A forum expert recently opined:
Unless there’s a compelling reason NOT to, always use CURRENT sources instead of voltage sources.
Like most online advice, the writer did not give reasons for is advice, and given the strength of it, an explanation is warranted.
Of course, many people giving compelling advice online don’t actually understand the matter and cannot give a rational explanation.
So, what is the difference between the application of current and voltage sources.
If there is a single source used in the model (the vast majority of models), then it make no difference to the validity of the model and results. Specifying a current source might scale the current report to more convenient / normalised values, but equally specifying a voltage source might be more convenient. So, use whichever is convenient for the purposes, and unless you dig into the text reports, it probably doesn’t matter in most cases.
On the matter of scaling, some tools rescale the calculated values for some given input power so that reports and graphs are in terms of that nominated input power (eg 4NEC2) and in that case, the reports are not simply in terms of current or voltage specified for the source, the magnitude and phase of the single source become largely irrelevant to the output reports.
If there are multiple sources, then specifying current sources allows direct control of their relative amplitude and phase and this might be of benefit for instance in a cut down model of a phased array where the phasing harness is not included in the model but multiple elements are driven directly by dedicated sources.
Here is an example of specification of a current source of 1<0 connected to tag 5 segment 1.
EX 6 5 1 0 1 0 0
This is NOT portable to NEC engines that do not support current sources, so that is good reason to NOT use current sources unless you have a compelling reason.
What if your version of NEC does not support current sources and you have a model where current sources are of advantage?
Early versions of NEC do not natively support current sources. Nevertheless a current source can be synthesised by a voltage source followed by a two port network specified by its Y parameters on an NT card.
Here is synthesis of a current source of 1<0 connected to a two port network between tag 5 segment 1 via wire tag 9901.
GW 9901 1 -0.0218426 0 9901 0.02184257 0 9901 1.09213e-3
EX 0 9901 1 0 0 1
NT 9901 1 5 1 0 0 0 1
Note that the auxiliary wire element needed to be created and the NT card to describe the two port network to link the new excitation connection at tag 9901 segment 1 to the original feed point at tag 5 segment 1. (I will leave it to the reader to brush up on Y parameters and the two port network.)
This may be done ‘under the covers’ for you, and you would be unaware of the detail unless you export the file to another tool which may not properly handle the synthesised element (even though it is bog standard NEC).
Again, woes of cross tool support provide good reason to NOT use current sources unless you have a compelling reason.
I receive lots of NEC model files from people asking for help. In most cases, they have used a synthesised current source as that is the default for one of the free demo packages. Invariably, these files are not opened by my preferred modelling front end, and to run them, I need to edit them by hand to fix them… which means any changes I make may not run on the correspondent’s tool.
Essentially, if they have used current sources, I don’t look any further and cannot help.
- Though it is popularly held that current sources should be the default choice, for most models there is no advantage and possibly portability disadvantages in using current sources.
- Unless there’s a compelling reason NOT to, always use VOLTAGE sources instead of current sources.
- Compelling advice without explanation is by dummies for dummies.