db units is described as (max-min)*0.001%. If max and min fixed then db unit (and db value) is also fixed - independent from measured value. It's suitable for processing instant values with constant or small-changing absolute noices (for example, frequency in range 40-60 Hz).
But for most measurements (currents, voltages) errors are dependent from measured value and usually described as percents from measured value. So fixed deadband will lead to flood for high values, and to triggering lost for low values.
For the same reason dbAng for measured vector values (currents, voltages, etc) shall be fixed (to avoid flood from near-zero angles).
Also describing dbAng and zeroDb as (max-min)*k% leads to complication of actual value, while this value can be presented in conventional SIUnits.
Proposal
The value of db shall represent:
a- percentage of (max-min) in units of 0.001%
b- percentage of previously transmitted value in units of 0.001%
The value of dbAng shall represent:
a- percentage of (max-min) in units of 0.001%
b- absolute value in units of 0.1 degree
The value of zeroDb shall represent:
a- percentage of (max-min) in units of 0.001%
b- absolute value in units of 0.001*multiplier*SIUnit
Definition kind must be described by vendor in MICS/PIXIT or (future) in a special attribute dbKind.
This tissue also impacts part 7-3 clause 8.
Discussion
Created
Status
I believe the proposal to relate the db in relationship to the last transmitted is reasonable.
Also in protocol IEC 60870-5-101 there are two limits (low and high limit for transmission of measured values). It can be used to adopt the communication to different situation (also dynamically after some time without transmission when the change of the value is too small for the high limit).
21 Oct 15
Discussion (red)
The problem is that deadband technology given by the standard does not provide the necessary functionality and requires to use of additional methods (polling, integrity).
21 Oct 15
Discussion (red)
In general, only SI units shall be used, and if a floating point representation is used, no multiples or submultiples should be used.
When a fixed-point or fractional representation is used, decimal fractions should not be coded in binary. Percentages given in decimal values such as 0,01% may be appealing to the human reader, but not for the computation, so binary fractions should be prefered.
12 Oct 15
Discussion (red)
The deadband ajustment is a compromise between the resolution to obtain at the client side and the reporting behavior. It has to be chosen accordingly.
Moreover, it is not the only means to control the reporting behavior in case of problems, e.g. polling or integrity reports can be used instead.
Finally, the design of the I/O scan and event monitoring processes lead to different implementations, e.g. a device may only issue reports at a minimum 2 Hz rate.
The proposal is not accepted as in a standard, even more in a standard that aims at interoperability, there cannot be two definitions of the same functionality.
01 Oct 15
Discussion (red)
Example: Current measurement range from 0.25 to 250 A. db=100 gives +-0.25 A Measurement accuracy is 3% (simplify, in all range). Measure I=1A gives noice 0.03 A - no flood, but can't see change from 1 A to 1.2 A Measure I=50A (fault) gives noice 1.5 A - network will be flooded during all fault.