Embedded Data Logger data drop out

[pmac]

I am using the Embedded Data Logger custom device that ships with NI Veristand 2013. When looking at the data in the TDMS file that the data logger generates there is a periodic "dropout" of data on all channels. This occurs approximately every 2 minutes and lasts for 1 to 2 seconds. See images below. The first one shows two data drops occurring in 3 minutes of data and the second is zoomed in on one of those drops.

I tried adjusting the FIFO depth in the data logger setup with values ranging from 500 to 50000 with no affect. I also removed any additional channel groups, reduced the channel count to 1, and tried different sample rates, all with no affect on the data drop outs. Has anybody seen this before? Any ideas what might be going on?

Thanks,

pmacArrows indicate data dropsZoomed in data drop

[pmac]

Since I started this thread I have done a lot of investigating on this issue and have determined that it isn't a problem with the Embedded Data Logger Custom Device since I can watch the dropout occur using the Graph Display without any data being logged. It appears to be an issue with my reflective memory devices.

I have two targets in my system and each target has a GE Fanuc 5565 reflective memory device. I use the Veristand mapping tool to pass data between the two targets via the reflective memory network. I have ~200 channels mapped from target#1 to target#2 and only 35 mapped from target#2 to target#1. I am seeing data dropout on all 200 channels mapped from 1 to 2 and on no other channels in the system.

Is there a known bandwidth limit. Is there something I can do with the RM configuration to fix this?

[joshe]

Hi PMAC,

The issue isn't with the Reflective Memory, but with NI custom device in VeriStand with the RM.  If you have a lot of channels, which you do, the RM Custom Device cannot right or read fast enough and if you look at memory usage, I would bet it is really high.  The way NI implemented the RM Custom device is to look at a channel, write or read, then go to the next channel.  This is very inefficient when using reflective memory.  When doing a read or write you want to do it as a block (all the channels at once), this is much faster.  We requested some custom device code from NI that does the block write/read, and then modified it to meet our application.  Talk to NI and see if they can provide something.

Good Luck