First, a lot of thanks for the pointer, that could really help.
What I really want to accomplish is to get at least a medium bandwidth interface
between the LAN and some real-time data aquisition units. The LAN side is
patterned after my Agilent 89441A vector signal analyzer. You simply open
port 5025 on 192.169.178.111 and dump/read there GPIB/IEEE488-like commands &
data streams. That side seems to work, although there is not yet much flesh
to it since the data collection side is still missing.
First I wanted to try the SPI interface, then the 16 bit multiplexed bus
to get an intelligent register interface to some FPGA.
For the SPI, I decided to attach a LT2500-32 ADC; that can do 1 Msample
at 24 bits. Add FFTW in the BBB and you have a respectable Fourier analyzer.
For the hardware, that’s all that is needed:
<
The Xilinx Coolrunner2 generates the sampling clock from the 100 MHz crystal osc. and collects
some left-over gates. $1.50 or so. The other small board is the ADC, its regulators and reference.
Home-etched and soldered. It’s completely open, in case someone wants it.
The SPI has soaked up much more time than I had planned. A new Debian image from
Robert and some other insights at least made that I don’t get a bus error for each SPI access.
One gets thankful for small advances… At least I can now create and start PRU programs,
talk to them via the shared RAM and transfer huge data blocks to Linux virtual memory land
through ping-pong buffers. It’s just that I can’t make the SPI say a word. Verbatim.
I also can finger the SPI pins when I re-assign them to the PRU and use R30/R31.
I also have a Red Pitaya. I like it architecture-wise, except that I’ll need a larger FPGA
to support fast ADCs with JESDI204B ports - and that it is based on that exotic Alpine Linux.
FPGAs are a home game for me, I’ve used them since there has been Xilinx.
best regards,
Gerhard