Hi David,
In regards to your post on May 24th:
Custom PWM functionality:
If you are talking about scan mode with a local cRIO chassis, you're currenlty limited y the functionality provided by Specialty Digitial, which consists of a fixed set of frequencies. This functionality is provided by the NI-RIO team, and is not likely to change any time soon. Since we also don't currenlty have a way to do hyrid mode from a cRIO in NIVS, my recommendation would be to not use the scan engine custom device, and to use an FPGA personality instead. Once hybrid mode is possible, it should be possible to use scan mode and implement the custom PWM in FPGA.
For an EtherCAT chassis, you can currenlty make a hybrid personality that uses scan + UDVs to communicate with your custom PWM code on the FPGA.
Simultaneous switching:
This behavior is not specific to the scan engine custom device; the custom device simply reads all relevant channels from NIVS. Since NIVS updates all channels each iteration of the loop and performs all mappngs at the same time, it is possible to use this for simultaneous updates. For example, you could have a single User Channel mapped to multiple digital outputs. When the User Channel changed, all digital outputs would update simulataneously. You could also direclty update all the digital lines from a stimulus profile and have them all update simultaneously (it is possible to update multiple channels in one time step).
In regards to your May 28th post, I agree with everything Stephen said. To directly answer your questions though:
a) EtherCAT through the Custom Device: it would be great because the "just configurable" approach, but I guess it won't work for the same reason (shortage of DMA channels) becuase I cannot use XNET, Scan Engine and Custom FPGA with NI Veristand on RT target. What do you think? In case it didn't work, would you have any solution development in the short term? In any case, what's the loop rate limitation of EtherCAT (1KHz? Would it work with 2KHZ loop rate, accepting to have the I/O updated every 1ms?)
This sort of configuration will work. The DMA FIFO limitation only applies to the FPGA of the 9082. Since the scan mode will only be used to support the EtherCAT chassis, and not the local FPGA, no local FIFOs on the 9082 will be used. So you can have custom FPGA + XNET on the 9082 and still use an EtherCAT chassis with the scan mode custom device. The max EtherCAT loop rate will vary greatly depending on the RT controller and the total amount of I/O channels. With limited I/O and a fast controller (PXIe-8133 or PXI-8108) we have seen rates >8kHz. So, for the 9082 I don't expect 1-2 kHz would be a problem
In the past a colleage ran a benchmark with a 9082 (running in single-core mode) and the scan mode custom device in NI VeriStand. The benchmark included both local and EtherCAT channels, a total of 184 analog channels, 64 digital channels, a model, and 4th order filter calculations. With the scan rate at 2kHz in this setup he was running at 48% CPU. So hopefully that should give you an idea of the approximate performance you can expect.
b) Ethernet expansion with NI 9148: hoping that 0,5ms (or 1ms) updates will be supported on a dedicated network, I'm worried about the effort required in order to allow easy and configurabile integration into NI Veristand? Do you have specific resources? What do you think about?
I haven't really used the 9148, so I can't confirm the rates, but I don't expect you could reach 1-2kHz single-polint updates. It uses non-deterministic Ethernet communication, so there are no guarantees on communication latencies. It also is not natively supported in NIVS and would need a custom device written for it. This code could maybe be based on the scan/EtherCAT custom device, but the RT code would need to e rewritten to accomodate the non-scanned IOVs. It would take a lot of time and coding to implement this, and the end result would not be as reliable as EtherCAT. So, I would not recommend this approach at all.
Hope that helps.
Regards,
Devin
