Average

SCPI Commands

FETCh:CDMA:MEASurement<Instance>:MEValuation:LIST:OBW:AVERage
CALCulate:CDMA:MEASurement<Instance>:MEValuation:LIST:OBW:AVERage
class Average[source]

Average commands group definition. 2 total commands, 0 Sub-groups, 2 group commands

class CalculateStruct[source]

Response structure. Fields:

  • Reliability: int: decimal ‘Reliability Indicator’. In list mode, a zero reliability indicator indicates that the results in all measured segments are valid. A non-zero value indicates that an error occurred in at least one of the measured segments.

  • Seg_Reliability: List[int]: 0 | 3 | 4 | 8 The segment reliability indicates whether one of the following exceptions occurred in this segment: 0: No error 3: Signal overflow 4: Signal low 8: Synchronization error If a combination of exceptions occurs, the most severe error is indicated.

  • Obw: List[float]: float Occupied bandwidth Range: 0 MHz to 8MHz (SDEViation 0 MHz to 4 MHz) , Unit: Hz

class FetchStruct[source]

Response structure. Fields:

  • Reliability: int: decimal ‘Reliability Indicator’. In list mode, a zero reliability indicator indicates that the results in all measured segments are valid. A non-zero value indicates that an error occurred in at least one of the measured segments.

  • Seg_Reliability: List[int]: 0 | 3 | 4 | 8 The segment reliability indicates whether one of the following exceptions occurred in this segment: 0: No error 3: Signal overflow 4: Signal low 8: Synchronization error If a combination of exceptions occurs, the most severe error is indicated.

  • Obw: List[float]: float Occupied bandwidth Range: 0 MHz to 8MHz (SDEViation 0 MHz to 4 MHz) , Unit: Hz

calculate()CalculateStruct[source]
# SCPI: CALCulate:CDMA:MEASurement<Instance>:MEValuation:LIST:OBW:AVERage
value: CalculateStruct = driver.multiEval.listPy.obw.average.calculate()

Returns occupied bandwidth (OBW) results in list mode (see method RsCmwCdma2kMeas.Configure.MultiEval.ListPy.value) . To define the statistical length for AVERage, MAXimum and SDEviation calculation and to enable the calculation of the results, use the command method RsCmwCdma2kMeas.Configure.MultiEval.ListPy.Segment.Spectrum.set. The ranges indicated below apply to all results except standard deviation results. The minimum for standard deviation results equals 0. The maximum equals the width of the indicated range divided by two. Exceptions are explicitly stated. The values described below are returned by FETCh commands. CALCulate commands return limit check results instead, one value for each result listed below. The values listed below in curly brackets {} are returned for each active segment: {.. .}seg 1, {…}seg 2, …, {…}seg n. The number of active segments n is determined by method RsCmwCdma2kMeas.Configure. MultiEval.ListPy.count.

return

structure: for return value, see the help for CalculateStruct structure arguments.

fetch()FetchStruct[source]
# SCPI: FETCh:CDMA:MEASurement<Instance>:MEValuation:LIST:OBW:AVERage
value: FetchStruct = driver.multiEval.listPy.obw.average.fetch()

Returns occupied bandwidth (OBW) results in list mode (see method RsCmwCdma2kMeas.Configure.MultiEval.ListPy.value) . To define the statistical length for AVERage, MAXimum and SDEviation calculation and to enable the calculation of the results, use the command method RsCmwCdma2kMeas.Configure.MultiEval.ListPy.Segment.Spectrum.set. The ranges indicated below apply to all results except standard deviation results. The minimum for standard deviation results equals 0. The maximum equals the width of the indicated range divided by two. Exceptions are explicitly stated. The values described below are returned by FETCh commands. CALCulate commands return limit check results instead, one value for each result listed below. The values listed below in curly brackets {} are returned for each active segment: {.. .}seg 1, {…}seg 2, …, {…}seg n. The number of active segments n is determined by method RsCmwCdma2kMeas.Configure. MultiEval.ListPy.count.

return

structure: for return value, see the help for FetchStruct structure arguments.