Doing some construction lately, I have noticed how inaccurate some
component values are.
Good quality capacitors that I have been using are a long way out from
the marked value. Also recommended turns for toroids to produce a given
inductance can be quite a bit out. The only solution is to use an
inductance and capacitance metre and  select the components. Your
matching network otherwise could be a long way out and you might be
wondering what is wrong.
There is a vast difference between powdered iron cores and ferrite
material for the same given inductance. ie a T200-2 powdered iron core
needs about 67 turns to achieve 54 uh whereas a ferrite core type 3c6
only needs 7 turns. There are other considerations when selecting cores,
power, Q factor etc. Having completed some calculations at the design
stage it is then worth experimenting with different cores to achieve the
most efficient energy transfer, avoid overheating and saturation of
core.
A lot of information about filter and matching networks is contained in
the ARRL handbook, this will get you started when comtemplating a LF tx
project.
Some marked component values are well beyond the 10 percent tolerance.
One capacitor marked 0.022 uf was in fact 0.033 uf (wrongly marked in
this case). That is why some projects do not perform as designed and you
wonder why.
73 G3KEV