Order Analysis Toolkit/ Reference Signal Processing: Even Angle Reference of of phase

[jenz]

Hi,

I try to use the reference signal proseccing for a shaft motion test but it doesn't work and I dont know why.

Please see piture 1:

(A) The aquired even angle signal contains one shaft imperfection in one revolution, which should be compensated. The diagram contains two revolution (No. 252 ... 254).

(B) The diagram of the extracted even angle reference shows also two discontinuities in two revolutions.

(C) Due to a unknown phase shift the correct compensation doesn't work.

Picture 2 shows the LV code. For the DAQ I use a NI-6122 PXI card and a shaft motion test stand with two eddy current distance sensors.

Can anybody help me please?

Thank you!

[dsb@NI]

What device are you using? Accordingly, set the device type (first field in device info cluster input to OAT Build Digital Tacho Info VI). If you have a device that isn't included in the device type ring, set device type to 'Custom' and be sure to set custom device delay to the digital delay for that device. To obtain this delay, you can measure, read the "input delay" from the datasheet, or query the input delay using a DAQmx Channel Property Node.

 

[jenz]

Hi Doug,

using a NI-6122 PXI card and I think it has no input filter.

Jenz

[dsb@NI]

sorry, i should have seen that in your original post. 

Have you been able to verify reference compensation if you treat one of your prox probes (or even the digital tach signal) as an analog tachometer channel? 

And on looking at your block diagram again, you may want to make sure that you are using a good even-angle reference. It looks good in your graph because you update your graph when you Extract RefData? = TRUE. When Extract RefData? = FALSE, the code passes a default value for the even-angle reference. You may want to store the even angle reference in a shift register when Extract RefData? = TRUE, and then use this stored value during compensation when Compensate Even Angle Signal = TRUE. However, the compensated graph does return something, and all we see is phase/time/sample shift, It may also be worth while to stop the DAQmx task after extracting the even-angle reference. Start that task again when performing the reference compensation. It would be nice to verify the compensation code independently of the synchronization code.

[dsb@NI]

Also worth noting that the build digital tacho info, digital tachometer processing, extract even angle reference, and compensate even angle signal VIs require continuous input and each has a reset input. You can monitor the error outputs of these VIs for warnings indicating automatic reset. More than likely, you will have to explicitly handle resets to these VIs in order to maintain synchronization (just on the software side). Please refer to the SVMS examples for reference processing. Here is a screenshot of Even Angle Reference Signal Processing (Digital Tach, DAQmx).vi. Note, the reset input is set to TRUE on i=0 which is also the start of the DAQmx analog input and digital input tasks.

[jenz]

Doug,

thank you for your information. I will reply if it don't work.

[jenz]

Doug,

I adapted the example VI (Even Angle Reference Signal Processing (Digital Tach, DAQmx).vi) to use it with my PXI-6122 card. But it dosen't work. Can you plesase check the attached VI (Step 4 - Reference Signal Processing). Is my DAQ code correct or do I create the phase shifts with my DAQ code?

Thank you!

Main VI: S&V Step 4 - Reference Signal Processing - 3.vi

SubVI: Digital Tacho.vi is a subVI

DAQmx Task: WBM_1E_Distance.png 

[jenz]

Does anyone have an idea?

[robo_Jeff]

Hi jenz,

I'm reviewing the code and I'll see what I can find.

[robo_Jeff]

The PXI-6122 has a 20 MHz timebase for the counter, which has 24-bit resolution, or aprox 16.8 million counts. That means the counter is going to roll over about every .84 seconds. This could be a problem at the low speeds of slow-roll compensation.  It's likely there is a counter rollover that's not being accounted for.

What you could try is performing period measurements with the counter, instead of edge counting. That should resolve the issue.

The PXI-6122 spec doc is here, for your reference.