Roach2 sync/1PPS input options are described at : https://casper.berkeley.edu/wiki/ROACH-2_Revision_2#Usage_Manuals.2C_Guides.2C_Memos.2C_etc.
Roach1 sync/1PPS input options are described at https://casper.berkeley.edu/wiki/ROACH#Usage_Manuals.2C_Guides.2C_Memos.2C_etc. On Tue, Jun 24, 2014 at 4:29 AM, Jack Hickish <[email protected]> wrote: > Hi Ramesh, > > There's no external connector to input a 1PPS signal, so if you need > one you'll have to wire it straight into the FPGA via GPIOs. Both > ROACH1 and 2 have some SMA GPIO inputs you could use for this purpose. > > Cheers, > Jack > > On 24 June 2014 13:04, Ramesh Karuppusamy <[email protected]> wrote: >> >> Hi Jack, >> >> Thanks - I ended up using the concat/convert modules from Xilinx blockset to >> typecast and tie the outputs of sample_and_hold modules. >> >> Does the x64_adc accept an external 1PPS as in iADC or the 5GSPS ADC? We >> normally use this signal for absolute timing of our data streams. >> >> Cheers, >> Ramesh >> >> On 24 Jun 2014, at 10:52, Jack Hickish <[email protected]> wrote: >> >>> Hi Ramesh, >>> >>> When the data gets to your simulink design it will be multiplexed with >>> 4 adc channels per adc yellow block output, so you'll have to demux >>> yourself or deal with the mulitplexing in your code somehow. I don't >>> think there's a casper block to demultiplex this for you, though there >>> is a serial-to-parallel block in the Xilinx library which probably >>> does what you want. >>> The sample and hold block in the setup you describe should work too >>> (with a sample rate of 4 cycles). You would want to wire the sync >>> input to the chan_sync output of the adc block, taking care to >>> properly account for the timing of the chan_sync relative to the data >>> (I can't remember if it precedes adc channel 0 by a clock cycle, or >>> comes at the same time -- the documentation should tell you). >>> >>> As you suggest, you can combine 4 12 bit outputs with the bus_convert >>> block to be clocked into the 10GbE block -- just cast from Fix12_11 to >>> Fix16_15 format. >>> >>> Cheers, >>> >>> Jack >>> >>> PS. don't forget to run the adc calibration script linked on the wiki >>> before trying to use the adc data! >>> >>> >>> >>> On 24 June 2014 09:27, Ramesh Karuppusamy <[email protected]> wrote: >>>> >>>> Hello All, >>>> >>>> I am seeking some guidance on a simple total power detector system that I >>>> am putting together. >>>> >>>> I would like to sample 8 to 16 signals at a low rate of 10 KHz with the >>>> ROACH1 and transmit the data on the 10GbE. Someone did suggest a simple >>>> USB-based ADC, but we like the ROACH based set up, as it involves little >>>> work on the receiving side. Therefore, I purchased 64-input 12-bit ADC >>>> board. The plan is to clock the ADC at 20 MHz and downsample to 10 KHz >>>> inside the FPGA. >>>> >>>> I read the yellow block documentation at >>>> https://casper.berkeley.edu/wiki/X64_adc, which says the FPGA will be >>>> clocked 4x the ADC rate and that the outputs are multiplexed on the >>>> dout<n> pins. Is there a casper block to de-multiplex the data on these >>>> pins? Alternatively, can this be a bank of 4 “sample_and_hold” (found >>>> inside Casper DSP block set) with 4-cycle delay, with the inputs delayed >>>> by zero, one, two and three cycles? >>>> >>>> Now can the four 12-bit outputs above combined using a bus_convert to a >>>> 64-bit input that can be clocked in to the 10GbE block? >>>> >>>> Thanks for your time! >>>> >>>> Cheers, >>>> Ramesh >>>> >>>> >> >
