flicker sensible beings. Some devices are less sensible, mainly those with inductive series reactances, which naturally block higher frequencies and their modulating effect. On the other hand, according to the same reference, electronic compact lamps show high sensitivity even up to the fifth and seventh harmonic modulation. As it was stressed in the paper's conclusions, that in order to capture the harmonics contribution to flicker, it is necessary to use a sufficiently high sampling frequency

more »

... for the input signal to include the corresponding harmonics and interharmonics in the anti-aliasing bandwidth. Since this is not specially recommended in the IEC standards, part of the effective flicker sensation caused by harmonics amplitude fluctuations may not be measured by the standard instrument. Finally, the authors appreciate the discussor's updating the mentioned standards. Since the original version of this paper was submitted in early 1999, the mentioned editions did not exist, as well as much of the added references. The authors should be commended for their interesting paper 1 on transformer modeling. They have proposed an electromagnetic (EM) transients model with parameters determined by measurements. The model is intended to describe the transformer's behavior at the high frequencies of lightning surges and of particular interest is the transferred voltage from primary to secondary. Therefore, the authors can properly neglect the effects due to core saturation and hysteresis without introducing large errors. Although the eddy current effects in the transformer windings are considered in the model, the fitted circuit for that purpose seems to be of an inadequate low order. In Z skin of Fig. 4 , in addition to the dc resistance, the authors have included only one series Foster branch. In my experience, for a frequency range of up to a few megahertz, a model of at least fourth order (four Foster branches) would be required to properly account for the eddy currents in the windings [1], [2] . I believe that the resistance at high frequencies will be underestimated in the authors' model. The analytical expression for the impedance of a winding is [2] Z W (!) = R W / A tanh d 2 0 B coth d 2 where A = 1 3 2m 0 1 2m ; B = 1 2m RW = 1 l ; = j! Manuscript