Tach Profile on a sbRio 9612

[erfigge]

I have a sbRio 9612 that I am using for pulling in a vibration and digital tach signal.  All the examples I find are all based of delta sigma modules and from my understanding the built in analog and digital inputs on the 9612 use the top level 40Mhz clock.  So, what is the easiest way to get a tach profile pushed through from the digital inputs to the real time?

Thanks,

Eric.

[preston_johnson]

You might consider a review of these:

http://sine.ni.com/cs/app/doc/p/id/cs-816

http://zone.ni.com/reference/en-XX/help/372416F-01/svtconcepts/svdigtacho/

http://www.ni.com/white-paper/3230/en

[erfigge]

Thanks for responding. However none if those really help with my specific issue. I need to push a DMA FIFO of tach time stamps through so I can use the order analysis routines. I can push a DMA FIFO through, but I'm stuck with generating the time stamp values. All the examples I find are explicitly for delta sigma modules and I'm not for sure how do do that with the built in digital input on my 9612 which is not delta sigma.

Thanks,

Eric.

[preston_johnson]

The time stamps used by the Digital Tachometer VIs in our Order Analysis Libraries are actually event counts, counts of the clock pulses driving the A/D channel. 

So lets say you are using the 40Mhz clock to pace the A/D (software timed in the FPGA) at 20kHz.  You could count 20kHz clock pulses (the AI convert logic) or any multiple of 20kHz. 

The counts (an array of integers, one count for each pulse detected) is then fed into the DMA fifo and used by the digital tachometer functions from the order analysis library in the Sound and Vibration measurement Suite. 

Let us know if this helps. 

[erfigge]

Preston, thanks for the reply. In the examples the hardware clock is used, could i just delay one tick in the middle loop and then every time the loop runs consider that a clock cycle front edge event? The example I'm looking at is this: http://zone.ni.com/devzone/cda/epd/p/id/6301

[preston_johnson]

I think that is right, just replace the "mod1 timebase" input to the U32 counter with your reference pulse from the A/D loop. 

[erfigge]

Preston, I found an example of yours that might make things even easier for me.  I found this message of yours: http://forums.ni.com/t5/LabVIEW/Order-analysis-speed-profile-from-measured-data-in-Hz/td-p/2229528

I pretty much have the same thing, at least as it currently sits, I have an RPM signal that comes in from a scada system.  So, for now instead of trying to build our system to have a seperate speed pickup I wanted to use your example.  However, I don't understand what I need to fill in your "TACHO SCADA" vi.  I am reading in the vibration data and doing and FFT for magnitude versus speed, and after that wish to do an order analysis so I can get the magnitudes of the 0.5, 1, 2, 3, and 4 multiple's.  Also, I am sampling at 32500 Hz on the FPGA. 

So, in the picture of my code below, if I know my rotational speed is 3600 rpm, how do I need to give that to the "TACHO SCADA" vi to make the speed profile?

Sorry for probably asking dumb questions on this and thanks for your help.

Eric.

[preston_johnson]

Here are a couple of comments. 

1) for your waveform data coming from the FPGA, you should consider the waveform reference design for CompactRIO: 

http://zone.ni.com/devzone/cda/epd/p/id/6206

this will insure your data is consistantly and accurately recorded, and that you do NOT burdon the CompactRIO controller with converting to floating point from fixed point

2) if you speed does not change, then you do not need a speed input at all, you can simply perform a power in band measurement on the FFT to get the magnitude (and phase) information you need, based on the power in band measurements.

3) If your speed changes, you will need to read the speed from the control system (SCADA) as fast as you can, and time stamp the speed values from the controller.  The SCADA tool then outputs a square wave analog pulse that can be used by the NI Order Analysis tools

http://sine.ni.com/nips/cds/view/p/lang/en/nid/12050

Hope this helps

[erfigge]

Preston, your point number 1 is a great link.  I will start implementing that as soon as I can get the order analysis working. 

As for the speed, it does change, but not very fast.  So, I have 2 clarification questions.  What format should the time stamp values be, actual time stamps, time since the last point ect?  Also, should I read and fill the data at the same rate as the actual vibration data is being grabbed, or does that matter?  I guess that leads to one more question, what is a reasonable number of values to have in the array?

Thanks again,

Eric.

[erfigge]

An update, I worked through the waveform crio examples and that has worked out great. I am still stuck at what exactly is going on with the frequency for the order analysis. I have the scada tacho vi, and even if I follow the other examples I just don't understand what is being passed through.

I have a frequency waveform in the realtime controller as technically that's what I'm being feed. What do I need to do to take a waveform (0's and 1's) and make that into what the order analysis needs?

Thanks again,

Eric.