[TowerTalk] Modifying Comtek 4-square controller(relay unit) for a different band?

[Michael Tope]

It looks like he introduces a little bit of resistance into each winding 
so the Y-matrix doesn't blow-up and the rounding errors aren't 
excessive. From the EZNEC+ v.5.0 help file:/
/

/"The models are not quite ideal. Because of the way they're 
implemented, the transformer models actually have a small but usually 
insignificant resistance in series with each winding. The value of this 
resistance is based on the specified port impedances, such that if the 
specified impedances are in the ballpark of the impedances to which the 
ports are connected, the resistance will be insignificant. There's quite 
a bit of leeway here, however, so you don't have to be precise about it. 
//Please note://The specified impedances aren't winding impedances, and 
they aren't the values of the series resistances. They should be chosen 
as explained."/

73, Mike W4EF....................

On 11/21/2021 12:02 PM, Lux, Jim wrote:
> On 11/21/21 11:11 AM, Michael Tope wrote:
>> Hi Jim,
>>
>> Starting with version 5.0, EZNEC implements the capability to model 
>> lumped transformers, inductors, capacitors, and resistors. Thus, it 
>> should have capability to model the Collins 90deg hybrid. The only 
>> caveat is that the transformer model is ideal (K=1.0), so including 
>> the impact of flux leakage and core loss would have to be captured 
>> with additional components in an equivalent circuit model. 
>
>
> I wonder how they do it. An ideal transformer has no Z or Y matrix - 
> for an ideal transformer you can't express the voltage in terms of 
> current or vice versa.  And yet, various analyses do use transformers 
> and admittance matrices.
>
> I1 = Y11 V1 + Y12 V2
>
> I2 = Y21 V1 + Y22 V2
>
> You can't find Ys from I1 = I2 / N and V1 = V2 * N.
>
>
> Lumped L R and C is easy.
>
> Maybe an extension to the underlying NEC core? But NEC is all about 
> creating a big Y matrix, and then solving it.
>
> _______________________________________________
>
>
>
> _______________________________________________
> TowerTalk mailing list
> TowerTalk at contesting.com
> http://lists.contesting.com/mailman/listinfo/towertalk