lost data from PXIe on Veristant

[bennyk]

Hello,

i am running a model on Versitand and am trying to find the max frequency at which i can run it.

i am using the configuration benchmarking tool to find the model limits. also is there any elaboration of what the configuration benchmarking tool dose. i assume it changes the PCL value periodically and sees if the model runs successfully?. 

i am noticing that on the workspace simple graph within veristand that the graph periodically has a blank period i.e lost data even when i set the pxie to run at the max loop max that the benchmarking tool specifies. i would have thought that the max freq from the configuration benchmarking tool would be sufficent for the model?

is the data presented on the workspace graph the real time data that the pxie is calculating? i am wondering is there not a limit set by the bandwith of the TCP/IP connection? i am comparing the pxie model to that of the simualted model and would expect to see the same shape i.e. steps as this is discrete.

thanks

[AmyNK]

Hi bennyk

Please see the following link for help with the tool:

Configuration Benchmarking Tool for NI VeriStand

http://www.ni.com/example/31441/en/

What version of VeriStand are you using?

In what software was the model created?

In the link above, there is a line which says "Please note, if more detailed timing information about the system is required, it can be discovered by performing an execution trace of the system." - have you tried this?

The 15 second warmup period mentioned in the link suggests that the benchmarking tool is only accurate after this time - could your "real life tests" be failing due to not allowing for this period?

What did the benchmarking tool suggest your loop rate should be? What loop rate are you using?

Amy

[bennyk]

i am unsing veristand 2014

the model was creat4ed in simulink

i am not authoristed to view that executation trace system link for some reason. i have not tried this.

i have two models a simple mdoel and a complex mode

the simple model; bechmarking tool gave 35.0653kHz

the complex model; bechmarking tool gave 15.4358kHz

when i run the simple model at 35kHz PCL rate on the PXIe target i can view the waveform ok on the workspace simple graph. however when i connect an oscilliscope to the 6704 card i am only seening a sqaure wave. i am trying to replicate the wave i am seeing in veristand physically as an analog signal at the pxie terminals

[AmyNK]

Hi bennyk

So the simple model appears OK within VeriStand it 35kHz, but the oscilloscope seems to show problems?

Could you try reducing the frequency, and see if it ever shows properly on the scope? This should give us an indication as to whether it's just the scope that is the problem.

Thanks

Amy

[bennyk]

Hello Amy,

thanks for the reply.

i have simplified the model further and carried out tests. the model is now literally a sine wave to an outport. i have complied and ran the model at 40kz and then at 1kHz. i nothice that the sample period of the digitialed sine wave is always the same with of 1.8ms? in both cases the sine wave is visialbe up to around 100Hz (freq of sinewave set in model). so the max frequency that the card seems to be able to put out is only around 100Hz and this is the case for both the model compiled at 1kHz or 40kHz.

i am confused as to why the sample period with is the same at 40kHz and 1kHz (ive also checked like 20kHz and 15kHz but sample width is the same) i would expect the sample width to change as i change the PCL rate as i would guess the bit calculation is carried out at a different speed? i have a 6704 card that i am measuring the signal from, i wonder is there a max sampling period on this card, as i cant see the 16bit resolution being a problem as regard representing the wave?

thanks

brendan

[AmyNK · Accepted solution]

Hi Brendan,

From the User Manual (http://www.ni.com/pdf/manuals/372110d.pdf) common questions section:

How fast does the NI 6704 sample channels?

The NI 6704 devices have one 16-bit DAC that is time-division multiplexed to create 16 voltage output, 16 current output, and four calibration channels. The channels are sampled at a rate of 50 μs per channel, meaning a channel can change value a maximum of 1.8 ms after it has been updated by software.

So looks like that particular issue is due to hardware restrictions.

Amy

[bennyk]

thank you amy thats great, it definitly solved the 1.8ms mystery 🙂 i was blaming my oscilliscope for a while.

i have also a 6341 card and am wondering if this is faster? i need three analog outputs but dont believe the 6341 has three analog output in any case?

thanks

brendan

[AmyNK]

The 6341 has 2 analog outputs, if you go to www.ni.com and use the search bar at the top, you will be able to find manuals and specifications for the hardware "of interest" (I'm guessing you're looking at the PXIe 6341).

Hope this helps

Amy