sbRIO 9629 and custom RMC for 96 digital I/O + 2 C Series Modules

[MattLawson]

Good day,

Forward: Our single DUT tester solution using a sbRIO 9638 and custom designed interface board (motor driver, signal selection, signal conditioning) is working great.

However, the 16 DUT durability cycler project has hit a proverbial wall.

The 16 DUT project was spec'd to use the sbRIO 9629 + both RMC breakout boards (project hardware chosen in collaboration with NI application team).
The sbRIO 9629 documentation and block diagrams indicate both 96 digital i/o and 2  C Series modules could be used at the same time, so I can see where there confusion occurred if they weren't super familiar with the sbRIO products.

Anyway, I now understand through some further reading (and speaking with a Samtec RMC supplier technical support), that simply getting a split 2 way ribbon cable to go from sbRIO to the 2 RMC modules would not work (despite the 96 digital i/o and 2 C Series being on different pins of the 240 pin connector). 

A customer RMC would be required.

They could design something but the cost would likely be prohibitively high for this project and low volume (~10 sbRIO 9629's) and turnaround long. 

They thought an NI partner would be a better bet.

NI has me in contact with Cyth.

I am awaiting to hear feedback if our project needs can be met with their existing custom RMC solutions. 

But I wanted to get the thoughts of the community as well in case they don't have an available solution.  

I can't be the first person to have this problem.....

I/O Requirements for the project

32 high frequency digital Inputs to read SENT signals (>5MHz)

32 medium frequency digital outputs (PWM and DIRECTION motor control outputs up to 20KHz)

32 other digital outputs (Motor and Sensor Enables)

(plan was to use the 96 digital i/o)

80 analog inputs @ 5Khz

(plan was to use built in 16 AI and 2 C series 9205 modules).

NI is also considering in parallel a 8 slot cRIO + 8 slot expansion chassis + multiple NI9205 and NI9401 modules solution.

However, this will be much more expensive (2-3x)  and take up more physical room.

Thanks!

Matt Lawson

Electrical Test Engineer

[RIObotics]

Hi Matt,

It sounds like the biggest challenge for your sbRIO-9629 project is trying to get 80 analog inputs @ 5Khz all into 1 sbRIO board.  Not easy.  I see some options for you, but each with pros and cons.

1. Leverage 2 NI-9205 C-series modules.  Like you said, this would give you a total of 80 AI channels.  However, to physically attach these modules, you need a NI-9697 2-slot mezzanine card adapter.  The downside is that, the NI-9697 only exposes 24 of the 96 DIO lines inside the RMC connector, so this won't meet your DIO requirements.  As a workaround, you could probably use a few of the exposed DIO lines as I2C or SPI interfaces, and then use 3rd-party I2C DIO modules.  Kind of clunky, but it is possible.

2. Keep the 96 DIO lines as-is, and use a NI-9694 board to break them out.  For the AI, use the concept similar to #1, take a few spare DIO lines and make a I2C interface out of it, and find a 3rd party I2C analog input module.  However, keep in mind that the DIO lines on the sbRIO-9629 are only 3.3V.  I see some of your DIO requirements may need 5V or even 24V, so you might need some signal conditioning in between.

3. There are various packaged solutions for the sbRIO.  Here's an example: https://www.wpc.com.tw/gstack-rio.html.  This solution uses the sbRIO as the core, and expands the I/O in a modular configuration.  The FPGA is a fixed personality bitfile that handles all of the module communication.  You can probably get a couple 32-ch AI modules, and a few DIO modules to build your system.

4. Make a custom RMC board that exposes everything, including the 96 DIO lines and the control lines for the C-series modules.  However, I don't know if in software, LabVIEW FPGA will allow you to use 2 C-series modules with 96 DIO lines.  Heck, if you were to go this route, I would just build a 80-ch analog input interface right on the custom RMC board, along with other DIO functionality.  But as you've mentioned, this custom project would take time and money.

5. It sounds like NI's suggestion is to ditch the sbRIO and go with cRIO.  This might be the fastest solution for you right now.  No uncertainties in the system.  But it'll cost you a pretty penny. 

Why am I telling you this?  I used to be a cRIO systems engineer for NI.  I also happen to run a secondhand NI store called Extend Test.  We have plenty of NI-9697's in stock, and can get NI-9694's, which are needed for options #1 and #2.

https://extendtest.co/products/ni-9697

Examples of I2C I/O modules: 

https://store.ncd.io/product/4-channel-i2c-0-10v-analog-to-digital-converter-with-i2c-interface/

https://www.ereshop.com/shop/index.php?main_page=index&cPath=143_179

My company is also the international reseller for WPC Systems, which covers option #3.  You can send me an email for more info: john.wu@tenet-tech.co.

One question, why is sbRIO required?  You could get away with a PC-based solution, such as PXI or cDAQ, it will be much simpler to program.

BR

John

Add motion to LabVIEW in 30min or less - TENET EMotion
Finding it hard to source legacy NI hardware? Try Extend Test

[MattLawson]

Hi John,

Thanks for your detailed reply, I really appreciate it.

In response to:

#1: I don't think breaking out some of the 24 DIO pins on the NI-9697  through I2C or SPI would work.   64 of the 96 pins require high frequency (>=5MHz).

#2:  I think this would be more feasible because the frequency of the analog channels is much less (5 KHz). However, I already had 100% utilization of the 96 digital i/o planned.

In regards to 5V requirement, that interface is performed by the custom interface board (1 per DUT) that we are already using.   This takes in the 5V DI signals from various sensors, performs the required signal conditioning, provides protection and a 3.3V output to the sbRIO.

#3: I will take a look at the custom package solutions, didn't know they existed, thanks!

#4: Thankfully, it seems NI partner Cyth Systems already has such a custom RMC board ready  - provide  inputs for 2 C Series modules and a direct passthrough for remaining digital IO which will be at least 96 DIO lines.  We will retain >5 Mhz bandwidth for all digital I/O.

However, my concern for #4 is what you brought up, is will the Labview support it (even though all the documentation indicates it would).

I will ask the Account Manager from NI and Cyth to get this clarified.

I see your point, as the provided NI solution NI-9697 has 2 C Series and breakout to only 24 digital i/o, the Labview software may not support accessing all 96 digital i/o and the 2 C series at the same time. 

I want to make sure it does before I know this solution will work!

5.  Yes, the cost is significantly different.    ~$40K cRIO vs ~$10K sBRIO.

The cRIO models with similar processor/FPGA as the sbRIO 9629 boards are almost $20K Canadian themselves (plus expansion chassis,  12 NI 9401 modules, 3 NI9205's). 

6. Thanks for the info on the custom interface boards!

7. In regards to "Why is sbRIO required?".  

The code is already completed with the xRIO platform for single DUT tester, and is being extended to 16 channel durability tester.

The sbRIO performs custom SENT position sensor decoding (includes serial/slow channel), custom digital decoding our sensor programming interfaces, motion profile generation and PID control compatible with our existing designs, etc.

I want to leverage all this work that I've already done.

Certainly, other implementations to do this would be possible (i.e. a CDAQ to acquire the analog and extra digital signals, which would talk to 1 or more controllers that would perform the controls and SENT decoding for 16 DUTs).

I'm not really familiar with the PXI platform.

Thanks again.

Matt Lawson