Uploaded image for project: 'SAFe Program'
  1. SAFe Program
  2. SP-1570

Assess additional firmware features 

Change Owns to Parent OfsSet start and due date...
    XporterXMLWordPrintable

Details

    • Feature
    • Must have
    • PI12
    • None
    • Obs Mgt & Controls
    • Hide
      • for each point, report a scope of work, estimate of size, proposed test strategy & requirements
      • Priority assessment for each feature - when needed, link to roadmap,
      • risk/concern addressed, other elements affected.
      Show
      for each point, report a scope of work, estimate of size, proposed test strategy & requirements Priority assessment for each feature - when needed, link to roadmap, risk/concern addressed, other elements affected.
    • 2
    • 5.5
    • Team_MCCS
    • Hide

      SP-1570 captures a lot of work that is critical to the sucess of the MCCS firmware implementation. However this feature was to large to be achievable in any single PI. However the information captured within the description preempted how the activities could be broken down into more manageable PI constrained features. As such SP-1570 has been suceeded by the following Features:

      SP-1962
      SP-1963
      SP-1964
      SP-1965
      SP-1966
      SP-1967
      SP-1968

      Show
      SP-1570 captures a lot of work that is critical to the sucess of the MCCS firmware implementation. However this feature was to large to be achievable in any single PI. However the information captured within the description preempted how the activities could be broken down into more manageable PI constrained features. As such SP-1570 has been suceeded by the following Features: SP-1962 SP-1963 SP-1964 SP-1965 SP-1966 SP-1967 SP-1968
    • 11.5
    • PI22 - UNCOVERED

    Description

      Additional feature have been proposed as result of CDR suggestions and ECP. These features were not present in the SPS DDD document. Some of them require an ECP to be raised, or respond to an ECP which changed L1 specifications. Some of the requirements do not have an ECP and are desirable, but not necessary to achieve.

      1. Improved filter design. In section 4.2.8 of the Low Signal chain document SKA-TEL-SKO-0000793 it is recommended to improve the combined LFAA+CSP frequency response, increasing the stopband rejection. This is possible to achieve by optimizing the LFAA filter tap coefficients, without changing the hardware design. As the ripple in the combined response is lower than the sum of the ripple in the two responses it is possible to trade more passband ripple for improved stopband attenuation. Most of this design work and the associated impact on the SKA-low spectral response has already been completed.
      2. Increased number of independent beams. SKA1-SYS_REQ-3039 requires up to 8 independent beams per station. Increasing the number of beams has been suggested during element CDR. Up to 48 beams can be easily accomodated without mayor modifications in the firmware architecture, and is sufficient for most applications. As in the current design, dynamic beamforming (source tracking) uses internally calculated phasors, calculated from delay and delay rate values specified by MCCS for each antenna. 
      3. Holographic mode – ECP-190004. Holography is a powerful calibration technique, demonstrated in LOFAR. In this technique a number of beams at least equal to the number of antennas are formed for a station, and pointed in different directions around a known source. By numerically inverting the beamforming process, calibration coefficients for each antenna can be derived. This mode can be impemented with minimal firmware modifications using MCCS provided beamforming coefficients, which include a geometric phase term. Each station computes 384 beams for the same spectral channel. Source tracking is not performed internally.
      4. Event capture of antenna samples - ECP-170037. This ECP requires antenna samples to be captured in a circular buffer, and the buffer to be frozen and transmitted in response of an external event. A prototype of the sample buffer has been implemented and used to collect digitized samples, for ADC and clock characterization. This must be integrated with the current machanism which allows samples to be transmitted to the MCCS server. The mechansim to provide a trigger, and to distribute it to TPMs must also be defined.
      5. RFI detection in the beam samples. As part of the Low signal chain review, it was suggested to include a RFI detection stage in the final stage of the station beamformer. The detection scheme shall be similar to the broadband RFI detection described in section 6.8 of SKA-TEL-LFAA-0500035 LFAA-SPS-DDD. It would allow detection of relatively narrowband impulsive RFI, with timescale of 0.2-4 ms and bandwidth of the order of the channeliser spectral resolution (0.8 MHz). It should be noted that fast transients have a signature comparable to these values.
      6. CSP packet sequence. Current firmware generates packets of beamformed samples for CSP using the same timing for each station. As a consequence, packets typically collide in the switch which aggregates the traffic on the LFAA  to CSP links. The switches operate in a “store and forward” mode, so packets are not lost, but packet arrival order to CSP is undeterministic. This problem is worsened if 100G links, with aggregation of 6 stations per link, are used, as proposed in the new LFAA-to-CSP ICD. Although some out-of-order packets are unaviodable, more deterministic packet ordering is possible if a slight time offset is added in the timing sequence for each individual station. Modification in the firmware is minor, but the exact implementation depends on the switch behavior so some trial-and-error will be necessary. No other modifications are expected in the firmware for the proposed change of links to CSP from 40G to 100G. In particular the link in the TPMs will still use the 40G standard. Packets will arrive to the cabinet switch with sufficient skew to be properly sequenced on the output 100G link. 
      7. Station beam power integrator. The last tile in the beamformer chain could integrate the station beam power and transmit integrated data to a programmable IP address as it is done for channelised data and tile beamformed data. This would facilitate health check and error detection. Currently the station beam power is calculated and integrated by the DAQ software, however this only supports a limited bandwidth and requires an high computational load. 

      Attachments

        Issue Links

          Structure

            Activity

              People

                j.santander-vela Santander-Vela, Juande
                r.braddock Braddock, Ralph
                Votes:
                0 Vote for this issue
                Watchers:
                0 Start watching this issue

                Feature Progress

                  Story Point Burn-up: (100.00%)

                  Feature Estimate: 2.0

                  IssuesStory Points
                  To Do00.0
                  In Progress   00.0
                  Complete11.0
                  Total11.0

                  Dates

                    Created:
                    Updated:
                    Resolved:

                    Structure Helper Panel