtherCat, then went to Line cam and Dalsa GEVA3000 interfacing with an Omron NJ via ethernetIP, which needed a different approach to get trigger outputs fast enough as Teledyne does not support EtherCat, then lastly I was introduced to FPGAs and what not with an NI compact RIO.
At face value the RIO seemed fantastic, but that was before they told me the software price. It had onboard and hugely expandable IO which the project needed, so no more interfacing via protocol to PLC, and hugely deterministic and simultaneous loops, rather than dealing with cycles and high speed comparisons.
To be honest the concept of FPGAs is awesome.

From what I can understand LabView uses some kind of function block diagram that then compiles to code when deployed. It looks like a Zelio in function block mode. For the price I am guessing this is right up there.

The application can be achieved using Teledyne, but I like the idea of doing away with complexity and using a single processor that is more like an industrial PC.
There is no real headroom with this approach, I think the NI route gives a lot more options.

I can lose the software price in the projects coming up, but it was unexpected. I have many software suites for different brands, but I had no idea a software package could be so spendy. Also I will probably pay someone else now to develop this application, then learn the software when I do not have buyers waiting for the working prototype. I was doing the whole thing myself with intuitive vision software, and standard PLC for IO.

Keen on experience with those in the know, as for me all this is unknown, apart from what I have been told by the suppliers, and the concepts online.
It all sounds shiny, but I am also keenly aware of the problems with say DCS vs PLC in the early stages - which brings me to me main questions is guess;

- Since I am only just aware of the NI gear, is the cRIO and LabView the powerful, stable and reliable product in 2016 that it appears to be?
- The high price of the software would indicate that it is a powerful tool, is this overpriced or justified given its huge range of application?
- The cRIO for my application seems to be huge overkill although I want the realtime and FPGA aspects of it which are the big ticket items. From what I can tell this product's main strength is data acquisition and realtime control, as opposed to PLC's strengths which are mainly process control with inroads to the above.
- I guess now PLCs are getting more into this realm anyway as they get more powerful, but is this all its cracked up to be or is it just an arduino on steroids?

It might be worth adding that my application has many small simple algorithms that need to run in parallel, and many outputs that need to operate at high speed, roughly 300 - 500us so it actually suits an FPGA in theory.

Thanks for reading,
Stew

I see Labview used a lot for test stands, universities, and scientific research. Also used by First robotics, incidentally. Quite powerful, been around a long time.

I wouldn't use a CRio to control an assembly line, but the fast, flexible IO makes a great fit for things where a PLC is pretty awkward.

We use LabVIEW on this site and at 3 other sites globally. They all run pretty much 24/7. We use PXI not cRIO, however I have seen the cRIO when I was over at NI UK HQ in 2014 for training and it looks like a good platform.

Personally, give me ladder logic or text based programming any day over NI's 'g-code'. Your mileage may vary. You can download a trial of LabVIEW (including the add-on's, of which you will likely need a few for your requirements). You get a 30 day trial I think and when that runs out you can click 'extend trial', I'm not sure how many times this can be done before they say - sorry, time to buy.

Thanks for replies,
I have a lot of IO on this that have to run in sub ms loops, and the type of application does not use complex algorithms, rather many small fixed ROIs and simple measurements within those ROIs so it seems like its a good fit for cRIO and this FPGA - what an amazing piece of hardware - I have only just discovered this.

Agreed about the programming, I also would prefer ladder/text but in saying that I think you can implement C++ through LV? (C++ is not my strong point)
Ah well, Patrick I'll start with the free trial I think - should be able to gauge it from there.

Like I mentioned in the first post, I got a shock when I was told what the software price was! Several times more than all of my major packages. Lets see where this goes....

Thanks,
Stew

Many PLCs can handle sub ms cycles, if you want to use LAD. You just need to get the more expensive CPUs and IO modules.

Also, if you only need a fast REACTIONS, and not actually a fast scan, hw based interrupts could be helpful.

Hi yes the original plc had 500us cycle time and supported etherCat. The vision controller only had ethernetIp support so to get deterministic outputs required syncronizing the encoder between units so that speed was not network dependant.
Ni unit was appealing as i could have onboard IO and less complexity while having plenty of leeway for speed given the amount of outputs. I would have needed 74 interrupts comparing 32bit integers which depending on how it was done can sacrifice rejection accuracy.
It is definitely possible using vision controller plus plc but i think i like the idea of having one embedded co troller doing the honors.
Cheers,
Stew