3 different I/O Nodes, getting data over to RT efficiently

[jonathandkr]

I am currently writing FPGA code (consider me a developing novice when it comes to programming), and have come across a few questions.  Before I pose the questions I'll shed light on the needs of the FPGA vi I am writing:

data acquisition I/O of:
16 thermcouple inputs (4 tc modules 9211s)
16 4-20mA inputs (2 9203s)
1 Digital Input (magnetic RPM pick-up via the 9401 module.... is there another way to bring this pick-up in?)
9 Analog Outputs (16 AO module) .... with 7 of thes AO channels I'd like to have PID capabilities for these.
Possibly using a 9237 to bring in my load cell/strain gauge input (I was toying with the notion of using a 4-20mA signal transmitter to amplify the load cell signal so as avoiding having to run a higher speed loop with the 9237 module).

I am looking to bring in the analog inputs as efficiently as possible, and thus came up with the question of whether or not I could have three separate while loops (loop 1 containing the thermocouple modules, loop 2 containing the 4-20mA modules, and loop 3 containing the digital I/O module) feeding into one main DMA FIFO.  I had come across a bit of info that this may screw up the order of the DAQ info coming across to the RT, but I still wasn't sure on this.... any input on how this would play out?   I could possibly do 1 other thing with both the TC and 4-20 mA nodes ( to combine the TC inputs with the 4-20 mA inputs in the same I/O FPGA Node, wire all of these inputs into a build array, and then wire the array into the DMA FIFO .... though would this be too many inputs per one node? 32 inputs.  And what of sampling the 4-20mA inputs 10 to 15 times per second, and only sampling the TC inputs at 3 to 4 times per second, would this screw anything up?

I possibly could take all the TC channels and build them into a cluster via a bundle, and use Front Panel Comm to pass the data from the FPGA to the RT... or I could possibly just use an interrupt for the TCs, and a DMA FIFO for the 4-20mA.

Where I am pretty hazy at right now is in regards to how efficient data transfer techniques from FPGA to the RT really are.



Message Edited by jonathandkr on 03-01-2008 12:24 AM

[xseadog]

Hi J,

interesting application.
Q1: What actual platform are you planning to use Compact RIO or ?

Q2. With regard to your thermocouples how often do you need to read them? Every second or what 30 secs. Thus you can reduce the data throughput if it is not essential to update so frequently.

Q3. Why don't you convert from mA to voltage thus reduce from 2off   9203 modules to 1off  9205 module. It also leaves you with the option of differential or referenced.

Suggest that you check out the tutorials on FPGA.'s

A link to get ya started:
http://zone.ni.com/devzone/cda/main

xseadog

[jonathanupr]

Reply to Q1: using the CRIO platform with a 1m chassis
Reply to Q2: I do need to keep a continuous measurement on all transducers (engine test-stand situation)
Reply to Q3:  the biggest reason for using the 9203 is mainly due to the transducers that I am using:  pressure transducers with built in amplification & transmission that give a 4-20mA output.  I could use unconditioned transducers though i would have to break-out a signal conditioning box, wouldn't I?

thanks

[xseadog]

Hi J,

with regard to your answer to Q2.What type of engine performance monitoring are you doing?

We utilise up to 256 thermosensors on our engine testing, where we can burn up to 10 tons of gas oil /hour Usually we are interested in the steady state perfomance and thus scan every 30 seconds as the system is thermally stable. However we do have the option of faster scan rates, if necessary.

You also mentioned CRIO, 1M gate, how many slots 4 or 8? Your configuration would have to be 8 to enable monitoring of all sensors.

Why don't you use a break out box to convert the current readings to voltage? A cheaper solution.

xseadog

[jonathanupr]

hi x,

I am using an 8 slot chassis at 1 m.  You mentioned it being cheaper for a breakout box, but on my end it seems so much cheaper to just take 4-20mA inputs from already amped pressure sensors (I've been looking at the majority of the products out there in regards to bringing in current and converting to voltage...though that doesn't seem to save any cash....  what does your DAQ set-up entail?  it may shed some light on how you are able to save some dough).  I have only used 4-20mA pressure sensors in the past, though I'd imagine they do make amplified pressure sensors giving a voltage output...and thus using the 9205 I'd definitely save a hell of a lot of cash with 2-4xs the inputs than the 9203 module.  In regards to the thermocouple inputs, what is the most cost effective method?  It seems that the 9211 modules aren't too shabby for the price, and the other signal conditioning modules (non-CRIO) were just as costly if not more.  I could always get a 4-20mA transmitter and convert the thermocouple voltage to the 4-20mA (if I needed to save space on the CRIO chassis).

The present application is in regards to making a DAQ system that gives our customers a wide array of choices for DAQ on their testbeds... It sounds like you guys hit it pretty hard with data-acquisition.  I feel greener and greener the more I read these boards, that is in skill level.

In regards to my first questions, anyone else have any other info?

Thanks for the reply X.

[xseadog]

Hi J,

 suggest you read up on thermal couples you actual signal level is usual in range of uV, though this does depend on temperature.

Also on what type of thermal couple you are using.

http://en.wikipedia.org/wiki/Thermocouple


With regard to your pressure sensors , that surely would depend also on the working temperature and pressure.ie sensors is designed to function in a particular environment, thus the cost  increases. In last application  the unit cost was  3000 dollars  and we had to use  20 ! however  the results justified the cost.

xseadog

[jkurtw]

Hey Jonathan,

Your application looks like it could benefit from the new programming model for CompactRIO introduced with LabVIEW Real-Time 8.6. However, you would need a 3M gate chassis and a 9012, 9014, or 9074 controller.

The new model, CompactRIO Scan Mode, can handle the RT/FPGA communication for you on I/O modules reading and writing at 1kHz or less, and you can also use LabVIEW FPGA on the other higher speed modules.

You can check it out here:

http://zone.ni.com/devzone/cda/tut/p/id/7338

 thanks,

Kurt