The transformer ferrite core.
A transformer made with a 61mm MIX#61 ferrite ring core (Fair-Rite Part Number: 5961003801) proved me to be the only suited for a transformer which magnetically couples RF energy between feeder and antenna.

Up until 20 MHz, it has low loss and good temperature stability. Its temperature coefficient is very low and the Curi temperature is very high. The transformers inductance will vary only very little between -25C and + 150C, resulting in temperature stable properties. The max flux of the core is 25mT, allowing maximal 400W RF power. 

            The feed point impedances of a 10m circumference loop. 
G0CWT measured in the current maximum of an 1/4 wave lambda loop a feed point impedance of 5.5 ohms.

A matching transformer for a 10m circumference loop, tuned to 7.1 MHz, should therefore have an impedance ratio of 50 : 5.5 ohms.
Its turns ratio should be : SQRT(50 : 5,5)  = SQRT(9 : 1) = 3 : 1. 

            1/8 lambda circumference loop :
The feed point impedance in the current maximum of an 1/8 lambda loop is 9.3 times lower (0.59 ohms) than in an 1/4 lambda loop.

A matching transformer for a 10m circumference loop, tuned to 3.65 MHz, should therefore have an impedance ratio of 50 : 0.59 ohms.
Its turns ratio should be : SQRT(50 : 0.59)  = SQRT(85 : 1) = 9 : 1. 

        The actual number of turns for a practical transformer.
1.  The parallel winding reactance of a transformer winding should be at least 4 times higher than its loading impedance.
2.  The number of turns must be kept minimal for low leakage inductance..
3.  The smallest possible number of turns for a winding is 1.
4.  A secondary winding of 1 turn on a 61mm #31 ferrite core shows at 3.65 MHz a parallel Xl of 3.7 Ohms, 
     which is 6 times higher than the 0.6 ohm loop load impedance.
5.  The 80m secondary winding therefore consist of 1 turn.
6.  For 40m then are 3 secondary turns needed.
7.  The total secondary winding therefore consist of 3 turns, with an 80m tap after 1 turn.
8.  As the winding ratio for 40m is 3:1, the primary winding consists of 9 turns.

            Winding inductance and reactance check :
-  Xl of every transformer winding should be at least 4x higher than its loading impedance.
-  L = n² x 0.17 uH. 
-  Xl = 2 x 3,14 x f x l .

       3.5 MHz :
Primary 9 turns. L = 13,8 uH.  
Xl = 304 Ohm +- 25%..
Xl : Zload = 304 / 50 = 6,1.

Secondary : 1 turn. L = 0,17 uH +- 25%.  
Xl = 3,7 Ohm  +-25%.
Xl : Zload = 3,7 / 0,59 = 6,3

        7 MHz :
Primary : 9 turns. L = 13,8 uH. 
Xl = 607 Ohm  +- 25%.
Xl : Zload = 607 / 50 = 12,1.

Secondary : 3 turns. L = 1,53 uH  +- 25%. 
Xl = 67,3 Ohm   +- 25%.
Xl : Zload = 67,3 / 5,5 = 12,2 .

        The disturbing leakage inductance in the transformer.
As its is impossible to cover the whole core circumference with only one or even three secondary turns, an optimal coupling between the primary and secondary transformer windings is  impossible. 

The unwanted effect is an extra, non visible, fixed inductance (the leakage inductance) seems to exist in series with the secondary winding. 

This leakage inductance in the transformer made a VSWR of 1 : 1.0 impossible. 

The in the prototype measured values of the leakage inductances were :
80m (1 turn)   = 117 nH.
40m (3 turns) = 358 nH. 

            The leakage inductance is small if :  
-  the number of turns is minimal,
-  all turns are wound very tight onto the ferrite core,
-  all turns are divided equally over the full core circumference.
-  primary and secondary windings are wound in between each other.

In this case has the secondary winding a very small number of turns, and its connections must be at the opposite side of the primary connections.
The secondary turns therefore could not be spread over the full circumference of the core. 

The leakage inductance can be kept minimal if the transformer is wound with 
- all turns are wound very close to the core surface, 
            and
- the secondary turns are wound in between the primary turns.

            Wire :
- two silver plated PTFE 13AWG wires of different colors connected in parallel and wound in parallel,
             or  
- a thick twin mains power wire with a max. width of 7mm wound without twists,
            or 
- a thick speaker wire with a max. width of 7mm wound without twists.

The in the prototype measured values of the leakage inductances were :
80m (1 turn)   = 117 nH.
40m (3 turns) = 358 nH. 

            Wideband match, developed by pa0nhc.

            My values for these series resonance capacitors are :
C80 => 4700 pF. Fres with transformer wires abt. 3650 kHz.
C40 =>   806 pF. Fres with transformer wires abt. 7100 kHz.
You could use them as a guide when setting up your match box.

        Capacitors :
With 400W power on 80m, 25.5 Arms runs in C80, and about 260Vrms (365Vp) stands across C80..
Best use high voltage / high current SMD capacitors, or wired mica capacitors.
Use enough capacitors of equal value in parallel, to keep the RFcurrent in each capacitor below capacitor specifications.
There is room enough for a large number SMD capacitors.
400V film or ceramic capacitors with low temperature coefficient are useable to. 
The new PCB is suited for all.

            Heath losses.
After 10 minutes 100W continues power, the temperature rise in the transformer was totally acceptable, and occurred mainly in the  copper inside the core hole in the secondary (highest current) winding. 

The ferrite core did NOT show warming up. The insulation between the windings and the ferrite core was totally satisfying, as no arcing nor insulation weakening occurred. 

Warning : MAXIMAL peak power on 80m is 400Wrms.