sbRIOs - easier to use and more versatile than they appear

[Mads]

Not a question (I might write a blog post someday instead about this) - but just wanted to share some insight about sbRIOs:

We needed a tiny computer in one of our subsea products that had to have dual networking, lots of serial ports - AND a very low power usage (preferably below 5W). Previously we had solved this by using the boards from the old cFP-2220 (which only drew about 3,8W and had 4 serial ports), but - apart from being increasingly pricey and soon obsolete (last time buy this year) - it was very slow, and its secondary Ethernet had no DHCP functionality. This turned out to be surprisingly difficult. We had a look at the sbRIOs from NI, but according to the specifications these would require waaay more power than 5W, and having a second network port and lots of serial ports seemed to require quite a bit of work....- We went looking for SBCs (of which there are not that many low power ones with dual networking, but there are some, and the number is growing)...but did not like how we would have to recreate the boot loader etc. functionality of a PAC for them. Using Windows would be the easiest for us where Linux RT is not an option, but putting Windows subsea was not too enticing either.

We tried the SOM from NI and retrieved offers from various design houses on making a carrier board for it. That proved to be a costly affair, either just for the design and/or for each board.
We were starting to get a bit pressed for time - so as a last resort almost we decided to return to the sbRIO option and see exactly how much power an sbRIO-based solution would require in real use, and how much work it would be for us to design a mezzanine board ourselves.

To our surprise the sbRIOs (9607 and 9627) were using less than 4W, even with both CPUs running at 100% (!)...Much less than the (obviously) worst case scenarios specified by NI. With that knowledge, we started looking at the design guide for the mezzanine board - and found out that the extra Ethernet port is pretty much a wiring job. We added serial ports, and SD card....and, as the RT clock battery on the sbRIOs only lives about 10 years, only a third of what we needed for our products, we added a new battery as well (40 year lifetime !). 

I'm writing this today as we have just tested the first prototype card, and it all works beautifully. The Ethernet port was up and running as soon as we connected the card, and the rest just required us to create a CLIP file and add that to the RMC socket...Quite simple (there is a huge lack of documentation here though, NI should make better step by step guides for this for us first time users).

We now have an sbRIO-9607 with dual networking, long-life RT battery, 3 serial ports (for now), and an SD card - and it all (including the mezzanine card) just draws 3,5-3,8W (!). Great stuff!!!

The only issue we have with the solution now is that the sbRIOs are IPC Class 2, and the oil and gas industry standards require class 3.....but for now we'll just have to live with that. Hopefully NI will be able to supply IPC Class 3 someday soon.

If NI were better at describing what the realistic power requirements are instead of just worst case numbers, and how easy it is to add functionality with the mezzanine card (or even offered some themselves), it would have saved us a *lot* of time of work. If there is interest for it we might consider offering our mezzanine card as a solution for others sometime in the future ourselves though. It certainly beats many of the alternatives out there, as long as you get passed the misleading specifications and lack of documentation.

[MarkCG]

That's good to hear, I'd like to use one for a project some day. I am kind of curious if NI is going to continue its investment in the area of embedded control and general machine control.  

[thoult]

I love the sbRIO platform for a number of reasons, mostly because it's easily integrated into so many different applications.

I'm pretty sure that flexibility is the reason why we don't have more specific numbers across much of the range. What DIO are you using? What are you powering from the 5V bus? Do you have any C-series modules slotted in? What are you running on the RMC? What's the operating temperature? What additional inrush current is expected? I know with the 9607 etc some of these are moot points, but still. My personal experience of butting up against specification is in digital timing. The maximum that NI like to quote for DIO speed is 10 MHz, but I know I can eke more out of that on the input side for some challenging encoder measurements.

For devices like the 9612 and 9636 I've used, there are definitely more detailed guidelines for power requirements in the spec because of aspects like these, but I've found that the best help tends to come from the sbRIO R&D teams, or Spex here on the boards when you have specific requirements like this. There are some other resources out there - I remember a white paper on thermal management in sbRIOs for instance - but when you're really pushing the boundaries as you are, you kinda hope there's a specialised sbRIO Apps Engineer who can be made available to you!

For the point about mezzanines, they're definitely getting better. There are design guides out there, and there have been a few case studies on RMC boards in the past few years that help give a flavour of what you can do and the considerations for it. If I can get the time, I'd like to sit down and redesign some of my electronics interfaces with 9627s and an RMC.

Glad to hear that it's all worked - if this ended up being a case study or presentation at an event vaguely near me, I would be really interested in reading or hearing more.

[skhanCRL]

Hello Mads,

I am doing a similar development and first time user of the SBRIO platform with the RMC card. Can we chat / email externally and would you mind sharing information?