Energy transfer question

hi, im new here, and here is my problem, which i cant solve by myself.
i'm playing SMP 1.2.5 with IC2 v1.97.
for constant energy generation i use several windmill towers and slowly add new ones. these towers have usual cross structure with 1 tin cable in the middle and 4 wms around it. the length of the cable is 39 blocks, and total amount of wms in the tower is 4*39+1(at top)=157. MV-t with a lever is placed on the bottom of the tower, so output is 512 eu packets.
the tower is placed between around 130-170 height.
i would like to build a massive MFSU storage (64 units), to use it with mass-fabricator and upgraded machines, which consume massive amounts of energy.
the problem is: i want to build lag-less scheme without losing energy during transferring it over hundred (or more) blocks.
1: if i use only fiber glass cables, starting at MV-t and ending at the MFSU storage, i will get reduced voltage plus each 40-th block of cable will make server's CPU calculate voltage drop at almost every tick from each wmt.
2: i can use 2 HV-t after each 39-th block of fiber glass to remove voltage drops (4*512 packets convert to 2048, and then convert back into 4*512), and it works fine, i have checked it. but im afraid amount of CPU operations in this case is much more, than in the 1st case.
3: i can use 1 MFSU after each 39-th block of fiber glass to remove voltage drop, but imo, its a heavy load for server. maybe im wrong. thats why im asking:

which scheme will produce less lags ?

for now, i use MFSU variant, and the whole scheme looks like that:

it's like that, because i cant use one MFSU for several towers: amount of energy from 1 tower is enough to charge one MFSU and keep stable 512 output (about 70%-80% of time).

here is the scheme which i'd like to build, with transformators :

and here is the scheme which i never used (1-st case):

there is also one more problem and its about MFSU storage.
i use cluster of 16 MFSU, working in parallel mode, so the output is 16*512=8192 eu/t, for my machines and mass fabricator.
but i noticed voltage drop in the fiber glass while the storage was charging (no output, all machines were turned off). it was showing about 508 eu/t instead of 512 eu/t.
and here is the question: where did 4 eu/t gone?
i logged off SMP and tried to build a bigger MFSU storage in SSP, and here is what i've got:
11*11=121 MFSU (square) is charged by 5 MFSU (the same energy source i have in multiplayer, 5 wmt).
5*512=2560, but it shows less, only 2263, energy loss is about 11.6%.
also, the closer MFSU is located to the energy source (if u follow along the wire from source to storage), the more eu/t it gets. closest MFSU get 41 eu/t, the farthest one gets about 10 eu/t (corner ones). it looks like "gradient charging" depending on how many storage devices is already connected to the wire.
if i connect only 4 MFSU to the fibre glass: no voltage drop happens, 512 eu/t everywhere, but if i connect more: voltage drops a little amount.

maybe its working as intended, but question is: is there a way to avoid this voltage drop while charging multiple storage devices?
here are only 2 pics of the experiment with MFSU, cant attach more
sorry for bad english skills, its not my primary language and i use it very rarely, but there is no better place to find the answers