Large rms seen in I-Dig



    project p2030 (pulsar search) uses both groups of the mock spectrometer. The setup is:

    On 18jun16 they (jason) reported that  b6s1g9 (beam 6, band1, group 1)  consistently had a higher power level than the other beams. The  output from the adjust power measurement showed:

From: p2030.cimalog_20170618
                                                                       PolA           PolB
                                                                   DigI   DigQ    DigI   DigQ
2017-Jun-18 00:51:26 INFO3   from_mock: From MOCK:  b6s0g0  rms:A  30.2   27.1 B  27.8   29.1 mn:A  -0.2    1.5 B  -0.4    0.6 gn:  9 14
2017-Jun-18 00:51:26 INFO3   from_mock: From MOCK:  b6s0g1  rms:A  28.1   27.2 B  27.4   29.2 mn:A   0.3   -0.1 B   0.5   -0.0 gn:  9 14
2017-Jun-18 00:51:26 INFO3   from_mock: From MOCK:  b6s1g0  rms:A  31.5   28.1 B  29.7   29.7 mn:A   0.7    0.3 B  -1.6   -1.0 gn: 11 16
2017-Jun-18 00:51:26 INFO3   from_mock: From MOCK:  b6s1g1  rms:A  63.7   28.2 B  60.9   30.2 mn:A   0.4    0.3 B  -0.2    0.1 gn: 11 16

The problem was with the bottom line : b6s1g1
  • b6s1g1 voltage rms for  digI polA and PolB was 60 counts
  • b6s1g0 voltage rms for digI polA and polB were about 30 counts..
  • But the signals for b6s1g0 and b6s1g1 come from the same power splitter. so the values should not be that different (6db).

Looking back through the p2030 cima logfiles, the problem first occurred on 03apr17.

    Felix and I did some snooping on 19jun17.

  • The signal path is:
    • 250 If ->   mock baseBandMixer ->  power splitter ->     mock b6s1g0  digitizers
    •                                                                                      ->     mock b6s1g1  digitizers.
  • We looked at the power levels going into the b6s1g0, b6s1g1  digitizers, and they were the same..
  • We swapped the baseband inputs, and the problem stayed with b6s1g1...
  • So the problem was in the mock spectrometer box b6s1g1...

    I took some cal on,off data (while the problem was present). I wanted to see of the problem was a gain problem ,  or added noise...

The plots show Tsys for the 7 alfa beams, for both groups (.ps) (.pdf).

  • Each frame is a beam from alfa (0..6)
  • This data was taken with the alfa cover on (in preparation for a radar run).
    • The amps see a reflection off of the cover of the signal input to the amps).
  • the colors are different pols and groups:
    • Black polA group 0, Green PolA group 1
      • These come from the same IF signal
    • Red polB group 0, Blue polB group 1
      • These come from the same IF signal.
  • Looking at page 2, bottom frame, there is no different between the Tsys of b6s1g0 and b6s1g1..
    • So the problem is not that extra noise is being added to the system...

    The  b6sxg1 spectrometer box was reset. This caused the fpga's to be  reloaded.
After reloading the fgpa's, the problem went away.....

    On 20jun17, the mock spectrometer tests were run. We found that:

  • b2s1g0, b6s1g1, b1s1g1  were failing the local test..
    • I ended up cycling the power on all of the spectrometer boxes.. After this all of the spectrometer passed all of the tests (more info)


  • b1s1g1 digI for polA and polB had voltage rms's twice as large as what was expected.
  • resetting the spectrometer boxes fixed this problem.. the fgpa's were reloaded.
  • Some of the other mock boxes failed the ltest (local transfer) diagnostic.
  • After cycling the power to the spectrometer boxes, all of  the boxes passed all of the tests..


<- page up